The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
FOREWORD -On October 29 and 30, 1979, the Federal Reserve Bank of St. Louis and the Center for the Study of American Business at Washington University co-sponsored their fourth annual conference. This volume presents the papers and comments delivered at the conference, entitled 11 Stabil i za ti on Po 1i ci es: Lessons from the ' ?Os and Implications for the '80s." The conference was divided into three sessions. The first of these considered recent developments in the theory of stabilization policy and empirical evidence on the effects of stabilization policies. The second session focused on evaluations of the monetary and fiscal policies pursued in the 1970s. The third and final session discussed international aspects of stabilization policies. This foreword pre- sents summaries of the three sessions. Stabilization Policies: Theoretical and Empirical Issues In his paper "Recent Developments in the Theory of Stabilization Policy," John Taylor focuses on current theoretical work on the response of output and employment to changes in aggregate demand. guishes between two approaches: He distin- information-based theories in which uncertainties about economy-wide disturbances are emphasized, and contract-based theories in which temporary riqidities in wages and prices are emphasized. The former set of theories, combined with rational expectations, are the foundation of the "new classical microeconomics." In these models only unanticipated disturbances affect real variables, and systematic policy has no effect on real variables. The contract models, on the other hand, allow for temporary rigidities in wages and prices and therefore yield more traditional conclusions about the short-run response of output and employment to demand disturbances and policy actions. Taylor contrasts the implications of these two approaches for two important issues of stabilization policy: the possibility of improving economic performance of output and employment through systematic variation in policy instruments, and the cumulative output loss associated with anti-inflationary monetary restraint. The information-based models suggest an absence of any gains associated with policy activism and an ability to decelerate inflation without a prolonged or serious rise in unemployment. The contract-based models suggest that there may be gains to policy activism and that there may be sizable costs in terms of foregone output associated with policies aimed at reducing inflation. In his comments on the Taylor paper, Hyman P. Minsky rejects both the new classical microeconomics and other theories based on "neoclassical" economics as meaningful frameworks for understanding the role and effects of stabilization policies. models: Minsky believes that these (l) lack the potential for economic instability that makes policy actions potentially useful; (2) ignore important developments, beginning in the mid-60s, that radically changed the environment in which stabilization policy must operate; (3) abstract from essential aspects of economic institutions, particularly the evolution of the financial system and of financial practices which have made the economy increasingly susceptible to financial instability with the accompanying threat of a serious debt-deflation process. ii The second paper in the first session, "Empirical Evidence on the Effects of Stabilization Policies" by Laurence H. Meyer and Robert H. Rasche, begins with a survey of monetary and fiscal multipliers. These are examined both across various large scale macroeconometric models and simple reduced-forms, and over time, to assess the degree of consensus and the nature of the evolution in policy multipliers as the various macroeconometric models have been refined. The authors give special attention to the difference in estimated fiscal policy multipliers between the large scale income-expenditure econometric models, on the one hand, and the St. Louis reduced-form equation on the other hand. Meyer and Rasche then develop the implications of both the largescale income-expenditure models and smaller monetarist models for the two issues highlighted in Taylor's presentation: the cumulative output loss associated with anti-inflation policies and the gains from policy activism. They contrast the large cumulative output losses implicit in both conventional estimated Phillips curve equations and monetarist models with the implications of rational expectation macro models. Meyer and Rasche note, however, the importance of balancing the gains from reducing inflation against the transitional costs associated with reducing inflation. They conclude with a survey of empirical evidence on the gains to policy activism, based on model simulations which compare the simulated performance of the economy under fixed rules, ad hoc rules with feedback, and optimal control. In his comments on the Meyer and Rasche paper, Neil Wallace rejects as useless any results based on the current generation of large-scale econometric models and reduced forms. ii; According to Wallace, these models are not "coherent'' in the sense that their conclusions are not derived from a mutually consistent and defensible set of assumptions. However, he admits that the same criticism can also be applied to almost all the recent rational expectation macro models. The important contribution of these models, according to Wallace, is not so much the policy ineffectiveness conclusion which has attracted so much attention, but the demonstration that the assumptions made about how economic agents forecast future values of variables have great influence on the response of real variables to macroeconomic policies. Stabilization Policies: Critique of the '?Os and Preview of the '80s The second session focused on evaluations of the monetary and fiscal policies pursued in the '?Os. In "The Case for Gradualism in Policies to Reduce Inflation," Allan H. Meltzer rejects as myth the view that the current inflation has its roots in the Vietnam War era deficits. Instead, Meltzer states that the proximate source of the current inflation is the monetary policy of the early 1960s, and that inflation persists because monetary policy continues to sustain anticipations of future inflation. Meltzer then develops the rationale for a policy of "gradualism'' -pre-announced, gradual, sustained declines in the rate of growth of money. Meltzer emphasizes the importance of conducting monetary policy in a way that permits individuals to quickly recognize permanent shifts in the rate of monetary growth. If monetary growth is volatile, indivi- duals have difficulty in inferring from observed money supply figures what direction the Federal Reserve is likely to take in the future. This situation results in a slow adjustment of expectations about iv future monetary growth and inflation to a permanent decline in the rate of monetary growth -- and, as a consequence, a serious cumulative output loss. By announcing its target and reducing the variance of actual monetary growth around its target, the Fed promotes more rapid revision in inflation expectations and minimizes the cumulative output loss associated with anti-inflation policy. In "Federal Budget Policies of the 1970s: Some Lessons for the 1980s," Michael E. Levy is critical of monetarist explanations of the persistent inflation of the last decade and a half. While recognizing the important role of monetary change in the inflation process, Levy argues that monetarist explanations, such as that provided in the previous papers by Allan Meltzer, fail to take the analysis far enough. Although they identify the Federal Reserve as the ultimate source of inflation, monetarists do not give the Fed's inflationary behavior adequate explanation. The fundamental source of the inflation of the last decade and a half, according to Levy, lies in the drastic changes in social attitudes and in economic policies that got underway in the mid-60s and persisted throughout the '70s. This new social activism resulted in large and rapidly growing federal programs designed to transfer real after-tax income from the productive sectors to nonproducers, a dramatic increase in both the size and scope of civilian programs, increased reliance on deficit spending, and a new wave of socially-oriented regulation. The dominant forces behind the persistent inflation were the following: the increased expansionary thrust of the budget; the acceleration in monetary growth in order to accommodate the deficit financing; the acceleration in wage demands as workers attempted to reverse the V decline in after-tax rea1 income associated with tax-transfer programs; the increased reliance on "inflationary" social security taxes; the increased business costs associated with regulation; and the slowdown in productivity and real growth resulting from disincentives, both to work and invest. In his comments on the Levy paper, William Poole suggests that the slowdown in productivity could have raised the inflation rate associated with a given rate of monetary growth by only one or two percentage points. The remainder of the rise would have to be asso- ciated with increases in monetary growth to accommodate the inflation initiated by the other factors cited by Levy. Poole states, however, that Levy does ·not provide any evidence that the factors he cited were quantitatively important sources of inflation pressure. Moreover, Albert Burger, in his discussion paper, notes that Levy gives relatively little attention to the behavior of the Fed and hence leaves unanswered the question that motivates Levy's objection to the monetarist explanation of inflation: "Why did the Fed accommodate these i nfl ati onary forces?" In his luncheon address, Lawrence K. Roos, president of the Federal Reserve Bank of St. Louis, described what he termed the "shortcomings" of the past monetary policy actions and announced his enthusiastic support for the Fed's recently announced change in the method by which future monetary policy will be conducted. Although the new policy approach, which places ·primary emphasis on the growth of reserves and monetary aggregates, holds the promise of avoiding the policy errors of the past, Mr. Roos cautioned that there are several steps which must be taken if the policy change is to fully achieve its vi desired results. Among the necessary steps are increased focus on the growth in the monetary base, the avoidance of monetary policy surprises, and a commitment to a long-run policy viewpoint so that neither political pressures nor false expectations force abandonment of the new policy. He emphasized that the new policy must be given at least a year to prove its value and should not be expected to dissipate inflation in a matter of months. Stabilization Policies: International Aspects Jacob Frenkel began the third session of the conference with a thorough analysis of the experience with flexible exchange rates in the 1970s. His paper, "Flexible Exchange Rates in the 1970s," sets forth the way in which the asset market or monetary approach to exchange rate determination helps to explain this experience, particularly the observed volatility in exchange rates and the relation between exchange rates and both domestic and foreign interest rates and price levels. Within this framework Dr. Frenkel highlights the central role of expectations, particularly expectations about future inflation, in determining exchange rates. An explanation of the volatility of ex- change rates is aided by the view that these rates are a financial variable whose value is sensitive to expectations about future developments and is capable of quickly incorporating new information about these developments. The central role of inflation expectations suggests, .ccording to Frenkel, an "intimate connection between monetary policy and exchange rate policy" and imposes a "unique responsibility on the monetary authorities in affecting the rate of exchange." vii H. Robert Heller outlines in his paper, "International Stabilization Policy Under Flexible Exchange Rates" the adverse effects that the move to flexible exchange rates has had on international trade, international capital movements, and foreign investment. Heller takes the position that the increased uncertainty about exchange rate fluctuations has resulted in a significant increase in costs to the business sector and that the adverse effect of this uncertainty has been particularly evident in the decreased willingness of investors to undertake direct investment and long-term construction activity abroad. He also suggests that speculative capital flows may have accentuated rather than reduced the fluctuation in exchange rates. These increased costs, moreover, were not offset by any benefits associated with flexible exchange rates, such as greater freedom for domestic stabilization policies. Heller notes that it will be impossible to return to fixed exchange rates as long as national inflation rates differ so widely. He concludes his paper with a series of recommendations for improving the functioning of the international monetary system under flexible exchange rates. To preserve the dollar standard, the United States must act to maintain the real purchasing power of the dollar. This, in turn, will require better control of monetary aggregates and will be facilitated by adoption of longer-term monetary aggregate targets. In his comments on the Frenkel and Heller papers, David Laidler emphasized the implications of flexible exchange rates for the response of inflation and output to deceleration in monetary growth by a single country. Flexible exchange rates, according to Laidler, impart an added degree of price flexibility; hence they permit both a more rapid viii deceleration in inflation and a reduction in the cumulative output loss associated with anti-inflationary policies. This fact suggests that empirical approaches which do not explicitly allow for the effect of an open economy under flexible exchange rates may seriously overestimate the cumulative output loss. In his comments on the Heller paper, Geoffrey Wood remarks upon the lack of evidence to support Heller's contention that flexible exchange rates have had a harmful effect on international trade. Wood also objects to Heller's contention that destabilization of capital movements has been an important source of volatility of exchange rates. In Wood's view, the volatility of exchange rates simply reflects the underlying volatility of national monetary policies. Washington University Laurence H. Meyer Associate Professor of Economics ix RECENT DEVELOPMENTS IN THE THEORY OF STABILIZATION POLICY John B. Taylor During the past decade the theoretical framework underlying macroeconomic stabilization analysis has undergone a number of significant developments. Theories designed to explain the crucial linkage between aggregate demand policy and real economic variables have been revised following the research on the "new microfoundations" of employment and inflation. Critical expectations effects of stabilization policy have been incorporated into the theoretical framework through the use of rational expectations. Optimal control techniques have become sophisticated enough to be used on large nonlinear econometric models, and more recently have been adapted for use in models with endogenous expectations. Supply considerations have been recognized as having important policy implications and, when necessary, have been incorporated into policy analyses. Theories underlying the choice between rules and discretionary policy have been altered and refined. These developments are likely to play an important role in the practical evaluation of economic policy in the years ahead. This paper reviews these developments in the theory of stabilization policy and outlines some of their implications for macroeconomic policy evaluation. The first section reviews the theories which have John B. Taylor is Professor of Economics, Columbia University. The author wishes to thank Robert Barro, Jerry Green, Dale Henderson, and Laurence Meyer for helpful comments on an earlier draft, and the National Science Foundation for financial support. been developed to explain the effect of policy variables on the real economy. As there is still little consensus here, a number of alter- native representative models are presented and compared. The second section examines the implications of these different theories for the problem of reducing the rate of inflation, which is likely to be one of the more important policy issues in the years ahead. The third section discusses a number of issues which have arisen in recent policy analyses and which are closely related to the changes in the theoretical framework: The Lucas critique of traditional policy evaluation procedures, the applicability of optimal control, the choice of rules versus discretion, and the applicability of the new equilibrium approach to stabilization policy. With few exceptions this review focuses on theoretical research on domestic stabilization policies. International considerations and empirical results are reviewed in other papers prepared for this conference. Some of the topics reviewed here have recently been the sub- ject of a large number of survey and expositional works. The variety of survey papers by Barro (1979), Buiter (1979), Fishcer (1979), McCallum (1979), Phelps (1979), Prescott (1977), Santomero and Seater {1978), and Shiller (1978) and the books by the Ball committee (1978), and Sargent (1979) provide further detail and alternative perspectives on the topics reviewed here. Expectations play a predominant role in any discussion of stabilization analysis. For the discussion that follows, the benchmark assump- tion will be that expectations are formed rationally. Variations from this benchmark -- due perhaps to the necessity of people gradually learning about whether the economy has undergone a structural change -- are -2- considered in the course of the discussion along with variations in the model underlying the policy analysis. THEORIES OF AGGREGATE DEMAND EFFECTS ON REAL OUTPUT AND EMPLOYMENT In the idealized world of complete markets with perfect information about opportunities in all markets, changes in the money supply or more generally, changes in aggregate demand -- do not affect real economic variables such as real GNP and employment. Apart from distri- bution effects, aggregate demand fluctuations are translated point-forpoint into price fluctuations. Money is neutral. Many of the theoret- ical developments in macroeconomics in t~e 1970s have been concerned with explaining, in more detail and with more rigor than earlier theories, why this neutrality is not observed in the real world. A reasonably firm understanding of the mechanism generating this nonneutrality is certainly necessary for evaluating stabilization policy because aggregate demand management tools, such as money growth and government expenditure plans, are the primary instruments of stabiliza t 1. on po 11. cy. 1 1The effects of government policies which impact directly on relative prices can be evaluated in principle using the standard allocative theories of microeconomics. Some examples: a relative lowering of tax rates on capital would be expected to stimulate investment by raising the desired capital-labor ratio; a higher steady rate of inflation has allocation effects by acting as a tax on real money balances; and unemployment insurance can raise the equilibrium unemployment rate by driving a wedge into the work-leisure tradeoff. Apart from disagreement over the magnitude of the relevant elasticities for measuring these effects, there has been a general consensus among economists that such policies have real effects. However, because these policies are used for allocative or distributional purposes, they are not generally flexible enough to be considered seriously in stabilization analysis. Nevertheless, their importance cannot be overlooked in analyzing macroeconomic trends. See Feldstein (1978) for a summary of such effects on unemployment. -3- Recent theories of the observed link between aggregate demand and real variables can be grouped into two types -- information-based theories in which the uncertainties about economy-wide disturbances are emphasized, and contract-based theories in which temporary rigidities in prices and wages are emphasized. At the risk of becoming too taxon- omic, it will be useful to further classify each of these theories. Among the information-based theories it is important to distinguish between those in which the uncertainty is whether an observed economic change is local or economy-wide, and those in which the uncertainty is whether an economic change is temporary or permanent. Similarly, among the contract-based theories it is important to distinguish between those that emphasize relative price shifts due to asymmetrical rigidities (for example, wages are rigid while prices are flexible), and those that emphasize the general persistence of all prices due to nonsynchronous price {or wage) setting relative to a prevailing trend in prices (or wages). Uncertainty about Local Versus Aggregate Economic Conditions Perhaps the most significant finding of the research 2 on the "new microeconomics" is that imperfect information about economic conditions outside an individual's own market or industry can have profound implications for the behavior of inflation and employment. Suppose aggre- gate demand increases because of a higher rate of money growth. Then individual firms will find an increased demand for their products, and will respond by increasing their production (and perhaps running down 2see Phelps et~ (1970). -4- :heir inventories of finished goods). But much of this higher real Jroduction may be due to the misperception on the part of each firm that the increased demand is a relative shift toward the product it sells. 3ecause there is always imperfect information about whether an increase in sales is a local phenomenon, this misperception and the consequent rea 1 output response is una voidable. If, on the contrary, each firm ,new that the increase in demand was common to all firms in the economy, 3nd was due to the purely nominal increase in the money supply, then its 1roduction response would be much smaller. If prices and wages were generally flexible, then firms would know that prices and wages should ➔ uickly rise to offset the increase in the money supply, and therefore that an increase in output would not be warranted. In the limiting case of perfectly flexible prices, good information about what is going on elsewhere in the economy enables fil11ls to respond just as they would be predicted to do in the money-neutral world of general equilibrium theory. But even with perfectly flexible prices, imperfect information creates a non-neutrality in which firms respond to aggregate demand stimulus by increasing real output. The link between aggregate demand and real vari- ables, according to this theory, depends in no essential way on price or wage rigidities. As long as there is imperfect information about the source of aggregate demand shifts, the correlation between aggregate demand and real output will exist. Of course, the possibility of a coincidence of perfectly flexible prices and wages with these well-known empirical correlations means that policy implications will be much different. Simple descriptions of this theory are found in Phelps et al. (1970) and Lucas (1973). The algebra of the Lucas presentation is -5- convenient for our purposes and can be represented in terms of a simple quantity theory of aggregate demand. (l) y + p = m+ v combined with an "aggregate supply" equation (2) y = u(p - p). All variables are measured in logarithms and should be thought of as deviations from secular trends: y is real GNP, pis the aggregate A price index, mis the money supply, and vis velocity. The p term represents a forecast of the price level before the information about m ~ and v becomes available. The difference between p and p represents the average difference between each firm's observation of demand conditions during the period and its guess about economy-wide demand conditions. This difference represents the misperception or mistake discussed above which causes firms to increase their production. production responses is y. The sum of all firms' (It turns out that it is convenient alge- braically to use prices to index demand conditions.) A Substituting from (l) into (2) and noting that from (2) that y=O, we find 3 (3) y u(m - m + v - v). 3we take p to be a rational (unbiased} forecast of p; hence E(p-p) = 0. "Biased" forecasts are treated in Section 1.2 below and arise because of information confusion about what is the actual model underlying policy or the structure of the economy. These ~"biased" forecasts have forms which resemble adaptive expectations, but unlike adaptive expectations are closely related to the structure of the model. -6- Hence real output responds positively to unanticipated money m-m and h unanticipated velocity v-v. This is the critical link between real variables and aggregate demand which the theory explains. However, because only unanticipated changes in aggregate demand affect real output, the policy implications of this linkage theory are striking: if the monetary authorities change their policy instrument m in a way which can be predicted by individuals in the economy, then in our notation m~m and the change in m does not affect real output at all. And from equation (1) the change in mis translated entirely into a point-for-point change in p, apart from any unanticipated shifts in velocity. This famous "policy-ineffectiveness" result, emphasized by Lucas (1973), Sargent and Wallace (1975) and Barro (1976), has understandably stimulated a large volume of research. The significance of this theory for practical stabilization analysis is not simply the neutrality result -- the idealized general equilibrium model has long been known to yield neutrality as discussed Rather the significance is due to the appearance of neutrality above. in a model which explains the empirically observed correlation between aggregate demand policy and real output. The theory would be of little practical importance if it did not generate this important empirical result. The econometric work of Sargent (1976) and Barro (1977, 1978), has been aimed at making this empirical connection more formal and rigorous. I think it is fair to say that this empirical work has demonstrated that the theory is consistent with these correlations. facts have been more difficult to reconcile with the theory. Other The per- sistence of unemployment is one regularity which does not emerge from -7- the simple theory, and was used as a critique of the theory by Hall (1975) and Modigliani (1977). A number of modifications of the theory to account for this persistence have been suggested. Lucas (1975) em- phasized that unanticipated shocks could cause firms' capital stock to get out of line, and this would have repercussions on production in later periods as the capital stock is adjusted. sized adjustment costs in changing employment. Sargent (1979) emphaBlinder and Fischer (1978) have placed more emphasis on finished-goods inventory being drawn down or accumulated. Optimal inventory adjustments in later periods will then require production changes and thereby cause a correlation between output changes at different dates. All these theoret- ical modifications of the basic information-based model with perfectly flexible prices can in principle explain persistence, but it has yet to be demonstrated whether actual inventory behavior or costs of employment adjustment are sufficient to explain the persistence. There is, of course, much other evidence which the theory can be tested against. Two pieces of evidence which seemingly run counter to the theory are procyclical productivity changes, and a slight tendency for real wages to vary procyclically, though the latter is much less pronounced. Sargent (1979), extending the work of Lucas (1970), has shown, however, that these observations are consistent with the limitedinformation flexible price models. His proof involves disaggregating employment into straight-time and over-time, and assuming that straighttime employment is more costly to adjust, but that over-time workers must be paid more on average. Under these conditions firms will find it optimal to employ more straight-time workers than over-time workers on average, but to make larger changes in employment among over-time workers than straight-time workers, when demand conditions change across the business cycle. This behavior implies that real average hourly earning will tend to increase during booms, because of the shift of the mix of workers toward higher paid overtime employment, even though real wages may fall for both groups of workers. Moreover, since fewer over-time workers are employed on average than strai~ht-time workers, their marginal productivity is higher. Hence, the shift to- ward more over-time employment causes average productivity in the economy to increase. Sargent (1978) has attempted to see if this intricate theory is sufficient to explain the phenomena quantitatively, and finds that, although there are some discrepancies, the theory generally conforms to the facts. Another explanation for the procyclical behavior of real wages is given in Phelps (1969) using a model of inventory behavior. New data now becoming available on real inventories may per- mit a check of this explanation. From the point of view of stabilization theory a number of extensions of the basic information-based model represented in equation (2) should be mentioned. Cukierman (1979) has shown that the limited- information assumptions can be generalized to permit firms to change their expenditures in order to better determine the source of economywide events. This makes the information structure endogenous to the rest of the economy, including policy, and thereby removes the criticism that the theory unrealistically places an exogenous information structure on economic agents. He finds that the general results of the theory are robust with respect to this modification. Mccallum and Whitaker (1979) have shown that the policy neutrality result does not apply to such aggregate demand tools as automatic -9- stabilizers because these react simultaneously to changes in economic conditions, rather than with a lag as in the feedback monetary policy discussed above. For example, with progressive taxes, after-tax income immediately changes as a fraction of total income when nominal income fluctuates. This can have direct real stabilizing effects. It should be emphasized, however, that in principle monetary policy could be made to operate just as simultaneously as the automatic stabilizers. This has not been the case in practice, however, except for extreme interest rate pegging where the central banks' supply of reserves responds instantaneously to changes in demand. Uncertainty about Temporary Versus Permanent Changes in Economic Conditions The theory discussed above emphasizes lack of information about whether demand changes are local or economy-wide. From the viewpoint of stabilization policy, an equally important type of uncertainty is the lack of information about whether an observed economic change is temporary or permanent. Theories which emphasize temporary versus per- manent effects are, of course, not new to macroeconomics, as exemplified by Friedman's (1956) original permanent income theory of consumption. Muth (1960, 1961) also emphasized the distinction in his original work on rational expectations. Here we are concerned with the importance of this uncertainty for the link between aggregate demand and real output. The general point is that a shift in nominal aggregate demand, which is expected to be permanent will have a much smaller effect on real output and a correspondingly larger effect on prices, than a shift which is expected to be temporary. -10- Suppose, for example, that in an attempt to reduce the rate of inflation the central bank reduces the growth rate of the money supply. The information problem which economic agents face is whether this change is a permanent one, or whether the central bank will soon give up on its resolve to lower the growth rate of the money supply. In reality, this information problem is not trivial, and cannot be eliminated simply by announcing that today's start at monetary restraint is the beginning of a permanent shift in policy. Lack of credibility about whether the shift is indeed permanent may be cured only by the public observing the results of the new policy. During the transition period when people learn whether the shift is temporary or permanent, the policy of restraint can have real output effects, even if prices are perfectly flexible. This can be illustrated using the algebra introduced above. 4 Equation (2) can be written in terms of inflation rates rather than price levels by subtracting the lagged price from p and p. This gives when rrt is the expected rate of inflation. Suppose that rrt = rrt so that there is initially no uncertainty, but that starting in period t+l the central bank reduces the rate of growth of the money supply to a level that will generate an inflation rate of rrs ' < rrt for s > t. If the new policy is not fully credible, then people will not immediately adjust their expectations to rrs. A reasonable assumption would be that they expect a level of inflation which incorporates the new information 4The following discussion is based on Taylor (1975). -11- about rr as well as the previously expected rate of inflation. In simple terms: s (5) = t+ l , t+2, ... Formula (5) can be derived more formally using Bayesian techiques which incorporate the uncertainty about whether the new inflation rate is permanent or whether the observed change is a temporary occurrence. The parameter\ will be time dependent in general, however, and this should be taken into account if one is interested in quantitative policy evaluation. To see the effects of the new monetary policy on real output assume for simplicity that ns is equal to a constant n* for s > t + 1. Then from (5) we have (6) for s rr > 5 = \ t + l. S-1 E i S' (1-\) rr* + (l-\) rrt i=O Hence, rrs converges torr*, but will be greater than rr*, if rr* is less than rrt (if the new monetary policy is to aim for a lower rate of inflation). The gap between the expected rate of inflation rrs and the actual rate of inflation rr* will be larger, the smaller is\. Hence, the less credibility there is about the new policy, the larger the inflation gap and the larger the reduction in real output. will be no reduction in real output if \=l. There In this way the uncertainty about permanent versus temporary effects has an important influence on the way policy is linked to real economic variables. The type of model represented here in very simple terms has been emphasized in stabilization policy analyses by Fellner (1976), -12- B. Friedman (1979), and Taylor (1975). A full macroeconomic model developed by Brunner, Cukierman, and Meltzer (1979) uses the distinction between permanent and temporary effects to examine the influence of supply shocks as well as demand shocks on production. Flood and Garber (1979) have provided estimates of similar credibility parameters in the case of monetary reform in the German hyper-inflation. These types of models have been criticized, especially when used for policy analyses of the type discussed here, because they appear to depend on policy deception (see Barro (1978)). While the potential for deception is clearly present in these models they are equally applica~e to situations where all parties disclose their intentions. nately, disclosure does not generate immediate credibility. UnfortuIt is the problem associated with this lack of credibility which these models emphasize. Contracts and Relative Price Effects Imperfect information is not the only reason that aggregate demand would be expected to influence real output. Temporary rigid- ities in prices or wages might force some of the change in nominal demand into changes in real production. Since casual observation suggests that such rigidities are pervasive either in the form of explicit contracts or less formal implicit contracts, economists have been willing to take these rigidities as given. The main theoretical development in this area during the past several years has been to recognize that the form W'lich these rigidities takes is important for stabilization analysis. Attempts have been made to model these rigid- ities with more detail than was previously available, and to trace out -13- the implications for policy. Two different forms of this type Of analysis can be usefully distinguished. The most common form of this type of model assumes that wages are at least temporarily rigid, but that prices are perfectly flexible in the sense that firms cannot directly influence profit margins by marking up their prices relative to wage costs. Firms simply adjust their demand for labor when the real wage shifts against them. Recent examples of this type of model are found in Fischer (1977), Phelps (1978), and Calvo (1980). Letting wt represent the nominal wage and keeping the notation introduced earlier, the most rudimentary form of this model is When the real wage rises firms reduce output and employment, until the marginal productivity of labor is increased. If wt is partially pre- determined, perhaps because of multiperiod contracts which were set in previous periods, then the link between aggregate demand and real output fo 11 ows directly. If aggregate demand is determined according to equation (1) then (8) a a(vt- wt) Yt = ~ t + l+a and clearly changes in nominal mt get translated into real output. The mechanism is simply that a higher money supply raises prices which lowers the real wage and stimulates employment and production. The major advance in using this type of model has been to develop the mechanism determining the nominal wage. -14- Fischer assumes, for ixample, that there are overlapping contracts with a fraction of the :ontracts set in each period so as to keep the expected real wage :onstant. A consequence of this assumption is that aggregate demand !ffects do not persist for longer than the length of the longest con:ract. Another consequence is that wage or price trends have no :endency to persist. In these two respects this type of model has many °eatures which are similar to the results of the information-based models. "his has led Gramlich (1979), for example, to conclude that wage-rigidities fo not add much in the way of policy implications to a rati ona 1 expecta- :i ons mode 1s. In pri nci p1e, of course, announced monetary po 1icy affects ·eal variables in such models, even with rational expectations. >een emphasized by Fischer (1977). This has The question is whether they describe ;he wage and price dynamics in an empirically accurate way that is rele1ant for policy analysis. The main feature of these models is their dependence on real wage :hanges for all employment effects. As discussed above, it has been Hfficult to find much variation in the real wage over the business :ycle. Empirical checks of this model along the lines of Sargent (1978) Jsing the distinction between straight-time and over-time workers would :herefore be very useful. On the other hand, there are important policy problems where :hanges in real wages are the central issue. For example, a supply ;hock could shift the marginal productivity downward requiring a reduc:ion in the real wage. With sticky wages, this reduction might be lifficult without monetary intervention. In effect the monetary ruthorities can use monetary policy to shift the price level to a msition such that the real wage is equal to the level which workers -15- would have aimed for, if they had known about the shock when they signed the contract. This is the conclusion of Phelps (1978) who bases his analysis on such a model. Gordon's (1975) analysis of agricultural supply shocks reaches a similar conclusion if farm prices shift up while industrial prices are assumed to be relatively riqid. Blinder (1979) also emphasizes these relative price rigidities in examining the appropriate response of policy to an oil price shock. One difficulty with all these analyses is the possibility that the assumed rigid wage (or price) eventually adjusts to offset the policy-induced shift in relative prices. In the Phelps analysis, this is not much of a diffi- culty in principle because the real wage is pushed toward what workers and firms would have negotiated otherwise. Another difficulty, already alluded to, is that the models do not capture much of the persistence effects of inflation and unemployment which now seem to present important policy problems. In this respect they are similar to the informa- tion-based models reviewed above. Staggered Contracts and Inflation Persistence By most measures the variability of the general price level in recent years has been larger than the variability of all but a small number of relative prices. For example, the real wage has been rela- tively stable compared with the sharp rise in nomina1 wages and prices. Moreover, changes in both nominal wages and prices are more highly correlated with business cycle fluctuation than changes in the relative wage. For these reasons, one might suspect that analyses which focus on real wage changes as the sole cause of employment shifts might be omitting at.her factors. -16- Another class of models which are based on rigidities in wages nd prices deemphasize the aggregate effects of relative price shifts nd focus on the problems of general price movements. These models mphasize the fact that all prices and wages are not set in unison cross the economy but are generally staggered, and that a primary eterminant of the price decision is the prevailing price outstanding n the market. Hall (1979) has recently developed a microeconomic odel which gives an explanation for the importance of setting prices elative to the prevailing price. An example of this type of model is given in Taylor (1979). irms and workers decide on a wage xt in period t which is to last for wo periods. The contract wage xt is set according to the expected revailing wage during the contract period with suitable adjustments to eflect demand conditions. Hence 9) here wt~ 1/2(xt + xt_ 1 ) is the average wage at time t. ions of Yt represent demand pressure on wage decisions. The expectaIf we make he additional assumption that profit margins are relatively stable hen pt~ wt+ y where y is a constant parameter which we can set to ero without loss of generality. By holding the relative wage constant, he model purposely abstracts from relative price changes and focuses n general price movements. In this model, as with the previous model based on price rigidties, aggregate demand policy has a direct effect on real output. If quation (1) is the aggregate-demand relationship, then the mechanism -17- works as follows: the price level is predetermined since the wage is predetermined and profit margins do not adjust. Hence, an increase in the money supply increases real balances, which tends to increase the real demand for goods. This results in an increase in production and hence an increase in employment. Eventually wages and prices will ad- just because the favorable demand conditions will give firms the incentive to pay increased wage demands. and reduce real money balances. This in turn tends to raise prices Eventually a new equilibrium is reached at a higher price level but with the same level of production. Money is neutral in the long run. What is different about this model compared with those discussed in the previous section is that convergence to the new equilibrium takes time, and there is never any important shift in relative wages (there is a period during which the workers who had settled their contracts when the money supply was changed tend to fall behind other workers but this is not necessarily integral to the workings of the model). The inertia in wage movements following the shift in money supply can be demonstrated by solving the model to obtain 5 where Band o depend on the parameter a. Hence, a change in the money supply sets off a series of changes in the contract wage xt and hence in the average wage wt. This series of changes in wt is matched by the price level Pt and, if the money supply is held fixed at the new level, 5The derivation requires the use of rational expectations to solve out for the expectation variables. -18- is reflected in a similar pattern of changes in real output. Because of these persistence effects this tyoe of model would seem to be more useful for examining stabilization problems associated with reducing inflation, or more generally achieving price stability, than the models discussed in the previous section. If changes in real wages are also thought to be important, then they can easily be incorporated into the analysis. Theoretical frameworks of this kind have been used for policy analysis by Phelps (1978a), Gertler (1977), Modigliani and Papademos (1978), Papademos (1979), and Taylor (1980). These models have some similarities to the "disequilibrium" models developed by Clower (1965) and Barro and Grossman (1976). Im- portant differences not generally found in "di seq uil i bri um" models are the use of rational expectations, a reasonably explicit description of the contract mechanism, and a reliance on the more traditional aggregate demand framework without the development of market spillover effects or of binding supply constraints. These differences largely reflect empirical considerations or modelling strategies. It is not yet clear what is to be gained empirically or theoretically from incorporating disequilibrium spillover effects. A recent paper by Green and Honkapohja (1979) has attempted to bring rational expectations into a framework which corresponds more closely with the disequilibrium models. However, their approach is designed to avoid explicit treat- ment of the nonlinearities caused by setting market transactions equal to the minimum of supply and demand. Rational expectations are much easier to deal with in linear models, and this is one reason the "demand is determining" assumption is used. Another reason is that the assumption seems to be empirically realistic in many situations. -19- ~omparison of the Alternative Theories What sets the contracting models off from the information-based models is of course the use of "sticky" prices, and the corresponding disuse of the market-clearing assumptions. In the contract models, markets "clear" in the short run in the sense that supply adjusts to meet the demand; in the long run, prices eventually adjust to clear markets. In the information models, on t~e other hand, prices instan- taneously adjust to clear markets in the short run. better? Which approach is I have used the contracting approach because it corresponds more closely with my interpretation of the market mechanisms in the real world. It is not just the widely discussed long-term labor con- tracts which suggest this interpretation, but also the much more common (at least in the U.S.) implicit contracts, which are much shorter and a re usually not called contracts. In fact, long-term labor contracts have so many indexing provisions that they probably correspond more closely with shorter contracts. Research in this area has shown that "contracts" do not have to be very long to generate a very lengthy persistence of wage and price inflation. ample.) (See Taylor (1980), for ex- But in using these contracting models, one has to be aware that without an explicit utility maximization framework, there is a possibility that the models are not robust to changes in policy. Again my preference has been to make the most of these models in situations where the contracting mechanisms are relatively robust. At the same time, it is difficult not to appreciate the theoretical elegance of the information models, and the potential to use the traditional tools of microeconomics to conduct policy analysis with these models. But even the information-based models have some ad hoc -20- assumptions, especially when they need to be modified for empirical work. One of the major recent developments in the literature on market- clearing rational expectations has been to pursue a more theoretically rigorous approach with the aim of omitting the remaining ad hoc features, in particular the money demand equation or quantity theory equations (such as equation (l} in this paper). Cass and Shell ( 1979) . 6 See Wallace (1977) and The work by Azariadis (1975}, Baily (1974), and D. F. Gordon (1976) does not provide as much of a foundation for contract models as one might have originally thought. These theories do not suggest why contracts are set in nominal terms without contingencies. In fact, Barro (1979) has suggested that these microeconomic theories are more useful in showing that the market-clearing models are useful "as if" devices. Calvo and Phelps (1978) and Hall and Lilien (1979) have pro- vided alternative theories of contracts which emphasize the practical and theoretical difficulties of making contracts contingent on everything. Most of the policy discussions associated with the theories reviewed above have been about the effectiveness of policy or whether policy activism is useful or not. In the market-clearing setting, only 6A useful appraisal of the overlapping generations model approach advocated by Wall ace is contained in Cass and She 11 ( l 979). The major appeal of this approach is the enormous theoretical mileage one gets from the disaggregation of generations. At an abstract level this disaggregation is very similar to the disaggregation of contracts according to when they are negotiated -- a feature of the contracting models discussed in Section 1.4. More generally one suspects that different types of disaggregation are likely to yield additional theoretical insights. Another example is the two-sector model explored by Sargent and Wallace (1971), Henderson and Sargent (1973), and Foley and Sidrauski (1970). -21- unanticipated changes in aggregate-demand policy matter, so announced policies do affect output. In contracting models aggregate-demand pol- icy has effect whether it is anticipated or not. Hence, in these models, policy is effective and, in certain cases, policy activism is desirable. Some examples of the optimal reaction to supply shocks were discussed above. Mccallum (1977) has argued that price rigidities are not really the source of the policy effectiveness in the contracting models. In criticizing the contract model used by Phelps and Taylor (1977) he shows that monetary policy is ineffective if one removes inventory effects on production, but uses the supply equation in the form of equation (2). However, inventory effects on production are an important part of models where prices do not adjust to clear markets. Firms will want to increase production, for example, if inventories are drawn down below optimal levels because price adjustments are not quick enough. This is the rationale behind the inventory effects on production in the PhelpsTaylor model. Omitting the term attributes suboptimal inventory management to rational firms. This point has been demonstrated by Frydman (1979) in a critique of McCallum's results. The main outcome of the policy-effectiveness debate is a general consensus that rational expectations per se does not rule out effective aggregate-demand management. It is the flexible-price market-clearing assumption that makes po 1icy ineffective for short run s tabi1 i zat ion policy .7 7Fischer (1978) and Lucas (1975) mention the nonneutrality that comes even in market-clearing models from the substitution out of money into real capital when the expected rate of inflation rises. However, this mechanism is not seriously considered as a tool of aggregate demand-management. Moreover it is likely to be offset by tax effects. A useful discussion of the relationship between rational expectations and policy effectiveness is found in Lucas (1980). -22- POLICIES TO STABILIZE PRICES The practical policy implications of these models can be alternatively stated from the viewpoint of price stabilization rather than from the viewpoint of policy intervention to affect output. Suppose, for example, that the rate of inflation is generally agreed to have become too high, either because of past policy mistakes or unavoidable velocity shifts, and that tne monetary authorities want to reduce the rate of inflation. The important question is whether the monetary re- straint necessary to achieve this goal of price stabilization will cause a recession and how large that recession will be. The answer to that question will obviously influence the policymakers' choice of how 11uch restraint to apply. If we take literally the information-based models, which emphasize the uncertainty between aggregate and local shocks, then if this policy of restraint is announced it will not have any effect on real output. There will be no recession since inflation will match the reduction in monetary growth point for point. This striking conclusion is, of course, contrary to the views of many economists and policymakers, and I think for this reason the model is still rejected by many economists as a practical guide to policy. On the other hand, if there is uncertainty about whether the changes in policy are permanent or temporary {as discussed above), then the real effects of policy will exist, and a recession would be expected to occur. The size and duration of the recession would depend on the speed with which people begin to believe that the central bank is firm in its resolve to restrain money growth. If the credibility is high or increases quickly, then the recession could be very mild. -23- Fellner (7979) indicates why he thinks that credibility is likely to increase quickly, if a clear announced policy of restraint is undertaken, and that people's expectations of inflation would be swiftly revised downwards. The contract-based models yield different conclusions. The models which emphasize real wage shifts because of asymmetric rigidities do not suggest any reason for a recession to last longer than the length of the average contract. The inflation rate could be put on its new target path in the first period; in the second period wages would adjust. In fact, if the restraining policy was announced and believed one period (year?) in advance, there would be no decline in output. In this case, this type of contract model does not give results that are much different from the market clearing models. The general staggered contract models suggest, on the other hand, that the recession would be somewhat longer because the adjustment process is passed on gradually from one contract to the next. However, because there are some forward-looking features to these models (see equation (9), the recession would not be expected to be as severe as would be implied by the simple reduced forms (see equation (10)). The policy of restraint (if it is believed) would change the parameters of (10), so as to reduce the size of the recession. Accurate quantitative estimates of how much the parameters would be expected to change have yet to be obtained, though simulation results in Taylor (1980) suggest that it is likely to be significant. In sum, each of the models reviewed here has implications about the real effects of a policy of price stabilization. (These models, ignore, of course, any direct positive real effects that a more certain -24- irice level might bring; see Fischer and Modigliani (1979) for a dis:ussion of these direct effects.) In the cases where the real effect s likely to be significant, it would be interesting and useful to :ompare empirically its magnitude with the estimates provided by con1entional econometric techniques as summarized by Okun (1978). l This is feasible and well-defined estimation problem as the discussion above 1akes clear. ALTERNATIVE TECHNIQUES FOR THE ANALYSIS OF STABILIZATION POLICY This section gives an overview of several recent developments :oncerning the choice of alternative techniques to analyze stabili~ation policy. Some of these issues are intimately connected with the ~heoretical developments summarized in the first section. rhe Lucas Critique of Econometric Policymaking Econometric models have played a large role in policy formulation in recent years. It is rare that the staff members of policymaking igencies do not run alternative policies through the major large scale ~conometri c models before meeting with their jo not have formal models of their own. II pri nc i pals," even if they Whether this heavy use of ~conometric models actually influences the decisions of policymakers is ~nother question. Political or other noneconomic considerations are frequently a factor. But when "pure" economic advice is sought, the results of the econometric models are certainly taken into account. For ~xample, the property of almost all econometric models that nonaccomnodative monetary policy has small effects on prices and large effects )n output, undoubtedly influences policymakers to choose more accomnodative policies than they otherwise would. -25- Lucas (1976) has criticized this type of econometric policymaking. He argues convincingly that the parameters of these models are not invariant to changes in policy, so that the policy experiments performed on these models (which treat the parameters as fixed) give misleading results. R. J. Gordon (1976) suggests that suitable modifications of econometric policy evaluation procedures could deal with the Lucas criticism. The parameters could, in principle, be made endogenous. The parameters of econometric models can shift for many reasons, but the one Lucas emphasized was that rational economic agents would forecast the future effects of policy, and accordingly, modify their behavior in a way not described in the econometric models. To deal with this problem it is necessary at least to reestimate the econometric models taking these expectation effects into account. The most prac- tical way to do this with existing econometric techniques is to use the rational expectations assumption. Having specified and estimated an econometric model with rational expectations it is then possible to perform a policy analysis to take account of the expectations effects. This is the approach taken by Taylor (1979a}. A simple quarterly econometric model of the U. S. economy was estimated during the 19541976 period, imposing rational expectations on economic agents. Using the estimated parameters of this model, alternative policies were compared, and for a given set of policy preferences, optimal policies were calculated. Because the model incorporated contracts of the kind dis- cussed above, a policy tradeoff between inflation and unemployment was implied by the model and this was calculated using the estimated parameters. The tradeoff was characterized by a "best" relationship -26- )etween output stability and price stability. 8 This optimal relation;hip apparent·ly dominated actual policy during the period as well as ~he policy of a constant growth rate for the money supply. Constant noney growth would have given better results than actual policy, how~ver, according to these estimates. Anderson (1979) and Fair (1979) have tried to estimate the quantitative significance of the Lucas critique by simulating conventionilly estimated econometric models, with rational expectations inserted. fhey both find the effects to be quantitatively significant, but their "esults are difficult to interpret because the conventional models were ,ot formulated as rational expectations models. For example, Anderson (1979) finds that the Phillips curve is much steeper when he imposes 0 ational expectations on the model. But clearly the specifiers of his nodel would have altered its specifications if they knew rational exJectations would be imposed. It is likely that the adaptive expecta- tions distributed lags used in such models are designed to caoture }ther dynamic properties than pure extrapolative forecasting. Quantitative work of this kind with rational expectations is only just beginning. More experience with these techniques will be 1ecessary before they can be accurately appraised as significant im)rovements over conventional econometric policy evaluation procedures. rhe results available thus far are promising, are already giving rough 8 Flemming (1976) p. 73 suggests that a tradeoff between output ;tability and price stability might be a good way to characterize the )olicy problem. Phelps and Taylor (1977), Taylor (1980), and Green and ,onkapohja (1979) have calculated theoretical tradeoffs of this kind. \n international comparison of such tradeoffs is given in Taylor :1980a). -27- empirical estimates of the effect of policy, and indicate that further research is fruitful. Two objections can be raised against these attempts to account for the Lucas critique. One is that the rational expectations assump- tion is not accurate because it does not incorporate learning on the part of individuals about the economy. If this learning problem is significant, then these techniques will have to be modified. Learning effects are likely to be a serious empirical problem immediately following a major economic reform. This was illustrated above for the case where the monetary authorities change their policy and people do not know whether it is a permanent or temporary change. However, even if learning problems are significant, these techniques will be useful for evaluating alternative policy procedures over a long period of time. For example, it is useful to know if a less accommodative monetary policy during the 1960s and 1970s would have increased the amplitude of business cycle fluctuations as much as conventional econometric models would imply. If the use of rational expectations gave results much different from other models over long enough periods for the rational expectations assumption to be realistic, then the results would be taken into consideration in recommending how accommodative policy should be in the 1980s. Another objection to the quantitative use of rational expectations as described here is that there are other reasons that parameters of a model could change. For example, even if rational expectations were used, behavioral relations for contract-wage determination might shift with policy as workers and firms change contract lengths. While expectations are probably a significant source of parameter drift, this -28- does not mean that models can ignore other behavioral shifts. Success- ful policy evaluation requires careful modelling of all behavioral relations. The New Equilibrium Approach to Policy Evaluation Lucas and Sargent (1978) have suggested that the pervasiveness of these other sources of parameter shifts means that minor modifications of econometric models are not sufficient. They recommend a "new equi- librium" approach to modelling in which all economic relations are based on explicit utility maximization analysis. If tastes and tech- nology remain relatively constant -- or can be modelled as exogenous factors -- then this approach, in principle, will avoid these other types of parameter shifts. The approach is attractive because once one has developed a model based on sound utility maximization principles, macroeconomic policy analysis is conducted like any other welfare analysis in microeconomics. Explicit externalities can be located and offset by optimal policies, and no approximate aggregate welfare criteria such as output and price stability are necessary. One would design policy to maximize the welfare of the representative individual. Attempts to design business cycle or econometric models along these lines include the work by Barro (1976), Lucas (1975), Hansen and Sargent (1980) and Kydland and Prescott (1980). This approach represents a fundamental change in macroeconomic policy evaluation and its full practical implementation will take a long time as emphasized by Lucas and Sargent (1978). As an alternative to the approach outlined in the previous section, several reservations about this new equilibrium approach might be mentioned. -29- Does utility maximization provide any additional constraints on an economic model which do not already come from a set of explicit decision rules and rational expectations? If it does not, then the gains from beginning each analysis with explicit utility maximization are not clear. For example, one of the major~ hoc features of decision rules designed for empirical work is that they include lags to capture the gradual adjustment of firms to new economic conditions. With utility maximi- zation, these lags are "exp 1ai ned" by adjustment cos ts which tend to make it optimal for firms to adjust slowly. But one has almost as much freedom to choose adjustment costs in a utility framework as one does to choose lag length when writing down decision rules. Unless good micro- economic or technological information is available to measure these adjustment costs, the utility maximization approach does not seem to provide additional information in this case. Another reservation concerns the practical use of the welfare of the representative individual as the criterion for stabilization policy. In principle this approach is better than the alternative approach of postulating an aggregate measure of welfare, which might include measures of inflation or aggregate employment stability. gate welfare approach has advantages in practice. But the aggre- It is very difficult to incorporate some of the welfare gains of price stability into individual utility functions. The gains from a relatively stable aggregate price level involve such considerations as providing a more certain framework for private decision making. Unti1 one finds a way to incorporate these complex effects into individual utility functions, the use -30- of aggregate criteria may serve as satisfactory and workable alternatives. Rules Versus Discretion The debate between those favoring rules versus discretion has not diminished in recent years but the arguments have been modified. A definitional change is that rules are now rarely taken to mean holding policy instruments constant. Feedback rules, in which the money supply responds in a systematic way to economic developments, are rules as much as constant money growth. Kydland and Prescott (1977) have suggested that the problem of time inconsistency (see also Calvo (1979)), implies that rules should be used rather than discretion. Time inconsistency can arise because of taste change or because people forecast future behavior of policymakers. In both cases policymakers may be tempted to change plans after they have announced the optimal path. Time inconsistency does not imply that optimization techniques cannot be used (see Fischer (1980) for a discussion of this issue), but it does raise questions of how policy should be implemented. Kydland and Prescott (1977) argued that rules would be a way to reduce the incentive for policymakers to change plans. Rules do not generally exploit the initial conditions of a maximization problem as much as fully optimal policies. If policymakers do not exploit initial conditions today, then people might expect that they will not exploit initial conditions in the future. But of course there is no logical guarantee. -31- This preference for rules over discretion is a practical, rather than a logical, implication of time inconsistency problems. 9 Another practical reason to prefer rules over discretion is that, especially with rational expectations, it is difficult to estimate the impact of alternative discretionary paths with great accuracy. The rational-expectations assumption is not accurate unless one can assume people are familiar with how policy works; this might require that they have experience with one type of rule for a long period of time. 10 Fischer (1979) has suggested a compromise resolution to the rules versus discretion debate: rules should be used in normal times, but in the case of an unanticipated disaster (such as a financial panic) discretion should come into play. It is difficult to disagree with this eclectic solution to the problem, but practical implementation might prove difficult. Objective measures of what is normal and what is ab- normal are difficult to obtain in economics. A less constructive, but perhaps more realistic resolution to the rules versus discretion debate comes from deemphasizing the distinction between the two. If policymakers make the same policy decision when- ever their staffs' econometric forecasts are the same, then in effect 9Monetarists who advocate the use of a fixed money growth rule, suggest that, because of initial conditions (a high inflation rate inherited from the past), the growth rate be diminished to the target path slowly when starting out on such a plan. There is a time inconsistency argument here. If higher rates of money growth are advocated because of initial condition, then what is to keep people from expecting a return to high money growth when similar conditions arise again in the future? 10 Another practical reason is that statistical estimates of policy effects are considerably less complex if one can focus on rules. -32- they are using rules. The rules might be difficult to describe and even more difficult to estimate, but they are rules nonetheless. If this is a good description of the way policy works, then research which focuses on alternative rules rather than discretionary paths might turn out to be the more practically useful type of policy research. Such research might suggest ways in which the policymaking process (rule) should be modified in order to improve the performance of the economic system. CONCLUDING REMARKS This overview has been aimed at recent theoretical research in stabilization theory. Earlier research on such issues as the choice of intermediate targets, problems of lags in the effect of policy, and the effect of parameter uncertainty on the choice of policy instrument has been omitted largely because theoretical developments in these areas have been relatively minor in recent years. It should be emphasized that these older problems continue to be of practical importance. The continuing efforts to persuade the Fed to switch to a reserve targeting procedure in their short-run operating strategy is a case in point. The practical interpretation of these earlier stabilization issues has been changed in some cases, however, by the theoretical developments reviewed in this paper. For example, Poole's (1970) analysis of the choice of policy instrument loses most of its practical relevance in the mark~t-clearing models where monetary policy is ineffective. But in the contracting models , where monetary policy effects on real output are significant, Poole's analysis needs only slight modifications to account for the rational expectations effects. -33- Interest rate pegging frequently leads to instability in rational expectations models, whether prices are flexible or temporarily rigid. This policy impli- cation, which was emphasized by Sargent and Wallace (1975), appears to be robust to change in the theory which is used. 11 That many other important policy implications are not robust to changes in alternative theories -- as was emphasized here for the policy objective of price stabilization -- suggests that additional theoretical and empirical work to sort out and test these theories should be high on any agenda for future research on stabilization policy. 11 such instability can occur in the model used by Phelps and Taylor (1977) for example. Because prices are set at levels which clear markets on average, market-clearing conditions are used to determine expected future prices which in turn are used to determine the current price setting. Extreme interest rate pegging can make future prices and hence the current price level undetermined. -34- REFERENCES Anderson, P. A. ( 1979) "Ra ti ona 1 Expectations Forecasts for NonRa ti ona l Models, Journal of Monetary Economics, 5, 67-80. Azariadis, C. (1975) "Implicit Contracts and Underemployment Equilitrium," Journal of Political Economy, 83, 1183-1202. Baily, M. N. (1974} "Wages and Employment under Uncertain Demand," Review of Economic Studies, 41, 37-50. Ball, R. J. et al. (1978) Report of the Committee on Policy Optimization, Her Majesty's Stationery Office, London. Barro, R. J. (1976) "Rational Expectations and the Role of Monetary Policy," Journal of Monetary Economics, 2, l-32 . . (1977) "Unanticipated Money Growth and Unemployment in the United States , " American Economic Review, 67, 101- 15 . --~-.--,- . (1978) "Unanticipated Money, Output and the Price Level in ---t~h-e-·united States/ Journal of Political Economy, 86, 549-80 . 1 . (1979) "Developments in the Equilibrium Approach to Business Cycles," unpublished, University of Rochester. ---=---.- Barro, R. J. and H. J. Grossman (1976) Money, Employment and Inflation, Cambridge, Cambridge University Press. Blinder, A. (1979) "Supply-Shock Inflation: Money Expectations, and J1.ccommodation," M. J. Flanders and A. Razin (Eds.), Development in an·Inflationary World, Academic Press, Forthcoming. Blinder, A. and S. Fischer (1978) "Inventories, Rational Expectations and the Business Cycle," unpublished. Brunner, K. , A. Cu ki erman, and A. Meltzer ( 1979} "Stagflation, Macroeconomics and the Permanence of Economic Shocks," CarnegieMellon Working Paper, February. Suiter, W. (1979) "The Macroeconomics of Dr. Pangloss, A Critical Survey of the New Classical Macroeconomics," unpublished, Princeton University. Ca1vo, G. (1979), "On the Time Consistency of Optimal Policy in a Monetary Economy," Econometrica, 46, 1411-1428 . • (1980), "Tax Financed Government Spending in a Neoclassical --~M~o~d-el with Sticky Wages and Rati ona 1 Ex pee ti ons, n Journal of Economic Dynamics and Control, 2, Forthcoming. -35- Calvo, G. and Phelps, E. S. (1978), "Employment and Contingent Wage Contracts," in Stabilization of the Domestic and International Economy, K. Brunner and A. Meltzer (Eds.), North-Holland, Amsterdam. Cass, D. and K. Shell (1979), "In Defense of a Basic Approach," California Institute of Technology, Social Sciences Working Paper No. 282. Clower, R. (1965), "The Keynesian Counterrevolution: A Theoretical Appraisal," in Theory of Interest Rates, Conference on the Theor of Interest and Money, F. H. Hahn and F. P. Brechling, ,eds. , London, Macmillan. Cukierman, A. (1979) "Rational Expectations and the Role of Monetary Policy: a Generalization," Journal of Monetary Economics, 5, 213-30. Fair, R. C. (1979), "An Analysis of a Macroeconomic Model with Rational Expectations in the Stock and Bond Markets," American Economic Review, 67, 539-552. Feldstein, M. (1978) "The Private and Social Costs of Unemployment," American Economic Review Papers and Proceedings, 68, 155-158. Fellner, W. (1976) Towards a Reconstruction of Macroeconomics: Problems of Theory andPolTcy, American Enterprise Institute, Washington . ~ - ~ - - ,. (1979) "The Credibility Effect and Rational Expectati ans," Brookings Papers on Economic Activity, l, 167-178. Fischer, S. (1977) "Long Term Contracts, Rational Expectations, and the Optimal Policy Rule" Journal of Political Economy, 85, 191-206. ~ - - - - · (1979) "On Activist Monetary Policy with Rational Expectations," in S. Fischer (ed.) Rational Expectations and Economic Policy, University of Chicago Press, Chicago (forthcoming) . . (1980) "Dynamic Inconsistency, Co-operation, and the --~B-e-ne-vo lent Dissembling Government, Journa 1 of Economic Dynarni cs and Control, 2, forthcoming. 11 Fischer, S. and F. Modigliani ( 1978) "Towards an Linders tan ding of the Real Effects and Costs of Inflation," Weltwirtschaftliches Archiv, 114, 810-833. Flemming, J. (1976), Inflation, Oxford University Press. Flood, R. P. and P. M. Garber ( 1979} "An Economic T!1eory of Monetary Reform," Jou ma l of Political Economy, Forthcomi nq. -36- Foley, D. K. and M. Sidrauski (7970), "Portfolio Choice, Investment and Growth," American Economic Review, 70, (March) 44-63. Friedman, 8. (1979) "Optimal Expectations and the Extreme Information Assumptions of 'Rational Expectations' Macromodels, 11 Journal of Monetary Economics, 5, 23-42. Friedman, M. (1956) A Theory of the Consumption Function, New York, National Bureau of Economic Research . . (1968) "The Role of Monetary Policy," American Economic --~R_e_v~iew, 58, 1-17. Frydman, R. (1979) "Sluggish Price Adjustments and the Effectiveness of Monetary Policy under Rational Expectations," unpublished paper, New York University. Gertler, M. (1977) "Rational Inflationary Expectations and the Disequilibrium Versus Equilibrium Methods in Macroeconomics," Stan ford Workshop on I nfl a ti on, Discuss ion Paper No. 15. Gordon, D. F. (1976} "A Neo-Classical Theory of Keynesian Unemployment;' in K. Brunner and A. H. Meltzer (Eds.) The Phillips Curves and Labor Market, North-Holland, Amsterdam. Gordon, R. J. ( 1975) "A lterna ti ve Responses of Policy to External Supply Shock," Brookings Papers on Economic Activity, 183-204 . "Can Econometric Policy Evaluation be Salvaged?: A The Phillips Curve and Labor Markets, eds. K. Brunner and A. H. Meltzer, North-Holland, Amsterdam. . (1976) ---~Comment" in Gramlich, E. M. (1979} "Macro Policy Responses to Price Shocks," Brookings Papers on Economic Activity, 125-167. Green, J. and S. Honkapohj a ( 1979) , "Variance Mini mi zing Monetary Policies with Lagged Price Adjustment and Rational Expectations," Harvard Institute of Economic Research, Discussion Paper Number 721. Hal1, R. E. (1975) "The Rigidity of Wages and the Persistence of Unemployment," _Brookings Papers on Economic Activity, 2, 301-35. - - - - ~ .( 1979) "The Role of Preva i1 i ng Prices and Wages in the Efficient Organization of Markets," unpublished paper, Stanford University. Hall, R. and D. Lilien (1980) "Efficient Wage Bargains under Uncertain Supply and Demand, 11 American Econor1i c Review, forthcoming. Hansen, L. P. and T. J. Sargent (1980) "Formulating and Estimating Dynamic Linear Rational Expectations Models," Journal of Economic Dynamics and Control_, 2, forthcoming. -37- Henderson, D. w. and T. J. Sargent (1973), "Monetary and Fiscal Policy in a Two-Sector Aggregative Model," American Economic Review, 63, 345-365. Kydland, F. and E. C. Prescott (1977), "Rules Rather than Discretion: The Inconsistency of Optimal Plans," Journal of Political Economy, 85, 473-492. .(1980), "Dynamic Optimal Taxation, --~R-at~,.-.o-n-a'-"'l~E~x_p_e_c~ta-t~i_o_n_s_,-and Control Theory" H Journa 1 of Economic Dynamics and Control, 2, l, forthcoming. Lucas, R. E. (1970) "Capacity, Overtime, and Empirical Production Functions," American Economic Review Paper and Proceedings, 60, 23-27 . . (1973} "Some International Evidence on Output-Inflation ---T-r-a-deoffs, t American Economic Review, 63, 326-34~ 1 .(1975) "An Equilibrium Model of the Business Cycle," ---o-f--=Political Economy, 83, 1113-44. Journal · .(1976) "Econometric Policy Evaluation: a Critique," Journal _ _ _o_f,........,_Monetary Economics, Supplement, 19-46 . . (1980} "Rules, Discretion, and the Role of the Economic ---A.-dr-"v~isor" in S. Fischer (Ed.), Rational Expectations and Economic Policy, forthcoming. Lucas, R. E. and T. J. Sargent (1978) "After Keynesian Macroeconomics," Federal Reserve Bank of Boston, After the Phillips Curve: Persistence of High Inflation and H1gh Unemployment. Mccallum, B. T. (1977) "Price-Level Stickiness and the Feasibility of Monetary Stabilization Policy with Rational Expectations," Journal of Political Economy, 85, 627-34 . ---. (1979) "The Current State of the Policy-Ineffectiveness Debate," American Economic Review, Papers and Proceedings, 69, 240-245. Mccallum, B. T. and J. K. Whitaker (1979) "The Effectiveness of Fiscal Feedback Rules and Automatic Stabilizers under Rational Expectations," Journal of Monetary Economics, 5, 171-86. Modigliani , F. ( 1977) "The Monetarist Controversy or, Should We Forsake Stabilization Policies?'',American Economic Review, 67, 1-19. Modigliani, F. and L. Papademos (1978) "Optimal Demand Policies Against Stag fl a ti on," We ltwi rtschaft l i ches Arch iv, 114, 736- 78 1 • -38- Muth, J. F. (1960) "Optimal Properties of Exponentially Weighted Forecasts," Journal of the American Statistical Association, 55, 299-306 . - - - ~ . (1961) "Rational Expectations and the Theory of Price Movernents," Econometrica 29, 315-335. Okun, A. (1978) "Efficient Disinflationary Policies," American Economic Review, Papers and Proceedings, 68, 348-352. Papa demos, L. ( 1979) "Stagflation and Monetary Policy," unpublished paper, Columbia University. Phelps, E. S. (1969) "Short-Run Employment and Real Wage in Competitive Markets," International Economic Review, 10, 220-232. _ _ _ _ et al. (1970) Microeconomic Foundation of Employment and Inflat.lorlTheory, Norton, New York . . (1978) "Commodity Supply Shocks and Full-Employment Monetary ---Policy," Jou ma l of Money, Credit and Banking, 10, 2, 206-221 . ---. (l978a) and Monetary "Disinflation without Recession: Adaptive Guideposts Policy," Weltwirtshaftliches Archiv 114 . . (1979) "Introduction: Developments in Non-Walrasian Theory" ---in E. S. Phelps Studies in Macroeconomic Theory, Academic Press, New York. Phelps, E. S. and J. B. Taylor (1977) "Stabilizing Powers of Monetary Policy under Rational Expectations," Journal of Political Economy, 85, 163-190. Poole, rJ. (1970) "Optimal Choice of Monetary Policy Instruments in a Simple Stochastic Macromodel," Quarterly Journal of Economics, 197-216. Prescott, E. S. ( 1977) "Shaul ct Contra 1 Theory Be Used for Economic Stabilization," in K. Brunner and A. H. Meltzer (Eds.) Optimal Policies, Control Theory and Technology Exports, North-Holland, Amsterdam. Santomero, A. M. and J. J. Seater (1978) "The Inflation-Unemployment Trade-Off: A Critique of the Literature," Journal of Economic Literature, 16, 499-544. Sargent, T. J. (1976) "A Classical Macroeconometric Model for the United States," Journal of Political Economy, 84, 207-238. _ _ _ _. (1977) "The Persistence of Aggregate Employment and the Neutrality of Money," unpublished. -39- Sargent, T. J. (1978), "Estimation of Dynamic Labor Demand Schedules under Rational Expectations, Journal of Political Economy, 86, l 009- l 044 . - - -• ( 1979) Macroeconomic Theory, Academic Press·, New York. Sargent, T. J. and N. Wallace (1971) "Market Transactions, Costs, Asset Demand Functions, and the Relative Potency of Monetary and Fiscal Policy," Journal of Money, Credit and Banking, 3, 469-505 . . (1975) '"Rational Expectations' the Optimal Monetary Instrument and the Optimal Money Supply Rule," Journal of Political Economy April 241-254. --~~~~~--,----;----,-~· Shiller, R. (1978) "Rational Expectations and the Dynamic Structure of Macroeconomic Models: A Critical Review, Journal of Monetary Economics, 4, 1 (January), 1-44. Taylor, J. B. (1975) "Monetary Policy during a Transition to Rational Expectations," Journal of Political Economy 83, 1009-1021. • ( 1979) "Staggered Wage Setting in a Macro Model , " American ---E~c-o-nomic Review, Papers and Proceedings~ 108-113~ _____ . (1979a) "Estimation and Control of a Macroeconomic Model with Rational Expectations," Econometrica, September, forthcoming . • ( 1980) "Aggregate Dynamics and Staggered Contracts," Journal ---o~f-Political Economy, forthcoming~ • (1980a) "Output and Price Stability: An International Comparison," Journal of Economic Dynamics and Control, 2, February, forthcoming. -----c,c--- Wallace, N. (1977) "Why the Fed Should Consider Holding M0 f"'.onstant," Federal Reserve Bank of Minneapolis Quarterly Review, Summer Issue, 2-10. -40- ON THE EFFECTS OF STABILIZATION STABILIZATION POLICY EMPIRICAL EVIDENCE ON H. Meyer and Robert H. Rasche Laurence H. Macroeconometric Macroeconometric research in the the 1970s 197Os has been been dominated dominated by by the the refinement of large-scale income-expenditure macroeconometric macroeconometric models, models, the the attenpt attempt to to reconcile the the policy multipliers derived from these models with yielded by simple reduced-forms, models with those those yielded by simple reduced-forms, the the refinement refinement and and estimation of the relation between inflation inflation and unemployment, and the the application of optimal control techniques to macroeconometric models. These four themes provide the focus for this this paper. The The first section reviews the implications of of various nacroecomacroecononetric nometric models for monetary and fiscal multipliers. We are particuparticu- larly concerned here with with the degree of consensus consensus across models and the evolution of of estimated estimated models over time. The second section section discusses discusses attempts to reconcile reconcile the divergent implications of of income-expenditure income-expenditure structural nodels models and the St. St. Louis Louis reduced-form for fiscal policy multipliers. In the third section estisection we develop the implications of esti- mated mated Phillips Phillips curve equations and monetarist monetarist models for the response of unemployment, output, and inflation to to traditional demand management management policies. And in the fourth section we consider consider the accumulated evievi- on the results of dence on on the gains from policy activism, drawing on optimal sinulations with optimal control control simulations with aa variety variety of of nacroeconometric macroeconometric models. models. Laurence H. Meyer is Associate Professor of Economics at Washington Associate Professor Scholar at at the Federal Federal Reserve Reserve Bank of St. St. University and Visiting Scholar Louis. Louis. Robert H. Rasche is Professor of Economics at Michigan Michioan State University. University. -41-41- During the last half of the ‘70s '70s increased attention has been focused on the way in which economic agents agents form expectations, particuoarticu11 1 arly inflation i nfl ati on expectations, ex pee tat i ans, and on on “equilibrium” equil i bri um!! macroeconomic macroeconomic mode 1s larly models embodying “rational "rational expectations.” expectations." These models yield dramatic dramatic concon- clusions about both both the costs of eradicating inflation and the gains from activism. We therefore consider consider the implications of of rational rational exex- pectation models in both both the third and fourth sections, although there is as yet only a small literature literature on empirical applications of these upon. models to draw upon. AA COMPARISON OF POLICY MULTIPLIERS ACROSS ACROSS MODELS AND TIME In this this section we review the evidence from structural models models and reduced-forms about the size and time time pattern of policy multipliers. We are interested in the average size multipliers, the the consensus consensus size of multipliers, estimated multipliers. across models, and the evolution over over time time in in the estimated AA Comparison Comparison of Multipliers_Across_Models Multipliers Across Models (1975) has surrgnarized summarized the the consensus across models rather Christ (1975) pessimistically: " ... though models forecast reasonably well over “. . . horizons of four to to six quarters, they disagree disagree so so strongly strongly about the effects of of important important monetary and fiscal fiscal policies effects monetary and policies that that they they cannot cannot be be considered reliable guides to such such policy policy effects, until it can be dedeof them are wrong in this this respect and which which (if any) any) are termined which of right.” right." (p. 54) 54) 1, 2, and 33 present Tables l, present policy multipliers from seven econoeconometric models models (Bureau of Economic Analysis (BEA), Brookings (B), (BJ, UniverUniversity sity of Michigan (MQEM), Data Resources, Inc. (DRI), (ORI), Federal Federal Reserve Bank of of St. Louis (St.L), (St.L), MIT-Pennsylvania-SSRC MIT-Pennsylvania—SSRC (MPS), and Wharton Bank St. Louis (MPS), and Wharton (W)) (W)) —42— -42- as reported in Fronin Fromm and Klein (1976). The multipliers multipliers are reported for the the first quarter quarter and fourth, eighth, twelfth, sixteenth, and twentieth quarters and and for three policy changes -- an increase increase in in real real —— government expenditures on on goods and services, aa decline in personal money supply or nonborrowed nonborrowed reretaxes, and an increase in either the money serves. The mean mean and coefficients of of variation for the the various multimultipliers are also reported.11 TABLE TABLE 11 Fiscal Policy Fiscal Po 1icy - Increase in in Government Government Expenditures Expenditures - Model IC* BEA BB MQEM DRI 74 ORI 74 St.L St. L MPS W w 62 62 561 56I 621 62] 611 61I 621 62I 651 65I RMSE(4Q)* 6.94 6.94 5.13 13 5. 6.20 4.60 4.98 4.23 4.64 (w/o St.L) 5t.L) Mean (w/o St. St. dev. dev. (w/o St.L) s.d./mean s.d./mean Mean (w/St.L) St. St. dev. (w/St.L) (w/St.L) s.d./mean s.d./mean Multiplier lQ lQ 4Q 40 SQ l2Q l6Q l 6Q 20Q 1.1 l. 1 1.8 7.8 1.4 l. 4 1.3 l. 3 0.5 l. 2 1.2 1.3 l. 3 2.2 2.2 2.8 1.77 l. 2.1 2. l 0.5 2.2 2.0 2.0 2.2 2.2 2.7 1.4 l. 4 2.2 —0.2 -0.2 2.2 2.2 2.4 1.88 l. 1.6 l. 6 2.0 1.0 7.0 1.7 l. 7 -0.2 -0.2 -0.5 2.4 1.33 l. 1.5 7.5 1.1l l. 1.77 l. —0.2 -0.2 1.35 l. 35 2.17 2. 17 0.36 0.17 1.93 1. 93 ..71 77 0.37 0.24 0.18 1.23 l. 23 .39 0,32 0.32 2.18 2. l 8 0.43 0.20 1.84 l.84 .98 . 98 0.53 2.4 1.0 7.0 2.0 2.0 —0.2 -0.2 0.7 2.6 2.6 1.75 l. 75 0.76 0. 76 0.43 0.43 1.47 7.47 1.01 l. 07 0.69 0.69 IC initial *IC= initial conditions conditions for for policy policy simulation; simulation; square error for four quarter forecast of of of dollars at 1958 prices) over 1961-1967 * = 1.37 l. 37 1.03 l. 03 0.75 1.14 7.14 1.11 l.ll 0.97 l. 1.99 1.17 l. 17 0.86 0.74 .97 .97 .94 .94 0.97 RMSE = root root mean mean RMSE real GNP (billions period. = 1The multipliers are reported with with and without the St. St. Louis model multipliers. The latter are based based on aa reduced-form income equaequation than on particularly in the case tion rather rather than on aa structural structural model model and, and, particularly in the case of of the fiscal substantially from the multipliers based fiscal multipliers, differ substantially the multipliers on on the structural models. —43— -43- The mean mean fiscal fiscal expenditure expenditure multiplier multiplier is is just just over 1-1/4 in in the the The over 1—1/4 first quarter and builds to 2—1/4 2-1/4 by the end of year year two; however, however, the cumulative multiplier is still over one after five years. there While there is about the multipliers through three is considerable considerable consensus consensus about the multipliers through the the first first three deteriorates sharply. years, the agreement deteriorates Note that in all cases the multiplier multiplier peaks within within three years, generally generally within four to to eight eight quarters; and cumulative fiscal multipliers fall to to zero or below below by the 16th quarter the fifth quarter for the St. Louis model, by the 12th 12th to 16th for the MPS model and by by the 24th quarter for the BEA model. But it it TABLE 22 Fiscal Policy - Tax Cut — Model Multiplier Multiplier ________ 1Q lQ 4Q SQ 8Q 12Q 120 l6Q 16Q MQEM DRI DR! 74 St.L* St. L* MPS Ww 0.4 1.0 0.6 0.9 00 0.4 0.5 1.22 1. 1.6 1.6 1.2 1. 2 1.3 1. 3 00 1.3 1. 3 1.22 1. 1.44 1. 1.6 1.11 1. 1.2 1.2 00 2.1 2. 1 1.7 1. 7 1.1 1.66 1. 1.1 1. l 0.9 00 2.2 1.9 1. 9 0.8 1.5 1. 5 1.22 1. 0.6 Mean (w/o St. 5t.L) L) St. St. dev. dev. (w/o St.L) St. L) s.d./mean 0.63 0.63 0.26 0. 41 0.41 1.30 0.16 0 .16 0.12 0. 12 1.52 1. 52 0.37 0.37 0.24 0.24 1.47 0.52 0.52 0.35 1.25 1. 25 0.47 0.38 Mean (w/St.L) (w/St. L) St. dev. (w/St.L) s.d./nean s.d./mean 0.54 0.34 0.63 1.11 0. 51 0.51 0.46 1.30 1. 30 0.66 0.51 1.26 1. 26 0.73 0.58 1.07 1.07 0.64 0.64 0.60 0.60 BEA B B 1.8 1.6 * Multipliers reported for St. Louis model are based on * Multipliers reported for St. Louis model are based on absence of aa tax variable in in the model’s model's reduced-form equation for income. income. -44- multiplier to reach zero in takes eight to ten years for the cumulative multiplier in the Brookings and and the Wharton and Michigan models and still longer in ORI models.22 DRI The tax multipliers multipliers are smaller than the expenditure expenditure multipliers; multipliers; they build from an an initial initial mean value of peak of 1.5 at at the of 0.63 to aa peak end of the the second year. In the case of aa tax change, change, there is less consensus in deterioration in in the first quarter, but no no deterioration in later later quarters. quarters. The The tax tax multipliers tend to peak aa bit later than the expenditure expenditure multipliers, multipliers, generally between between the 8th and 12th 12th quarters, and then then decline. TABLE 33 Monetary Policy Policy Monetary Model BEA ORI DRI St.L MPS Ww Multiplier MV* RU RU Ml RU RU RU Mean (w/o (vito St.L) St. St. dev. (w/o) lQ lQ 4Q 4Q SQ SQ l2Q l2Q 16Q l6Q 00 D.3 0.3 1.1 l. l 0.3 0.3 1.44 l. 0.2 4.1 4. l 4.4 3.2 4.5 4.5 0.4 8.3 2.8 2.8 8.4 7.2 7.2 0.7 6.5 1.2 l. 2 12 .4 12.4 8.6 8.6 0.7 2.8 2.8 —0.4 -0.4 14. 14.55 8.0 8.0 0.5 1.24 l. 24 3.0 0.65 0.65 6.08 0.63 0.63 7.05 0.69 6.50 6.50 0.95 0.95 (Ml == narrow money supply; * MV MV == monetary monetary variable variable (Ml supply; RU RU nonborrowed as in non borrowed reserves; initial conditions same as I. Table l. * = 2Note also that the fact that the cumulative multiplier multiplier turns negative negative does not guarantee guarantee a negative long-run multiplier since sinte these models are subject to oscillatory convergence convergence to their their long—run long-run values. —45— -45- ~ There are only four comparable multipliers for monetary policy (those using nonborrowed reserves). The initial quarter quarter mean mean multiThe multi- plier is small small and the mean multiplier peaks at the end of of the third year at at aa value of 7. 7. There is less consensus about monetary comcom- pared to to fiscal policy; the coefficient coefficient of variation is larger in all but but one quarter quarter for monetary policy multipliers. While the St. St. Louis cumulative multiplier peaks in the the fourth quarter and goes to to zero by the 16th 16th quarter, large scale model multipliers generally generally peak peak after 88 to to 12 12 quarters quarters and the MPS multiplier reported by by Fromm and Klein is still rising from the 12th still 12th to to 16th quarters. quarters. The The large large scale models thus suggest that monetary monetary policy has aa more persistent effect on outoutput than is the case in in the St. St. Louis model. The exception is is the DRI DR! model in which which the cumulative monetary policy multiplier multiplier falls to to zero by the 20th quarter. While the multiplier results do differ across models there there is is clearly considerable considerable consensus particularly over over the first two years in the fiscal policy policy when when we we exclude the the St. St. Louis Louis results. results. the case case of of fiscal The problem is evaluating how much divergence divergence in the multipliers is conconsistent with using the models for policy recommendations. recommendations. Later we we will discuss the use of of stochastic simulations which which allow allow for multimultiplier plier uncertainty within within aa particular model, model. valuable approach suggested by Chow Chow (1977). (1977). valuable approach suggested by Here Here we want want to note note the the Chow notes that Chow notes that while while policy recommendations from alternative alternative structural models models recommendations derived derived fron differ from each other, they may may nevertheless be closer to each other than to a passive policy of constant growth rates in the policy instruinstruments. The comparison Chow Chow suggests and implements implements is the improvement policy derived from in economic economic performance in one model using optimal policy -46-46— a second model relative to to the economic performance under under passive policy. Chow uses the the multiplier properties of of the Wharton and MichiMichi- gan models to construct reduced-form reduced-form equations for real and nominal GNP including government expenditures and nonborrowed reserves as the policy instruments and and employs aa conventional quadratic quadratic loss function involving deviations deviations in in real and nominal from their targets (in involving real and nominal GNP GNP from their targets (in each case average historical values over each case average historical values over the the period period in in question). question). The results of this this experiment are mixed. mixed. If If the Michigan Michigan model structure and were the the true structure and the policy policy recommendations recommendations were derived from the Wharton Wharton model, model, active would improve improve performance from the active policy policy would performance relarelative to aa passive policy; policy; costs under the the active policy would be be under 25 25 percent of those under under aa passive policy policy although although they would be 70 percent greater than if the the policy percent greater than if policy were were derived derived using using the the true true strucstructure. ture. hand, if On the the other hand, if the the Wharton Wharton model model were the the true structure structure and the policy policy recommendations were derived model, and the recommendations were derived from from the the Michigan Michigan model, the cost under an active policy would three times cost of the cost under an active policy would be be three times the the cost of aa passive policy and and about 17 times about 17 times the the cost cost when when the the true true model model was was passive policy used. used. And, of course, the Michigan and Wharton Wharton multipliers multipliers are quite close at least for fiscal policies, compared compared to say the Brookings and the St. Louis models. Thus there are other comparisons that would would lead to to even less less favorable favorable results results for for activism. activism. AA Comparison of Policy Multipliers Over Time We expected expected to find aa secular We to find secular decline decline in in the the value value of of fiscal fiscal multipliers and aa secular secular rise in in monetary policy multipliers for large scale econometric models from the the late ‘SOs '60s versions versions to the versions of of the midmid- to late ‘70s. '70s. However, published information on such —47-47- multipliers multipliers is relatively scarce and what is available available is frequently not constructed on aa comparable basis. This, of course, increases increases the value of the NBER/NSF model comparison studies but makes multiplier multiplier comparisons pieced together together from the literature hazardous. Perhaps Perhaps the most serious problems for comparing multipliers multipliers across nodels models or over time are differences in specin initial conditions and differences in the spec- ification of policy instruments, particularly particularly for monetary policy. The large scale models are invariably nonlinear, nonlinear, implying implying that that their multimultipliers are sensitive to initial conditions, particularly particularly the the degree of of economic economic slack. But there is painfully little reported evidence of the the degree of this sensitivity. There are aa bewildering bewildering number of possipossi- bilities for aa change in tax rates and even differences in nultipliers in multipliers for different different government expenditure components. The The most serious problem, however, may be differences in assumptions about the monetary policy policy instrument. instrument. Monetary Monetary policy, policy, particularly particularly in in the the late SOs 60s verver- sions, has been identified with with changes in short—tern short-term interest rates. with either In other other cases, monetary monetary policy policy is is identified ,1ith either the the money money supply or some often nonborrowed reserves. supply or some reserve reserve aggregate, aggregate, most most often nonborrowed reserves. The The choice affects both monetary monetary and fiscal multipliers since fiscal fiscal multimultipliers assume unchanged monetary monetary policy; policy; fiscal multipliers will, of course, be much larger under fixed short-term short-term interest rates than under fixed values of the money money supply or or nonborrowed reserves. In Tables 44 and and 55 we have have pieced together some policy multipliers multipliers for alternative versions of Michigan, Michigan, Wharton, Wharton, and MPS models. The The Michigan ‘70 '70 and and Wharton ‘68 '68 models models assume assume constant short—term short-term interest Michigan rates while while the others others assume assume constant unborrowed reserves. reserves. It is sursur- prising (to us at least) that the the fiscal multipliers multipliers in in the the late ‘60s '60s -48- TABLE 44 TABLE Real Non Defense Government Expenditure Multipliers - Real GNP - QQ Michigan 70 a Michigan 75b 70a 75b 1.55 1.4 l. l. 4 c Wharton 68 68c 2.0 Wharton 75 b 75b 1.3 l. 3 Wharton 79 d 79d 1.1l l. e MPS 69 69e 1.3 l. 3 MPS 75 b 75b 1.22 l. 4 2.1l 2. 1.7 l .7 2.0 2.0 1.77 l. 1.8 l.8 2.2 8 1.99 l. 1.44 l. 2.0 2.3 1.8 l.8 1.6 l.6 2.2 12 12 n. a. n.a. l. 1.00 2. 2.1l 2.6 1.77 l. 1.1 l. l 0.7 ,,' a "'' aa S. H. Hymans Hynans and H. T. Shapiro, "The “The DHL-III OHL-III Quarterly Model of the U.S. Economy,” Economy," Research Seninar Seminar in Quantitative Economics, University University of of Michigan, 1970, Table Table 4, p. 22. bb C. R. p. c Fromn and L. R. Klein, "The “The NBER/NSF Model Comparison Results,” in Fromm Comparison Seminar: Seminar: An Analysis Analysis of Results," in L. L. Klein and E. Burmeister (ex), Econometric Model Performance, Pennsylvania, 1975, Table 6, 402. M. K. Evans and L. R. Klein, The Wharton Econometric_Forecastjj9j~o4~j, Econometric Forecasting Model, Econonics Economics Research Unit, ed. , 1968, 1968, Table 5, SB. University of of Pennsylvania, 2nd ed., 5, p. 58. dd Unpublished Wharton multiplier simulations kindly provided by R. M. Young, Wharton Econometrics Forecasting Associates. e DeLeeuw and E. M. Granlich, Policy,” Federal Reserve Bull Bulletin, June F. Deleeuw Gramlich, “The "The Channels of Monetary Policy," et in, June 1969, Table 4, p. 489. Shock applied fully to to federal real wage payments. three models (including the two with constant short— shortversions of the three term rates) are so small; they peak at 2.0 or or less. less. One important difference in the later versions versions of of Michigan and and MPS models is the sharp decline decline in the cumulative cumulativ2 multiplier from its peak peak value value by the the 12th quarter. 12th There was aa. tendency in in earlier versions for multipliers multi p1i ers to stabilize stabilize at to at about 7.5-2.0 1.5—2.0 for aa longer period. This contim:,es This continues to be the case in the Wharton model; model; in both both the the ‘75 '75 and ‘79 '79 versions the be fiscal multipliers are stable or rising during the first three years. We have been able to find comparable unborrowed multiunborrowed reserves multi- pliers at different points in in time for only two models: pliers model and the MPS model. These S. These are reported in Table 5. the Wharton In In these models there is is aa fairly dramatic evolution of the nonetary monetary policy policy multiplier. 1968 Wharton model the unborrowed reserves multiIn the 7968 multi- plier for real real GNP reached aa fairly constant level in in the 11.5 .5 to to 2.0 after about one year. range after In the MPS model the multiplier multiplier is stable In in the 10.0 70.0 range during the second and third years. In the later TABLE 55 Unborrowed Reserve Multipliers (Real GNP/Nominal Reserves) 68c Wharton 68 c 0.0 1.5 l. 5 2.1 2. l 1.7 l. 7 l1 44 88 12 72 75bb Wharton 79dd l,harton 75 Wharton 79 l.4 l. 1.4 1.22 4.5 4.8 4.8 9.1 7.2 9. l 8.6 13.33 13. Notes - See Table 4. — -50So — - 69ee MPS 69 MPS 0.7 5.4 5.4 l10.0 0. 0 12.4 72.4 MPS 75b b 75 0.3 3.2 8.4 9.4 1ersions ,ersions of both models, the multiplier is continually growing over the First three years. Note also the substantial substantial increase in the size of ‘68 verthe monetary policy multipliers in the the Wharton model from the '68 ver- ;sion ion to to the ‘75 75 and ‘79 79 versions. versions. 1 1 We view the Wharton ‘68 68 multipliers as 1 fairly typical of the conventional wisdom wisdom of the midmid- to to late ‘SOs, '60s, prior to the development of the MPS model. “ST. LOUIS” COMMENTS ON THE "ST. LOUIS" EQUATION Since the original Andersen-Jordan Andersen-Jordan article article (1968) (1968) (AJ) that proproposed aa single equation equation test of the the relative importance of of monetary monetary and fiscal policies on nominal nominal GNP, nunerous numerous replications replications have been perper- formed, across time, across countries, and across functional forms and and aa number of of criticisms, mostly statistical in in nature, have have been levied against against the equation. The purpose of of this section is to review review the criticisms that have been raised against the the equation equation and to to evaluate evaluate criticisms. how robust the the equation equation appears to be against these criticisms. The conclusions of the the Andersen-Jordan investigation are are by by now now almost universally almost universally known. known. The conclusion that The conclusion that remains remains most most controvercontrover- sial is the zero cumulative fiscal multiplier multiplier for nominal nominal GNP. This This conclusion conclusion did not conform well to the conventional wisdom of the late 1960s, nor was it it consistent with other econometric results. ConseConse- quently, for the past decade there has been considerable considerable skepticism of of that yields this conclusion. conclusion. the specification that Functional Forms, and_Distributed_Lags and Distributed Lags_ Time Periods, Functional_Forms, The Ad equation was estimated The AJ estimated over over the period 52/1-68/Il 52/1-68/II and subsub- over the sequently reestimated by Andersen and Carlson (1970) (AC) over 53/1-69/TV period as part of 53/1-69/IV of the St. St. Louis model. model. -51—51— In each case case monetary policy had aa powerful and significant effect effect while the tax variable (change in in high employment receipts) was insignificant and exex- government expenditure cluded from their preferred regression and the government variable had only a small small and transitory transitory effect. Silber (1971) subsesubse- into Republican Republican (53/I-60/IV) and Democratic quently split the period into (61/I—69/IV) sigthat fiscal variables variables were sig(61/I-69/IV) administrations and found that nificant in the latter but not in the former. former. Silber argued that these results are are consistent consistent with more systematic systematic use fiscal policy results with the the more use of of fiscal policy in in the latter period. At aa minimum, these results suggest that the the time time period used in estimation can can dramatically dramatically affect the conclusions in the estimation and that the estimates estimates may may reflect the particular policies pursued over the estimation period. More recently has extended extended the More recently Friedman Friedman (1977) (1977) has the sample sample period period employed employed by AC through 76/Il 76/II and and concluded concluded that “even "even the St. Louis equation now believes in fiscal policy.” policy." In Table Table 66 we report the rere- sults of the the Ad AJ and AC equations along with with estimates estimates over over alternate time periods including Silbers Silber's two subperiods (Sl and S2), S2), Friedman’s Friedman's extended period (F), and the period 1960/1-1976/Il 1960/1—1976/Il (MR). The results The matter~ The size and sigsuggest that both money and the time period period matter'. significance nificance of of fiscal policy policy multipliers multipliers is is not not definitely definitely settled by by these results. In response to Friedman, Carlson (1978) (1978) has has pointed pointed out that the first difference form of of the estimated equation, equation, while appropriate over the AC period, is not appropriate over over the longer period period because of of heteroskedasticity, heteroskedasticity, implying that the tt values of coefficients coefficients reported by by Friedman are unreliable. When all variables variables are defined defined as as rates of two periods are change, Carlson finds that the results of the two —52-52- TABLE 66 TABLE Time Periods Time Periods A-l AJ 52/l-68/l 52/1—68/IlI AC 53/l-69/IV 53/I—69/IV Sl 51 53/l-60/IV 53/I—60/IV M M 5.83 (7.25) 5.57 (8.06) 5.58 (. 43) (.43) GG 0.17 0. 17 (0.54) 0.05 (0.17) -1. 77 -1.77 (. 90) (.90) 2.36 2.36 (. 67) (.67) Sample ' w "' ' T T R2 .60 .66 Se 4.01 4. 01 3.84 .652 4.23 F F 53/1-76/1 53/1-76/IlI MR MR 60/1-76/1 I 60/1-76/11 9.20 9.20 ( 2. 35) (2.35) 4. 94 4.94 (6.3) 5. 72 5.72 (1.07) (1,07) l. 75 1.75 (2.11) (2.11) -3.92 -3.92 (2.78) (2.78) l. 42 1.42 ( 4. 3) (4.3) 2.44 2.44 ( 5. 57) (5.57) -1.67 - l.67 (2.90) (2.90) S2 52 61/1-69/1 61/1—69/I .73 .73 .66 .69 .69 3.30 7.54 7.84 7.84 consistent with with the the hypothesis hypothesis that that the the specification specification is is stable stable and, and, consistent like like the original AC equation, equation, indicate indicate that any effect of government expenditures expenditures is small and temporary. temporary. Allen and Seaks (1979), using the growth rate specification, find that the the fiscal variable sums sums to zero in both both Silber subperiods (Eisenhower and Kennedy-Johnson) Kennedy-Johnson) but but is sigsigin subperiods (Eisenhower Nixon-Ford era (69/II-77/I). nificant in the Nixon—Ford (69/11-77/I). Over the period 60/1-76/Il we find that that both expenditure and tax variables enter enter significantly into both first difference and and rate of change change specifications. specifications. into both first difference rate of In In Table Table 77 we we report the results of the AC equation in in difference difference form over both the original period (AC) and over over Friedman’s Friedman's extended extended period (F) and and in rate of change form over Friedman's Friedman’s extended period (C) along with with the Allen-Seaks the Nixon-Ford Allen-Seaks results results over over the Nixon-Ford period period (AS) (AS) and and both both functional functional (MR1 and MR2). forms over the 1960/1—76/Il 1960/1-76/II period (MRl From From these results we we can conclude conclude that money, money, time time period, and functional form matter. Ad type The results of of AJ type equations are estimated using using polynomial distributed lags. lags. This technique requires selection of of lag lag length, degree degree of polynomial, polynomial, and and end point point constraints. constraints. Schmidt Schmidt and Waud Waud (1973) caution caution that introduction of inappropriate constraints can result in biased and inconsistent estimates and demonstrate demonstrate how changes in degree of polynomial polynomial and and end point constraints can can substantially substantially alter the the conclusions about policy multipliers. Others Others have have found of lag length of lag can affect conclusions also. We can conclude, conclude, therefore, therefore, that that the choice choice of time period, period, funcfunctional form, and and lag constraints matters aa great deal. deal. money appear very robust. robust. money appear very The results results for The results results for for fiscal policy are are dramatidramatiThe fiscal policy cally affected by these factors. -54-54- TABLE 77 Functional Functional Form Sample Form* ' en u, ' AC F C AS MR1 MR2 AC 53/T-69/IV 53/1-69/1 V F 53/1-76/11 C 53/1-76/lI 53/1-76/]] AS 69/11—77/I 69/11-77/1 MRl 60/1-76/Il 60/1-76/11 60/1-76/Il 60/1-76/11 Delta Dot Dot Dot Delta Dot Delta MR2 M M 5.57 (8.06) 4.94 4. 94 (6.3) 1.06 1.06 (5.59) .90 (1.93) ( l . 93) 5.72 5. 72 (5.31) ..75 75 (3.08) (3.08) G O 0.05 ((0.17) 0. 77) 1.42 (4.3) .03 (.40) (. 40) .36 (2.07) ( 2. 07) 2.44 (5.57) ..37 37 (2.82) -1.67 - l . 67 (2.90) -.29 -.29 (2.25) ( 2. 25) T T FR22 .66 .66 .40 Se Se 3.84 77,54 .54 3.75 3.75 Delta: * Delta: * Dot: Dot: first difference difference specification specification first rate of change specification ..56 56 .69 .42 7.84 3.02 3.02 Biases Associated With Choices of Independent Variables Biases reduced-form multipliers and The inconsistency between the AJ/AC Ad/AC reduced-form the multipliers in large-scale large—scale econometric models models generated generated aa search (on both sides of the controversy) for an an explanation. Monetarists criticized large-scale econometric models for failing to to capture the crowding—out money demand crowding-out phenomenon through misspecification of the money equation (e.g. excluding excluding a wealth wealth effect) effect) and and failure failure to to explicitly explicitly ininclude aa government financing constraint. The income expenditure counterattack focused on on the unreliability of of reduced—forms reduced-forms due due to to aa variety of of problems, some some more more easily correctable correctable than than others, associassociated with the choice of of independent independent variables. variables. been: The key issues have What are appropriate measures measures of the policy instruments? can the possibility possibility of of reverse be avoided? avoided? can the reverse causation causation be How How What biases are What biases are introduced by omission of nonpolicy exogenous variables? The Measurement Measurement of Policy Instruments There with specifying the policy There are two interrelated problems with instruments. The first is the problem of specifying the instrument that the policy authority authority directly controls. For example, if Fed if the Fed sets sets policy by controlling the value of of the monetary base, employing employing a monetary aggregate other other than than the monetary base as aa proxy for the policy instrument may bias the policy multipliers if the other aggreother aggregate varies endogenously relative to the base. AA second problem arises arises even if the instruments themselves are included if policy itself syssys- responds to economic developments. developments. tematically responds this case, case, the In this policy instruments themselves become endogenous endogenous and reverse causation again may bias the multiplier results. results. -56—56— In this section we take take up the in the next the the problem of specifying the policy instruments and in of endogeneity of of policy. problem of noted in in aa DeleeuwThe problem of reverse causation was noted DeLeeuw- Kalchbrenner (1969) (1969) comment on on the Ad AJ paper. Indeed it was the concern over this issue that arose arose out o~the o• the Friedman-Meiselman debates that motivated the the choice of the high high employment fiscal policy policy measures measures by Andersen and Jordan. Andersen Deleeuw and Kalchbrenner’s Kalchbrenner's main concern is with with DeLeeuw monetary base base or money supply as as the variable variable the Fed the choice of the monetary directly controls. They point out that the choice choice among the monetary base, the nonborrowed base, total reserves, and nonborrowed reserves depends on whether whether the Fed Fed offsets offsets the effect of movements movements in member bank on the base and of movements bank borrowing on movements in currency currency holdings on reserves. They They express no no special preference among these these alternate alternate measures suggesting only that results which hold for some some measures and not for others should be be viewed with great caution. caution. Their Their empirical results indicate that fiscal multipliers are affected by the the choice of monetary instrument; instrument; in particular, fiscal multipliers multipliers of approximately produced in the MPS model result when nonborrowed nonborrowed reserves are the size produced substituted for the monetary base. substituted The treatment of fiscal instruments instruments in the Ad/AC AJ/AC equations equations has also drawn considerable considerable comment. In order to to avoid the bias bias associated tax revenues and expenditures expenditures with the income induced movements in tax transfer payments) under under preexisting schedules of tax tax and (mostly transfer Ad/AC equations use high employment expenditures. transfer rates, the AJ/AC High employment receipts were tried but dropped from the the preferred equation equation due to lack lack of significance. The high employment employment surplus surplus was also employed in in an alternate alternate specification. —57— -57- is clearly an inappropriate measure of of stimulus stimulus assoassoThe latter is actions because because it groups components which which are exexciated with fiscal actions pected to have different multiplier responses. The same problem arises even in the high employment employment expenditures even in the case case of of high expenditures because because this this variable variable includes both both expenditures on on goods goods and services and transfers while while economic theory suggests that transfers should be be netted against taxes. Suggestions for improved specification of of fiscal variables have been DeLeeuw-Kalchhrenner (OK), Gramlich (1977), (1971), and Corrigan made by Deleeuw-Kalchbrenner (1970). (1970). Gramlich employs purchases of goods and and services of goods services Gramlich employs government government purchases rather than high high employment expenditures, expenditures, and assumes assumes no adjustment adjustment is necessary to purge it of effects of changes in income. Government Government exex- penditures are employed in in aa composite composite variable including grants—in—aid grants-in-aid and exports exports with with an an adjustment adjustment introduced for defense inventory accumulation. Deleeuw high employment DeLeeuw and Kalchbrenner suggest adjusting high receipts changes in in this endogenous receipts to to purge purge changes this variable variable of of the the effects effects of of endogenous movements in prices. Gramlich uses uses high high employment employment net tax tax revenues (taxes minus transfers) also also adjusted along lines suggested by by DK. DK. The difficulty difficulty with with all all these series series for for tax tax revenues revenues is is that that the series series for changes include nonzero entries in periods periods during during which no changes in tax rates or transfer transfer programs occurred. occurred. Corrigan Corrigan has has suggested an alternate alternate tax variable, the initial stimulus measure, measure, that indicates the tax revenues released released or absorbed by tax tax rate changes. This series This has has plenty of zeros~ zeros: For each each tax, the initial stimulus measure is the the change in tax rates times the lagged tax base. An unweighted sum for all taxes is the variable variable Corrigan Corrigan used and it continues to be be used in the New York Fed version of the St. Louis equation. equation. -58- The discussion above suggests that the simple specification specification of of both equaboth monetary and fiscal instruments instruments employed in the Ad AJ and and AC equa- tions may be improved improved upon upon and that such improvements might alter the the relative importance of monetary and fiscal multipliers. multipliers. However, the the modifications modifications suggested suggested above above have not generally generally resulted in dramatic dramatic changes in the estimated estimated multipliers in simple reduced—form reduced-form equations. While many of these suggestions seem valid, they have not helped to resolve the differences between the St. Louis equation equation and econometric econometric model s. models. Endogeneity of Policy Even if direct policy policy actions, our estiestiif we obtain measures of direct mates of their their effects will be biased if these actions themselves are systematically systematically related related to economic developments. This This problem has has widely been noted in comments on the Ad AJ equation, equation, but most critics critics inincluding DeLeeuw and Kalchbrenner considered the problems in measuring cluding Deleeuw the instruments the more likely source of bias. with with endogenous policy are easy to illustrate. illustrate. The biases associated If If aa policy instrument eliminate completely the varies in response to to disturbances so as to eliminate instability in income, the regression of variof the change in the policy varion changes able on changes in income (zero by assumption) assumption) will yield aa zero coefcoefficient on the policy instrument. Thus, endogeneity endogeneity of policy may result in aa downward downward bias bias in the policy multiplier, with the downward bias policy. bi as a funucion func i.:. ion of the effectiveness effectiveness of po 1 icy. We He can, therefore, interpret the zero multiplier on instruments as evidence of interµret the zero multiplier on fiscal fiscal instruments as evidence of their effectiveness rather than of their insignificance~ insignificance'. While the endogeneity of of policy may introduce biases into the estimates of policy policy -59—59— multipliers from both reduced-form equations and structural models, on the bases of simulation simulation results Goldfeld and Blinder (1972) suggest on that the bias is much more serious for reduced-forms. reduced-forms. If If policy responds to economic developments with aa lag, the bias bias is is reduced but not eliminated. Omitted Variables Omitted Exogenous Variables The The third major source of bias in the choice of of independent variables in the Ad/AC non— AJ/AC equation is alleged to be the omission of of nonpolicy exogenous variables. Andersen and Jordan explained explained in in an an apap- pendix to their original paper why they believed that the omission omission of other other exogenous variables variables did not bias bias their measured impact of the the monetary and and fiscal policy variables: these variables are presumed to be independent of monetary and fiscal policies and their average average effect effect is registered in the constant term. Modigliani Modigliani (1971) (1971) made the the first detailed detailed critique of the St. Louis reduced-form model on on the grounds of of (1976) reported aa more exexomitted variables and Modigliani and Ando (1976) tensive simulation results tensive set set of of simulation results supporting supporting their their view view that that omission omission of of exogenous variables variables may severely bias bias the results results of reduced forms. The ingenious ingenious simulation experiments experiments involved estimation of an an Ad AJ type equation on data generated generated by by non—stochastic non-stochastic simulations simulations of aa model. economy. economy. The model represents represents the known structure structure of aa hypothetical The simulated values values of of nominal income from the the model are the “actual” "actua 1" values va 1ues of of income income in in the hypothetical hypothet i ca 1 economy. AA reduced-form is is estimated estimated using using these these simulated values values for for income, income, and the the resulting resulting estimated multipliers are compared compared with their “true” "true" values (the values implied by structural model). by the structural The comparison comparison of the reduced—form reduced-form -60-60- multipliers with their “true” "true" (structural model) values tests the ability of simple reduced-forms, only aa couple of of policy inability reduced—forms, including only instruments, to replicate the true true value of the policy multipliers. In the 1971 paper, Modigliani emphasized the the finding that the of the St. Louis Louis equation on MPS simulated values yielded yielded aa estimate of money multiplier in excess excess of of the “true” "true" MPS MPS multiplier multiplier and reached reached the 11 “unequivocal conclusion” that reduced-form money multipliers unequivocal conclusionu multipliers are upward biased. biased. This bias was attributed attributed to positive correlation between the money supply and omitted exogenous exogenous variables. For example, example, if the Fed Fed attempts to stabilize interest interest rates (as monetarists assert they often attempts do), then the money supply will be positively correlated correlated with real sector exogenous exogenous demand variables variables and the monetary monetary policy multiplier multiplier can be be expected expected to to be biased upward. Modigliani and Ando (1976) turned their their attention to to biases in in the estimates estimates of fiscal effects and suggested that correlation correlation between omitted exogenous variables variables and and fiscal instruments in this this case might account for the small small size and transitory effects effects of fiscal fiscal instruments Louis equation. in the St. Louis Estimates of the AJ Ad type equation equation on values of of the change in nominal income based on on simulations simulations with the MPS model yield fiscal multipliers like like the original Ad AJ equation and contrary contrary to structure of the MPS model. the structure They concluded concluded that the St. St. Louis approach is "a “a severely biased and quite unreliable method of estiestimating the response response of a complex complex economy to fiscal and and monetary monetary policy actions actions” (p. 42). 11 To To demonstrate the role role of omitted variables in in the bias bias in the Ad equation, they remove any correlation between policy instruments and AJ nonpolicy exogenous variables in the structural models by assuming all -61—61— nontrended nontrended exogenous variables are constant constant at their their means and all all trended exogenous exogenous variables grow along aa constant constant trend. trend. trended The predicted The value of nominal income for this adjusted structure structure is is computed and used used to reestimate the Ad AJ equation. Fiscal approFiscal multipliers now now of aopro- priate size and magnitude magnitude confirm the crucial role role of omitted omitted exogenous variables in biasing the estimates of of the policy multipliers in in the initial Ad AJ equation. In both papers, papers, Modigliani Modigliani and Modigliani Modigliani and and Ando (MA) are carecareful to note that the evidence they present does not permit them either either to accept the MPS multioliers multipliers or reject the St. St. Louis ones. ones. But their results should make those who use St. Louis type type reduced-form equations uneasy about the validity validity of of the the multiplier results, results, particularly those for fiscal instruments. omitted variable variable bias may may be While the analysis demonstrates that omitted aa source of serious inferential inferential error in the impact of policy actions, the to be be nonconstructive nonconstructive in that it the conclusion conclusion appears appears to in the the sense sense that it does does not provide any evidence on on the particular source source of of the bias in the experiments that were conducted conducted and it it suggests abandoning the entire approach approach without attempting to investigate the issue of biases biases in in the St. Louis results directly. St. It would be useful to identify the the sources of bias bias in the estimated multipliers by introducing the the most important directly into the reduced-form equation. exogenous variables directly AA number of studies studies have have attempted attempted to address the alleged alleged biases in the the St. Louis approach directly by including nonpolicy nonpolicy exogenous variables. variables. Gordon (1976), fur for example, example, added aa “shock "shock proxy,’ proxy," concon- sisting exports, consumer sisting of of the the sum of net exports, consumer expenditures on automobiles and non—residential non-residential fixed investment investment to the St. St. Louis specification. -62— -62- Although monetary multipliers multipliers decline and fiscal multipliers increase over his his longer sample period, the the multiplier results with and without the shock proxy remain qualitatively alike; monetary monetary multipliers are significantly positive while the the sum of of the lag coefficients on the the government expenditure variable is not significantly different from zero. Recently, Dewald and Marchon Marchan (1978) (1978) have estimated estimated expanded St. six different different countries, including including the United Louis equations for six States. States. They included exports as aa separate independent variable, variable, disdis- missing the the conglomerate variable constructed by Gordon as including including too too many endogenous influences. For the United States, the Gordon result is replicated; the impact imimpact of monetary policy is reduced, the impact of of fiscal policy is left left essentially unchanged, and the exports exports variable has aa significant contemporaneous impact. AA major monetarist contention contention is that the influence of aa maintained change in in the monetary rate should should be be aa proportional change in the the growth rate of nomnomgrowth rate inal income. This This hypothesis is is alleged alleged to be aa universal phenomenon. phenomenon. However, while Dewald and Marchon Marchan cannot reject this hypothesis hypothesis for the U.S. data, the monetary response for the the U.S. is the strongest strongest of of any of the six six countries countries investigated. The long-run elasticities of nomnom- inal GNP with respect to the money stock in in the other five countries never exceed exceed .5. In France they found this elasticity only .07 elasticity to be be only .07 and in in two countries (France and the U.K.) this this estimated estimated elasticity is not not significantly significantly different from zero. zero. -63-63— Resolvjflq hePuzzleLReduced—Fonn Resolving the Puzzle: Reduced-Form Versus Versus Structural Structural Model MultiDl4~!_ Multipliers Two further tests by by Modigliani (1977) (1977) attempt to to resolve the puzzle of conflicting multiplier results. First of all, he suggests that despite the apparent large differences in the the AC and and MPS multimultipliers, the two sets of multipliers differenU multipliers may not be ~ilinificantly different'. To To test test for significance of the difference in multipliers, multipliers, Modigliani Modigliani presumes presumes that the MPS multipliers are the true ones and tests whether the AC multipliers differ significantly from the MPS multipliers. The The result result is that that they they are not significantly different different at the Si) 5.0 percent level. Modigliani Modigliani concludes, concludes, “This "This test resolves the puzzle by by showing showing that there is really no puzzle: the the two alternative estimates of of the expenditure multipliers are not inconsistent, given the margin of error of the estimates. It implies that one should accept whichever of two estimates more reliable reliable and estimates is is produced produced by by aa more and stable stable method, method, and and is is generally more sensible. To me, these these criteria call, without question, question, for adopting the econometric model estimates." estimates.” (p. 10) 10) For For those who would would still still opt for the reduced-form multipliers, multipliers, Modigliani compares the the post-sample prediction performance of of the AC equation with one in which the coefficients of of government expenditures plus plus exports were constrained to equal those based based on multipliers multipliers dederived models. rived from simulations with the MPS models. begins 197011. begins in in 197011. dominates: The The post sample simulation For the first first four equation For the four years, years, the the MPS MPS based based equation “distinctly larger” larger" errors errors in eight eight the AC equation yields "distinctly only three quarters, quarters, and results in a quarters, smaller errors in only squared 1/3 larger than for the the MPS based equation. squared error l/3 Over the next “miserably” but next two years, both both equations equations perform "miserably" but the MPS based equation is still still 11“aa bit better.” better. 11 -64-64- Conclusion The income income expenditure counterattack on on reduced-forms, particuparticularly the Modigliani-Ando results on on the implications of of omitted exogeexogenous variables, and the ability to to dramatically dramatically alter the fiscal policy substanmultipliers by by choice choice of time period and functional form, have substan- on reduced-form equations for small small and tially weakened the case based on reduced—form equations transitory transitory fiscal effects on on nominal income. income. The implied monetary policy multipliers, on the other hand, have proven robust, at least for the United States. ASSESSING THE CUMULATIVE CUMULATIVE OUTPUT LOSS LOSS OF OF ERADTCATING ERADICATING INFLATION prominent policy issue of the the ‘70s '70s and one that seems certain AA prominent '80s is the appropriate policy policy response response to dominate at least the early ‘SOs to aa prevailing high rate of of inflation. inflation. The view that there is long— is aa long- run trade-off trade—off between inflation and and unemployment, unemployment, widely widely held held at the end of the ‘60s, '60s, is now held by by only aa small minority. The key issues are the nature of the short—run short-run relation between inflation and unemunemployment exployment and the process by by which economic agents form inflation inflation ex- pectations. Macroeconomic Macroeconomic models, both both income expenditure expenditure and and none— mone- tarist versions, suggest suggest that while the the traditional demand management techniques remain quite capable capable of of reducing the rate of inflation, inflation, the cost of such such a policy policy in in terms terms of of cumulative output output loss loss would be be great. Despite Despite the importance of of the the issues, there is substantial substantial disdis- agreement about the cost of eradicating inflation inflation and and little evidence evidence on the benefits benefits derived derived as aa consequence. In this this section we present present evidence on on the cumulative cumulative output loss estimated Phillips associated with with reducing inflation inflation based on on both both estimated -65- curves and monetarist monetarist models. Then we discuss discuss the most serious limitalimita- tion of these results -- the failure to allow the results to be influinflu-- enced by the the degree to to which the public public believes policy authorities are committed to aa consistent anti—inflation anti-inflation policy. In the final analysis, analysis, the cost of anti-inflation policies in the the form of output loss must be balanced against against the benefits associated with aa reduced rate of inflainflation. tion. Empirical evidence on the cost of of inflation and hence the benebene- fits fits of of reducing inflation inflation is is quite limited. limited. Our discussion discussion of of the the benefits of of anti-inflation policies is therefore therefore confined to to deterdeter- mining how large the per period gains gains would have to be in in order order to justify incurring the cumulative output loss which we we calculated calculated from the Phillips curves curves and monetarist models. Econometric Evidence on the the Size of the Cumulative Output Loss Loss Three alternative sources of evidence on on the cumulative output loss associated with the use of of demand management policies to moderate inflation are discussed discussed below. estimated Phillips curves. curves. The first is evidence evidence directly directly from Here we we calculate how long long unemployment unemployment must be increased by either either 11 percentage point or 33 percentage points points to reduce inflation by by above the rate consistent with steady inflation to 7.5 7.5 percentage points. points. The second and and third sources sources use monetarist models whkh which include either aa Phillips curve or aa reduced-form reduced-form equation relating inflation inflation to monetary monetary change. Here we we simulate the effects on inflation inflation and output of of aa phased phased deceleration in monetary growth. Results Based on Estimated Estimated Phillips Phillips Curves Three recent recent studies studies have considered the cost of reducing inflainflation in the context of of traditional Phillips curve regressions (Perry —66-66- (1978), Okun (1978), and Cagan Cagan (1978)). Perry’s Perry's results are based on aa wage change equation using the inverse of his weiqhted weiqhted unemployment rate and lagged wage change estimated using annual observations observations over over the the 1954-77 period. period. ‘nonaccelerating His preferred equation yielded yi e 1ded aa "nonacce 1era ting employment (NAIRU) of 4.0 in inflation rate of employment" in terms of of his his weighted unemployment unemployment rate (corresponding to about 5.5 percent in the official unemployment rate in ‘77): '77): unemployment (1) Mn W W -1.88 nln -1.88 ++ 7.44 (1/Uw) ++ 0.79 A1nW tlnW_11 ++ 0.21 A1nW AlnW_22 ++ 1.07 ONIX DNJX (—2.2) (-2.2) (3.5) (4.6) (1.1) (2.9) = S.E. •= 0.70 where where WW adjusted adjusted hourly hourly earnings in the the private private nonfarm sector and and = DNIX is aa dummy for the controls equal to to —1 -1 in 1972 DNTX 1972 and 1973 1973 and +l +1 in 19744 and and 1975. 197 Any unemployment rate in excess of of the critical unemployment rate, if maintained long long enough, enough, will permit aa cycling down down of inflainflation. tion. To compute the the cumulative cumulative output loss of eradicating eradicating inflation, we we begin with Mn Aln PWset equal to to 10.0 10.0 in the two two lagged years years and at NAIRU. NAIRU. Our "moderate" policy consists consists of of increasing the weighted unemOur moderate’ policy weighted unem- ployment rate 1.0 point above above NAIRU in period 11 and holding it here until ~1n P declines declines to \ln W to 2.5, the the rate presumed equal to trend trend growth in labor productivity and, and, therefore, consistent with price stability. stability. The wage inflation rate falls from 10.0 to 9.6 percent percent in the first year and declines about 0.3 percentage points per per year thereafter year taking 23 years to reach aa 2.5 percent rate. rate. An alternative "radical" An radical policy is modeled modeled as aa 33 percent point increase in unemployment unemployment beginbeginning in period one and again sustained until until wage wage change declines declines to —67— -67- 2.5 percent. percent. This takes ~k 1111 years: takes~ years! Note Note that the nonlinearity nonlinearity in in Perry’s Perry's wage equation ensures that that the cumulative excess of person years of unemployment and, hence, cumulative cumulative output loss will will be greater in in the more radical policy case. Using Okun’s perOkun's estimate of 3.2 as the impact on output of aa 1l percent cent point increase in in unemployment, unemployment, we can convert convert the excess unemunemployment unemployployment into into output loss. 3 One One percentage percentage point increase in unemploy~ ment ment reduces output 3.2 percent or $45.6 billion dollars (calculated (calculated at 1978 value for real potential GNP). The The 33 percent point increase in an initial year output loss unemployment involves an loss of $136.7 $136.7 billion. To find the undiscounted output potenTo find the cumulative, cumulative, but but undiscounted output loss loss we we assume assume potential output will rise at a 3.3 3.3 percent rate. rate. This yields yields aa cumulative loss of $1532.6 billion for the moderate policy and $1778.0 $1778.0 billion billion for 4 4 the radical policy. The discounted output loss is essentially radical policy. essentially the initial year year loss and the number of years years required to product of the initial complete the program (not accounting for the 3.3 3.3 percent rate of growth growth potential output is the same as discounting by aa 3.3 3.3 percent percent rate); in potential the discounted losses are $1047.9 billion and and $1503.6 $1503.6 billion in in the modest and radical cases, respectively. Charts l1 and 2. The results are depicted depicted in in (Perry 1 refers to to the the moderate case and Perry 22 to the the radical radical case.) 3Estimation of sugEstimation of the Okun law relation over more recent data suggests that 3.2 may be an overestimate of the output loss associated with aa one percentage point increase increase in unemployment; unemployment; the the recent estiestimates mates are about 2.5. 41f 1f the Okun’s Okun's law coefficient is 2.5 instead of outof 3.2, these outlosses should be reduced by by about 20 percent. put losses -68- PP or or WW CHART I1 11.0 n.o CHART MODERATION IN INFLATION VIA MONETARY DECELERATION MODERATION IN INFLATION VIA MONETARY DECELERATION 10.0 9.0 9.0 8.0 7.0 6.0 5,0 5.0 "'' <.O 4.0 PERRY I ' 3.0 PERRY 2 2.0 1 .0 1.0 0.0 CAGAN cAGAN -1 .0 -1.0 -2.0 -3.0 -3.0 1 2 3 3 4 55 6 77 88 99 10 11 12 13 14 15 16 17 10 19 20 21 21 22 23 23 2 IEAR Okun finds that that aa variety of estimated estimated Phillips curves (PC5) (PCs) in in the literature quantitatively similar similar conclusions. the literature yield yield quantitatively conclusions. The six six The equations considered by Okun Okun yield aa first year reduction reduction in in inflation of from 1/6 1/2 percentage point l/6 to l/2 point and an average of of 0.3 percentage points for aal1 percentage point increase in unemployment. Gramlich Gramlich (1979) similar conclusion. (1979) reached aa similar There are are two two aspects of the which deserve aspects of the Perry Perry specification specification which deserve There unemfurther discussion: expectations are formed adaptively and and the unemployment ployment rate enters nonlinearly. The The Phillips curve curve is uniformly uniformly drawn as a nonlinear nonlinear relation and and there have been been aa number of of theorettheoret- ical explanations (including (including Lipsey and Tobin) and some empirical supsupport port (Perry’s (Perry's influential i nfl uenti al l1966 966 study, for example). However, nonlinear nonlinear and linear specifications specifications seem seem to do about as as well over sample sample through existence of nonlinearity would provide the mid-197Ds. mid-l970s. 55 The existence provide aa rationrationale for the gradual as opposed to radical policy approach; the the greater the nonlinearity, the the greater greater the cumulative output output loss loss under the radical as opposed gradual policy. The inflation inertia implicit in the Perry equation derives from two sources: sources: actual inflation is is built into into expected inflation inflation with aa lag and actual inflation responds gradually to unemployment in in excess of the critical critical rate. To the extent that the lag in incorporating incorporating actual negotiations is actual inflation inflation into into future future wage wage negotiations is long, long, indexation indexation might substantially substantially reduce the the inflationary inertia. ation, there would be aa lag. Even with with index— index- Assuming that the full effect occurs 5Cagan cagan (1977) has recently noted the surprising surpr1s1ng lack of evidence of nonlinearity nonlinearity and this has been been confirmed in aa careful examination examination by Papademos Papademos (1977). (7977). —70— -70- rL EL 2,200 CUMULATIVE OUTPUT LOSS BASED ON PERRY & CURVES & CAGAN PHILLIPS CuRVES CHART 2 2,000 2,000 1 ,800 1,800 PERRY 2 1,600 1 ,600 1.400 1 ,400 1,200 1 .200 PERRY 11 1 ,000 1,000 ' ' ___, 800 600 400 CAGAN 200 00 -200 -400' —400 -600 1 22 3 4 5s 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 YEAR 24 cumulative out— outwithin the first year would not dramatically reduce the cumulative put costs. The cumulative output loss would would decline about 20 percent percent in each case. Thus, the critical determinant determinant of the gradual decline decline in in inflation is the extremely extremely small per period deceleration deceleration in in inflation inflation associated with labor market disequilibrium (excess unemployment) in in the conventional Phillips Phillips curve, not with the slow response of inflation inflation expectations to to changes in in the actual inflation rate. Cagan develops a PC equation beginning with the natural rate specification and assuming adaptive expectations expectations Cagan’s Cagan's estimated PC PC is: (2) ( 2) Pt = u—u ut - ut 22 Pt_i 0.95 (~ ~ ) Pt-i - o.95 2 - l - u +u (ut + ut-l - 0.23 ( ~ t- 3 - + ut-2 t-2 -- ul where PPis is the quarterly rate of of change in in the CPI, uu is the unemployunemploy- ment rate for prime age males and ii u is estimated estimated from the constant constant of the regression regression (3.7, (3.7, for this this regression) regression) and and the equation is estimated estimated 1953—1977. over the the period 1953-1977. using quarterly observations over As is clear in Charts 1 1 and 2, the Cagan equation generates generates aa dramatically more more rapid decline in inflation inflation and smaller smaller cumulative cumulative output loss. 0 at a 7.5 percent inflation rate (in Beginning in period Oat the current and last last period) and at NAIRU, aa one percentage point point inincrease in the unemployment rate reduces inflation by the full 7.5 perper- centage points by by the eighth year with cumulative output loss of $4.2.9 billion, about aa quarter quarter of that associated with with the Perry and Okun results. —72— -72- Evidence Evidence Based on the St. Louis Model To provide additional evidence on on the output output effects of using using stabilization stabilization policy to reduce inflation, we ran ran simulation simulation experiments with with the St. Louis model. model. 66 We begin begin with a base run in in which the rate rate of monetary growth is at aa steady 7.5 percent percent rate beginning beginning in 1968/111 of 1968/TI! through 1978/IV. 1978/TV. This builds in inflation inertia and provides the debase against which we can evaluate the effects of gradual monetary de- celeration. Beginning in in 1973/I 1973/l we gradually decelerate monetary growth by l1 percentage point in the first quarter of each year. We then compare the policy runs with base base run and compute the cumulative output loss associated with the policy. The first set of simulations with with the St. Louis Louis model employ the l953/I-78/IV. version of the model estimated estimated over the the sample sample period 1953/1-78/IV. The general practice practice at the Sank Bank is is to to employ employ the estimates estimates of the model using all all available data data for forecasting and policy simulations. 78/TV, however, has aa very large The version estimated estimated through 78/IV, large coefficoefficient on the the demand slack variable in the model’s model's Phillips curve, curve, almost three times the size of the coefficient estimated with data through through 71/I! or 75/1, 75/I, for example. 71/ll by the the lines labeled StL1. StLl. The results are reported in in Charts Charts 33 and 44 There There is a rapid deceleration in inflation and aa low cumulative cumulative output loss. The inflation rate begins to decline very slowly; it takes two years years to reduce the inflation rate by Il perpercentage point. Thereafter the deceleration speeds up so that after 66For a description of of the St. St. Louis model model,, see Andersen Andersen and Carlson (1970). The model includes aa reduced—form reduced-form equation for nominal income and aa Phillips curve equation for price change; output output is then solved for via an identity. -73— -73- 5-1/2 years, years, inflation has declined declined by 7.5 percentage points. 5—1/2 unThe un- employment rate rate rises slowly at first and the maximum maximum increase is is only only 1.8 percentage percentage points, during the the sixth sixth year. The cumulative cumulative output loss loss is is only about $200 billion. The output loss is, of course, sensitive sensitive to the coefficient coefficient on the demand variable in the Phillips Phillips curve. Using aa version of the model estimated estimated through through 71/111, where the coefficient on on the demand smaller than in the first version discussed, variable is substantially smaller inflation decelerates much more gradually; gradually; after six years the inflainfla- points below that tion rate in the policy run is only four percentage points in in the base run. At this point unemployment unemployment is is four four percentage points points higher than in the base run. loss is $350 The cumulative output loss billion at this point and escalating rapidly. These These results are dede- Charts 33 and 44 by the lines labeled Stl2. picted in Charts StL2. Evidence Evidence Based on Reduced—Form Reduced-Form Equations 7 Given reasonable doubt about the the validity validity of the the Phillips Phillips curve, 7 it it is useful to consider the implications of of reduced—form reduced-form models that are not not tied directly to an explicit Phillips Phillips curve. curve. We consider two examples: Stein’s Stein's (1978) two equation model of inflation and unemployunemployment and AJ AJ type equations for nominal nominal income income and inflation. The The results are depicted in in Charts 33 and and 44 by by the lines labeled Stein (Stein 11 for the moderate case and Stein 2 2 for the radical case) and and StL3. 77 See, see, for example, Stein (1978). —74-74- VIA MONETARY DECELERATION INFLATION VIA IN INFLATION MODERATION IN MODERATION MONETARV DECELERATION CHART CHART 3 Pp 10.0 9.0 9.0 8.0 77 .0 6.0 ""'' ' STL 22 '~ 5,0 5.0 ST L~ ~. 4.0 STL I 3.0 3.0 STEIN STEIN II 2.0 STEIN 2 11 .0 00 1 22 33 44 5 66 77 88 9 10 11 12 12 YEAR YEAR Stein model -· In the Stein Stein model model,, both both unemployment and The Stein -- inflation are driven by by the rate of of monetary growth. Stein’s Stein's two equation model is: is: u(t) (3) A 6 (4) At ~t) dt) = 33 · 0.6 0.6 u(t-1) - + + 0.4 (t-i) (t-1) · 0.4 0.4 ~ µ7 (t-i) (t-ll c - •~04~ Ul (t~i) ·0.4 TT (t-i) (t-1) ++ 0.4 lll (t-1) is the unemployment rate, where uu is rate, rate rate of monetary growth. w ·rr is the the inflation rate and and ~ JJl is is the The critical unemployment unemployment rate is 5.0 and and the equilibrium rate of inflation is the rate rate of monetary monetary growth. ning at u u = 5,0 and ~(t) +(t-l) 5.0 rr(t) == TT(t-1) = = BeginBegin- 7.5 == ,,7(t) p~(t) = ~~ we µ7(t-l), we deceldecel1(t-1), erate the rate of monetary growth either (a) (a) gradually by 11 percentage point per ~ per year until µ7 = = 0 or (b) immediately to to 0. In the gradual policy, unemployment rises beginning beginning in year 22 and peaks in year 8 at 6.6 percent returning to almost 55 percent 16. percent by by year 16. The The inflation inflation rate rate begins to to decelerate in year year 22 initially at aa 0.4 percent point aa year rate but ultimately reaches 1.0 1.0 point per year by year 7. 7. The The inflation rate is is down to 22 percent by year 88 and thereafter declines gradually gradually to about zero zero by by year 16. $687.5 billion. billion. The The cumulative output loss is Interestingly, the gradual policy incurs aa smaller smaller cumulative loss, $613 $613 billion. cumulative output output loss, The St. St. Louis reduc reduced-form equation for income income with aa reduced— reduced—for~euatjon A second simulation equaform for inflation -- A simulation based on reduced-form reduced-form equa—- tions combined the reduced—form reduced-form for nominal income income in in the St. Louis model with aa reduced-form equation for inflation. inflation.88 The inflation 88The reduced-form equation for The inflation used in in this section was developed developed by by Jack Tatom of the Federal Reserve Bank of St. Louis. An earlier “Does the Stage earlier version of of this equation was used by Tatom in in "Does of Rate?” Federal Reserve Bank of the the Business Cycle Cycle Affect the Inflation Rate?" of of St. Louis Review, Review, September September 1978, pp. 7-15. —76— -76- 700 CHART 4 CU*JLATIVE CUMULATIVE OUTPUT LOSS ASSOCIATED WITH NONETARY MONETARY DECELEMTION DECELERATION 600 STEIN 11 STEIN 2 500 400 ""' ' STL 33 STL 300 200 STL 1 100 100 0 2 3 4 55 6 77 8 9 10 11 12 13 14 15 16 17 18 19 YEAR 20 reduced-form includes aa twenty period distributed lag on the rate of of reduced-form change in the money supply and aa four quarter distributed distributed lag lag on on the differential in in the rate of change in producer prices for energy and the price index index for the nonfarm business business sector, and two dummies for the effects of the freeze and and Phase II and for the subsequent subsequent catch up effects. The St. Louis equation yields yields values for nominal income; income; the the inflation reduced reduced form is employed to generate price level level predictions; and the price level is used to to deflate nominal nominal income to to yield real output output predictions. The results in in Charts 33 and 44 depicted depicted by by the line labeled labeled StL3, reflect the response to the same phased monetary decelerdeceleration St. Louis Louis model described ation employed employed with with the the other other St. model simulations simulations described above. Note the similarity with the St. St. Louis Louis results with with aa Phillips 71/Il), StL2, curve (based on the the sample period through 71/11), StL2, in in Charts Charts 33 and 4. With the reduced-form equation equation inflation inflation declines more rapidly, rapidly, by about .20 - .30 .30 percentage over most period; corabout .20 percentage points points per per year year over most of of the the period; cor- respondingly, the output loss is somewhat somewhat smaller. But the time pattern and magnitude magnitude of of both the deceleration deceleration in inflation and the cumulative output loss are remarkably similar. Again note that the output loss per quarter has not peaked after six years of the phased phased deceleration deceleration so so that the cumulative output output loss loss is still rising raoidly rapidly at the end of six six years. Qualifications of the Empirical Empirical Analysis Analys_is_ Qualifications The results reported above are derived both both from explicit explicit Phillips curves, and from monetarist reduced-forms. The existence existence of aa cumulative output loss associated with eradicating inflation is -78- therefore therefore generally consistent with with both income-expenditure income-expenditure structural models and monetarist reduced-forms. reduced—forms. The emThe major deficiency of the em- pirical analyses on which the the results described described above are based is the failure to to allow allow the public’s public's perception of current current and future policy policy to affect expectations about future inflation. inflation. The Credibility Effect The results reported above based on Phillips curves all related inflation in the current period to aa distributed lag on past inflation rates where the latter are intended to reflect the rate rate of inflation expectations exexpectations (and/or direct the influence of past inflation as as for example via catch—up catch-up effects). This specification does not allow allow the degree of credibility associated with announced anti—inflation policies degree of credibility associated with announced anti-inflation policies or or even the the expected expected influence influence of recent recent policy actions to influence influence inflation expectations. The estimates estimates of cumulative output loss gengen- erated overerated by by such such models models are, are, therefore, therefore, almost almost certain certain to to be be over- estimates. Fellner (1979), (1979), for example, maintains maintains that that" ... the ... standard model coefficients coefficients... ... would would change significantly significantly for the the better -- in in the the direction direction of more rapid reduction of better of aa much much more rapid rate rate of of reduction of -- inflation given slack inflation for for any any given slack -- if if aa demand demand management management policy,.. policy ... -- changed to aa credible policy of consistent demand disinflation." disinflation.” But by much does does the overestimate inflationary by how how much the standard standard model model overestimate inflationary inertia? inertia? By 10 10 percent, 50 percent? We do not have any reliable quantitative quantitative estimate of the degree deoree to which the deceleration deceleration of of inflation to which policymakers policymakers can can speed speed the inflation by by clearly defining their anti—inflation anti-inflation policies and convincing the public that they intend intend to to follow through. —79— -79- Nevertheless, Nevertheless, there would be nearly universal agreement that anti—inflation anti-inflation policies ought to be be set out Clearly clearly and supported by both the Treasury and the the Federal Federal Reserve in such aa manner as to maximize maximize the Credibility credibility effect. Rational Rational Expectations and the Cumulative Output Loss of rational expectations models advocated, In the extreme form of for example, by by Sargent and Wallace (1976), (1976), the cumulative output loss associated with aa credible credible policy of monetary deceleration should be zero. zero. These models have two essential features: 1) they are equilibl) equilib- rium models in which prices respond immediately and fully to monetary monetary change and real variables variables such such as as unemployment unemployment and output output respond resoond only only to unanticipated inflation; and 2) inflation expectations are formed rationally, taking into into account knowledge both about the structure structure of the economy and the systematic features of policy. In such aa model, inflation should should moderate imediately immediately in in reresponse to to the monetary monetary deceleration, deceleration, provided, of course, course, that that the policy was announced in in advance and believed (or otherwise otherwise expected). We had had thought thought of running running simulations with an an RE version of the the St. lines suggested suggested by by Andersen (1979). Louis model along lines On aa moment’s moment's sufficiently obvious that computer computer reflection, the implications were sufficiently simulations could be dispensed ,dth. with. The St. Louis model has has aa Phillips curve in which inflation inflation depends on aa demand variable (x) and expected inflation (Pe) an adaptive adaptive expected (pe) where the latter is determined from an expectations expectations model with weights taken from aa regression of of the nominal interest rate on past inflation rates: (5) p P = + Sx + ~P -80-80- Andersens RE version imposes the condition that Pe Andersen's = E(P); E(P); i.e., i.e., that that subjective inflation expectations equal inflaequal the model’s model's forecast forecast for inflation. In this case: (6) C E(P)a+Sx+eE(P) a + sx + sE (Pl (P) ( 6' ) (6’) E E (P) C l (a+ S x) l~E(sx) l -s and Andersen substitutes substiti~tes (7) p =~(a+ S x) 1-c for the St. Louis Louis Phillips curve. curve. = .86, its value in the St. Louis model. Andersen sets c~ = HowHow- ever, if if cc is meaningfully viewed in this case as as the coefficient on expected expected inflation, inflation, the value of .86 estimated in the St. St. Louis Louis model should not be be accepted as the magnitude magnitude of that that parameter parameter in the the RE RE version of the St. St. Louis model because the value of of cc was estimated estimated under under the assumption that expectations were formed fanned adaptively. Taking 1, as seems essential to to the RE model, model, equation 77 no longer longer is aa c == l, meaningful equation equation for for P. P. meaningful (6’) (6') Instead we obtain from c = l Instead we obtain from (6) (6) where wheres= 0 = ct + sx O=a+sx so that there is aa unique value of x* = - a/s s/B corresponding, of course, course, to the natural natural rate of unenployment. unemployment. xx can differ differ from x* only on on — account of random disturbances (with zero zero mean). In In this this case any effect effect of monetary deceleration on on the rate rate of growth of nominal income transformed immediately and fully into is transfonned into aa decline in in inflation inflation without any cumulative output loss. This This seems to us a more -81-81- meaningful RE version of the the St. St. Louis Louis model than than that employed by Andersen. 9 Andersen Balancing the Gains from Reducino Inflation~g~jpstthe Transitional Balancing the Gains from Reducing Inflation Against the Transitional Costsi0 Costs ~ The cumulative output loss is is aa measure of the cost of of antiinflation policies. To To evaluate the desirability of of such such policies we we also need to assess the gains from reducing inflation. Unfortunately, the costs of inflation inflation (and hence the the benefits benefits of reducing inflation) inflation) are not as or easily easily quantifiable cost of unemployment. are not as clearcut clearcut or quantifiable as as the the cost of unemployment. Fischer Fischer and Modigliani (1978) provide aa careful careful outline of the costs of inflation. The costs include the the welfare loss loss associated with the incentive reduction in capital acincentive to to economize economize on on cash cash balances, balances, the the reduction in capital ac- cumulation due to disincentives for saving saving and investment that reflect after—tax the way in which the tax system system permits inflation to affect after-tax 9There is aa second second and related objection objection to to Andersen’s Andersen's approach. In the the St. St. Louis model ca is not the sum of the coefficients coefficients on on lagged inflation the sum of the inflation rates. Indeed thesum the coefficients coefficients is generally generally about 1.0. The reason for this curve does 1.0. this is that the St. St. Louis Phillips Phillips.curve does not estimate the weights on on lagged inflation inflation directly directly within the estimation of of the Phillips Phillips curve itself. First, First, an an equation equation for aa short-term interest interest rate rate is estimated as as aa function of of the rate rate of of monetary monetary growth growth and distributed lags on on both both the rate of change in in output and on on past past inflation inflation rates divided by the ratio of unemployment unemployment to the full— fullemployment rate. The sum sum of of the the coefficients on lagged prices from from the interest rate equation in the original Andersen/Carlson Andersen/Carlson article was 1.27 1.27 so so the sum of weights on on lagged inflation inflation rates in the Phillips (1.27/(u/uf)), approximately 1.0. The sum curve is .86 .86 (l.27/(u/uf)), sum of the the inflainflation coefficients from the interest rate rate equation equation vary vary considerably considerably estimate of over different sample periods periods and the the estimate of ac always always compensates compensates to yield yield a sum sum on past inflation rates of about about 1l .0. This This reinforces our view that that the value of as in equation (6) should be taken as 1.0. 1.0. 10This section was added to the original paper and was motivated lOThis section was added to the original paper and was motivated by tzer at by comments by Jerry Jordan and Allan Mel Meltzer at the conference. conference. -82-82- redistriburates of return and the cost of capital, and the arbitrary redistribution of income and wea1th due to unanticipated inflation. wealth due While Fischer and Modigliani Modigliani do do provide estimates of of some compocompo- nents of the costs of inflation, inflation, neither their study nor others others permit permit us to compute aa meaningful estimate estimate of of the benefits that would would accrue from reducing inflation inflation which which could in turn be compared compared with the cost cost in in terms of cumulative output loss. What we we can compute is the minimum minimum size of of the permanent gain gain in in output per year due to to eradicating eradicating inflainfla- tion which would just justify incurring the cumulative cumulative output loss assoassociated ciated with the transition to price stability. per year. benefits as a gain in real output per We will refer to the the of the gain Some components of may, however, be be welfare or or utility gains that would would not necessarily necessarily show up in computed measures measures of real output. While While such welfare gains gains difficult to evaluate evaluate than than output gains, they are no no less are even more difficult in developing developing aa measure of the benefits of reducing inflation. inflation. important in Figure l1 depicts depicts the comparison we wish to make. Figure dashed XX The dashed line is the rate of growth of (potential) output if if inflation inflation remains Figure Figure 11 x X — t n 0on -83—83- indefinitely at at 7.5 percent. percent. If anti-Inflation anti-inflation policies are are pursued, If output is assumed to follow the solid solid line. occur between between t = 0 and t O = The transitional costs nn as unemployment rises above the rate associated with potential output. However, if if there are costs of of inin- flation, output will rise above the level that would have prevailed prevailed if the initial initial steady inflation rate rate had continued. continued. We define as the define GGas present value of of the permanent per period output gain, gain, evaluated from period nn to (8) (8) oo GG = r i=n (l+r)’ This can be be compared compared to to the the present value of the cumulative cumulative output output loss (L) ( L) (9) LL = n—l L. n-1 z E i=O (l+r)1 i0 where L~ Li is the output 1loss ass in the ith ith period ((i=0, i =0, . . . n—l). n-1). . . . Assuming Assuming that the unemployment rate is maintained above the rate consistent with potential output by aa fixed amount for nn periods, periods, the the loss in period i can be be expressed expressed as ((10) l O) L~ E (1+p)1 where [U is the loss in the first period and oa is is the the rate of growth in potential potential output. ((10’) 10' ) If r=p, the expression for LL simplifies to L L == n[ nTi This is precisely precisely the the way way we calculated the discounted value of of the cumulative output loss above for the Perry and Cagan equations. -84-84- To simplify further, we ~ for all ii we assume g~ gi is is aa constant g > > n. g which first equates the cost of unWe then then solve for the value of g unemployment employment and the gain from eradicating inflation -- the minimum minimum value -— of the permanent per period gain from eradicating eradicating inflation inflation that would justify justify incurring the transitional costs. The The value of ~ g for the Perry, Stein, and Cagan results are presented presented in in Table Table 8; we calculated calculated them under the assumption of aa 3.3 percent discount rate and for two TABLE 88 The Minimum Value of the Per Period Period Gain that Justifies Eradicating aa 7.5 Percent Inflation Rate Value of ~ g (billions of 72 $) 3.2 3.2 2.5 Equation/ Model Perry Perry Cagan Stein Stein Stein 11 22 73.0 70.9 70.9 16.66 16. 31. 31.00 25.4 25.4 11 22 57.0 57 .o 55.4 13.0 24.2 19.8 respecalternative values of the Okuns 0kun's Law coefficient coefficient (3.2 and 2.5, respectively). The minimum value of ~ g varies from $13 $13 billion per year year based on Cagan’s Cagan's Phillips curve to $73 billion based on the Perry’s Perry's Phillips curve under aa moderate policy. Note that this this analysis analysis provides an alternative perspective perspective on Mote the case for gradualism. Under gradualism, the costs may may be be reduced if if the Phillips curve is nonlinear. But the benefits are are also more more gradual (in our analysis, postponed until until inflation is eradicated). Thus, we find that although the costs are smaller under the gradual gradual pol-~ pol1), the size size of of the per period period gain gain icy using the Perry equation (Perry 1), -85- justify eradicating inflation inflation is is smaller under under the more required to justify radical policy (Perry 2). The radical policy also yields yields aa smaller minimum per period gain using the the Stein model, although this result result was expected in this case because the cost turned out to be lower in in the using Stein’s Stein s model. radical case using 1 The calculations reported above presumed that the gains from rereducing inflation could be be meaningfully represented represented as as aa fixed real real sum per period. What if the gains are more meaningfully meaningfully specified specified as as aa real real sum which grows at the the same rate as potential output? For For example, the cost of aa fully anticipated increase in in inflation inflation is is generally generally measured by by the reduction in the area under the demand curve for money money balances as as wealth owners reduce their demand for money money in response to the associated rise in in nominal interest rates. rates. The The decline in in demand for real generally viewed real money due to to aa rise in in the interest rate rate is generally as proportional to to the overall scale of money holdings which, in turn, is determined by by the level of transactions (e.g. real income). The cost of a given rate of inflation and and hence the the benefits benefits of eliminating eliminating the inflation may therefore grow grow at the the rate rate of of increase of of potential output. (8’) (8') In this case where where ~ g is is the value of the the gain in in period nn (the G= ; G=~ i=n i=n g (1+r_l2_ (l+r)i1 (l+r) first period in in which aa gain is is registered). For r~ .: :_ r, GG -+~ co. > 00 • corresponds to the result recently recently derived by Feldstein (1979): (1979): This if the cost of inflation grows at aa rate rate equal to or or greater than the the discount rate, any positive initial gain (any ~ g > > incurring incurring any finite transitional cost~ cost'. -86- 0) is sufficient to justify resu1ts suggest that the case for anti—inflation ctnti-inflation policies These results should not be dismissed lightly, even when there are large transitional costs of eradicating Inflation. inflation. The range of the estimates of the cumulative output loss, the uncertainty about the adjustment in those results required to allow allm,· for the tr12 credibility effect, and the lack of aa quantitative estimate of the cost of Inflation inflation makes it extremely difdifficult fic.1lt to make a meaningful comparison :::omparison of the costs and benefits of 1 anti—inflation anti-inflation policy. It should not be surprising therefore therefore that that indecisive and often lacking In in coninitment commitment policymakers generally seem indecisive to reduce Inflation. in-~latior.. Narrowing the range of estimates of output loss the cost of inflation should be high on the and developing a measure of the priorities for macroeconomic research in in the 1980s. RULES VERSUS ACTIVISM two propositions. The case against activism rests on two The first µrivate sector of the economy is inherently proposition is that the private stable. This is a major tenet of monetarism and suggests the absence of a need for •or stabilization policy. Indeed, monetarists generally concon- tend that the iinstability nstabi1 i ty observed in the economy results mainly mainly from government rather than private sector decisions. stabilThe inherent stabil- Ity of the private sector results In ity in part from the absence of large and persistent exogenous shocks and and in part from the fact that the shocks that do occur have relatively small and only temporary effects on outout- put and employment as a consequence of the economys economy s built-in stability. 1 The second proposition in the case against activism is that even if the economy were subject to cumulative <umulative movements In in output, employemploy- ment and inflation relative to target levels, discretionary policy -87- might only compound the instability rather than dampen it. The danger The that that policy will turn turn out to be be destabilizing follows from the long inside lag, the long and variable outside lag, and the general unceruncertainty about the the effect of policy policy on on the economy. economy. The case for activist activist policy involves involves aa rejection of of the the two two proppropositions developed above; above; the economy needs to and can be be stabilized by appropriate manipulation of policy policy instruments. The first proposition in is that the economy is subject to in support of policy activism, activism, then, is to substantial and persistent sector. substantial and persistent disturbances disturbances arising arising from from the the private private sector. In addition, addition, nonmonetarists contend that policy policy can can be implemented with sufficiently sufficiently short inside lags and with sufficient precision qiven our our understanding understanding of the the structure of the the economy economy to to yield an improvement in economic performance relative to to aa policy of aa fixed rule. rule. Relevant empirical evidence on on rules versus activism activism includes: (1) the relative size of exogenous exogenous impulses arisinq arisinq from policy and nonpolicy sources policy and nonpolicy sources (2) the degree of persistence in the response to such such disturbances disturbances (3) the ability ability of active policy to improve improve economic economic perperformance in the face of the disturbances. formance in the face of the disturbances. Stability of the Private Private Sector The issue of the stability of of the private sector sector has been categocategorized as aa fundamental difference between monetarists and the convenbetween mon~tarists conventional Keynesian tenets (See Andersen (1973) and Mayer Mayer (1975)). (1975)). Nevertheless, it appears to be be an an issue on which little, if any, relerelevant vant empirical evidence evidence is available. The evidence that is conventionally conventionally cited in in response to the allegation the Keynesian Keynesian position as allegation that that the position regards regards the the private private sector sector as -88-88- of simulation simulation experiments with inherently unstable is the result of various econometric models. These experiments suggest suggest that the models are stable, usually exhibiting highly damped oscillations back back to equilibrium following some some shock (see Klein (1973)). Such results under the postulated experimental conditions conditions are probably aa necessary necessary condition, but not aa sufficient condition to substantiate substantiate the the monemonetarist proposition. We would need need to look at at the the degree of damping under aa policy of fixed rules relative to the damping under an endogeendoge— nous policy with feedback from current economic developments. The case The for rules is enhanced if if endogenous endogenous policy reduces the the degree degree to for rules is enhanced policy reduces to which which disturbances are damped. Evidence from Model Simulations Discussions of the effectiveness of of policies often focus on the size of po policy multipliers. 1icy multi p1 i ers. Such Such measures of the the leverage 1everage of policy on goal variables variables are critical to to setting setting policy, but do do not provide any evidence on on the usefulness usefulness of of discretionary policy unless they are zero. Indeed as Cooper and Fischer demonstrate, even even if if the the policy instrument has aa zero cumulative multiplier it may be useful as aa stabilization stabilization has tool as long long as it has aa nonzero short-run multiplier. tool important More important is is the pj4jç~jjit predictability of of the outcome of policy policy actions which which is is more the goal variables. closely related to the errors in forecasting the The The case case for discretion, therefore, has little little or nothing to to do do with with the size of policy multipliers, moving multipliers, unless unless there is some concern about about movinq the “penalty the policy variables too far or too fast such as as when aa "penalty function" is added to the "goal function” “goal function." function.” The time pattern pattern of the the response as well as the predictability multipliers, on predictability of the policy policy multipliers, —89— -89- the other hand, do matter. discretlon, thereEvidence on rules versus discretion, fore, generally involve model simulations simu1ations and these are most useful If if allowance is made made for uncertainty about the the multipliers. multip1iers. Below we review the evidence on the comparison of of economic economic perperfonnance formance under rules and discretion based on simulations with macromacroeconomic models. First we must define a set of alternative policies; four alternatives have been Investigated. investigated. 1) Actual policy: instruments Historical simulations in which policy Instruments take on their historical values provide the benchmark of actual policy, discretion as it was implemented as opposed to what would have been optimal in the context of the model under consideration. 2) 2) Fixed rifles r~les or rules without feedback: FIxed in which the Simulations In policy instrument instrument is constrained to grow at a constant rate provide evidence on the effect of fixed rules; for example, example~ a constant rate of monetary growth as advocated by Friedman. In this case case. the policy policy inin- strument is totally independent of current economic developments. 3) Active rules or rules with feedback: /J.,n alternative to both disdisAn cretion and fixed rules is an active rule or a rule which ~thich requires policy instruments to respond systematically to current economic develdevelinvolving opments. This approach introduces Phillips type ad hoc rules involving proportional and derivative controls. Some experimentation is underunder- taken to identify “good” good" rules but short of full optimization. 11 Such Such simulations can be viewed as a way of modeling systematic systematic discretionary policy without the blatant policy errors that in retrospect always mar mar the historical runs. 4) Optimal control: The benchmark for identifying the best that is possible under discretionary policy is an optimal control simulation simulatio~ in -90—90- inwhich policymakers are viewed as selecting a time path for their instruments that that minimizes minimizes the losses associated with deviations deviations of their goal variables from their target levels. It, therefore, therefore, requires imim- posing an an explicit exolicit loss loss function inc1uding designation of of relative including the designation weights miniweights on competing objectives and solving the model subject subject to minimization of of the losses. 1osses. The an ininThe solution allows the selection of an strument path to to reflect knowledge of of the the structural structural parameters parameters of of the model and forecasts of future performance based on current and past values of exogenous exoge!'lous variables and the dynamic structure of the model. A superior eccmomic performance performance under superior eccnomic under such such circumstances circumstances hardly hardly proproA vides convincing support for discretionary policy, policy, although although it it provides of tne potential for discretionary policy to to improve economic economic evidence of performance. The various policy regimes can be difbe simulated in aa number of different ways. In In aa deterministic deterministic simulation the error error terms in the various estimated equations are set to zero. various esti111ated zero. This immediately removes aa potentially important important source of instability in the private economy economy and should be be expected to bias bias results in favor -Favor of fixed rules. There are two two basic types of stochastic stochastic simulations reflecting the two sources of random disturbances: estithe additive error error terms in the the esti- mated equations equation2 and the estimated coefficients. Simulations allowing allowing for random additive error disturbances are generally labeled stochastic simulations while those that randomize both both parameters parameters and additive errors are referred to to as as fully stochastic simulations. —91— -91- Actual Policy Versus Fixed Monetary Growth Rules Modigliani reports reports two simulations simulations with aa fixed monetary growth rule rule over the period beginning in 1959 1959 and ending in in mid-1971. mid-1971. at aa 33 percent percent annual rate. case Ml is constrained to grow at In each In the first simulation all shocks are eliminated by by substituting constant trends or means for untrended exogenous variables. variables. In the second, second, hishis- torical values of exogenous exogenous variables are employed. In the first case case the monetary monetary rule stabilizes the the economy, economy, but, allowing allowing for historical shocks the economy "was “was distinctly less less stable stable than actual experience, by aa factor of 50 50 percent percent [p. 12].” 12]." Eckstein investigates the the implications of of smooth growth in in non— nonborrowed reserves over the period of 1964 through 1975. 1975. (Nonborrowed ‘64, accelerate reserves grow at at aa 44 percent rate in '64, acce 1erate 1/4 1/ 4 percent percent point each year year until until they stabilize at aa 66 percent rate rate during and after after 1972). 1972). Eckstein finds that smooth growth in reserves does result in in “a more stable growth pattern” "a pattern" but does not dramatically dramatically alter the overall results for economic performance. Versus Fixed Rules Rules Active Rules Versus In aa series of papers employing simulations with both the MPS MPS and l972b, 1974) compare St. Louis models, Cooper and Fischer (1972a, (1972a, 1972b, Phillips type feedback control rules with with fixed growth rate rules. They conclude that that there are active rules which dominate fixed rules for both models, under deterministic, stochastic and fully stochastic stochastic simulations. The dominant active rules generally generally involving involving strong derivative controls and some proportional control. criterion was The criterion the average standard deviation deviation in the unemployment unemployment and inflation rates. rates. —92-92- For the St. St. Louis model, for example, the average standard deviations for each variable were reduced reduced by about 20 percent percent in in the deterministic deterministic simulations (over the period 56/1-68/IV), 56/I-68/IV), between between 50 - 70 percent percent in in - the stochastic simulations (over the same same period) period) and and by about 50 50 perpercent in the fully stochastic stochastic simulations (over (over the period 55/I 55/1 - 7l/IV). 71/JV). — The improvement v-ws was more modest, however, model , where however, in the MPS model, where the standard deviation deviation of unemployment fell by by 44 - 24 percent and that of - 32 percent in stochastic simulations simulations over the period inflation by 77 - 32 - 1956/I - 68/IV. 1956/l 68/l V. — Optimal Control Simulations There have been been numerous attempts to compare fixed rules with optimal control simulations including Chow Chow (1972), Garbade Garbade (1975), Cooper and Fischer (1975), Crane, Ravenner Havenner and Tinsley (1976), and Crane, Havenner and Berry (1978). studies find that that The first four studies fixed rules are uniformly inferior to optimal control control (and generally inferior to historical historical policies). These simulaThese studies use use stochastic simula- tions but actual actual values of exogenous variables variables and, with the exception exception of Cooper and Fischer, constant parameter values. Garbade Garbade for example example finds that “discretion,” discretion, in the form of optimal control, reduces reduces the 11 11 50 percent compared to aa fixed rule, aa result in close expected loss loss by 50 agreement with Chow. Chow. Garbade views views his his results as as adding to to the “accumulating evidence” !!accumulating evidence" of of the gains gains associated associated with discretion “when when aa 11 valid representation of of the economy economy is available.” available." But that, after all, all, is the major major element in in the controversy. Cooper and Fischer find find that their active rules perform perform quite well in relation to to optimal control solutions using the St. St. Louis Louis model. -93—93— Costs are reduced by about 45 percent relative to fixed rules, but fixed rules outperform historical policy In in this case due due in part to greater instability in instrument movements movements in the latter case. The Cooper-Fischer paper produces a possibly valuable insight about the Cooper—Fischer relative performance of rules and discretion. Stochastic simulation requires multiple simulations for alternative realizations of the the stochastic disturbances. They found that the poor overall performance of fixed rules resulted from from their “spectacularly "spectacularly bad” bad performance performance in 11 replications where where losses turned out to be above average for all all policies. Where average performance Is is good, on the tf!e other hand, hand, fixed control. rules perform about as well as optimal control. imoly that This may imply optimal policy is nonlinear—restrained nonlinear-restrained to t8 fixed rules within within a band around target values of goal variables and active only outside those bands. Thus, “fine tuning is rejected, but activism In in the face of aa ' fine tuning” 1 11 major disturbance disturbance has a substantial payoff. This conclusion is reinforced by the Crane, Havenner and Tinsley 1911/1-1974111 period using a condensed version of study of the 1971/1-1974/11 of the MPS model, MINNIE. Optimal policy is not especially volatile after an initial aggressive expansionary policy pol-icy in the first first two quarters to to offset the recession implicit in the initial conditions. The optimal policy again dominates fixed rules, in this case by about 40 percent; and fixed rules would have increased expected losses by about abnut 45 perper- cent relative to historical policies. Rational Expectations and the the Limits of Activist Policy The traditional arguments against activist policy focused on the Implications implications of long inside lags, long and variable outside lags, and -94- multiplier uncertainty; there was a general emphasis on the limitations of policy po1icy in an environment characterized by insufficient knowledge knowledge of the economys economy 1 s structure. The Lucas—Sargent—Wallace Lucas-Sargent-Wallace rational expectaexpecta- tions models suggest aa dramatically different different basis for fixed rules. These models suggest that policy is doomed doomed to ineffectiveness ineffectiveness in an environment in which economic economic agents have knowledge both about the structure structure of the economy and the way in which policy authorities rerespond to economic developments. In this this case case too much knowledge rather than than too little little knowledge knowledge underlies the ineffectiveness of policy. Real variables according to to these models respond only to unanticipated price or or inflation shocks. produce produce surprises. Systematic policy, by definition, cannot Therefore, although there exists aa trade-off trade-off bebe- be systemsystemtween unanticipated inflation and unemployment, it cannot be exploited by policy authorities; this atically exploited this is is generally generally referred to as the neutrality proposition. The theoretical structure of these models and the implications of of aa number of qualifications, qualifications, particularly particularly of nominal contracts, contracts, have been been thoroughly developed developed in the existence of the paper by by Taylor. The role, operational operational specification, and implicaimplica- tions of rational rational expectations in macroeconomic models is the the central issue issue in in macroeconomic theory theory today and and empirical empirical investigations investigations of of these models is certain to SOs. There to be be the growth industry of the the '80s. are, however, only only aa handful of of empirical studies to date date that attempt attemot to to test the the neutrality proposition. Mccallum (1979) (1979) in aa recent survey of this literature notes that McCallum the formal evidence is not inconsistent while "the inconsistent with the neutrality proposition. proposition . . . the power of of existing tests tests is not high and, in any any . . event, the the evidence evidence is not entirely entirely clearcut. clearcut." -95— -95- The two most most important important empirical studies are the Barro Barro papers (1977, 1978) on the effect effect of of empirical unanticipated monetary growth on output and Sargent’s Sargent's on unemployment and output paper (1976) applying Sims and Granger tests for causality causality to movements movements in the unemployment rate, the money supply, government government expenditures and other other macro variables. Barro estimates estimates aa reaction to isolate Barro reaction function function to isolate unanticipated unanticipated monetary examines the role role of of unanticipated unanticipated and and anticianticimonetary growth growth and then examines pated monetary monetary change on unemployment and output. His results results are are rere- markably one sided, supporting the hypothesis hypothesis that only only unanticipated markably policy actions affect real variables. But But his his empirical methodology has been convincingly critiqued critiqued by been convincingly by Small, Small, Fischer Fischer (1978) (1978) and and Gordon Gordon (1979). {1979). Sargent is somewhat more cautious in interpreting interpreting his findings as indicating indicating that “the "the causal structure imposed on the data by the the at variance classical model. model . . . is not obscenely at variance with with the data data [p. 233].” 233]." . . We think this this means the results are mixed, which indeed they they are. There movements in There is some some evidence, evidence, for example, that that movements in the money money supply “cause” movements in the unemployment rate (using the Granger test) and "cause" movements some evidence that it does not (using the the Sims test). Summary The evidence accumulated over the ‘70s '?Os has has at best only only aa consensus over the gains associated with modest role in increasing the consensus activist policy. of the ‘70s '?Os has clearly eroded eroded the the The experience of optimism about the potential activist policy that characterized the apparent success of of the 1964 tax tax cut and the long expansion of of the ‘60s. '60s. There is wider recognition today compared compared to the mid-l96Os mid-l960s among proproponents ponents of active policy policy of of the limitations of of active policy policy and and the the -96-96- difficulty of “fine "fine tuning” tuning" the the economy by responding responding to even even small difficulty departures of output and employment from target levels. Active Active policy, policy, however, continues to have wide support in situations where where aa sizable displacement has occurred, as in in the the 1973—75 1973-75 recession. On the other other of rules, such such as Friedman (1968), also allow for hand, many proponents of the 14].” the use of discretionary policy to to offset “major "major disturbances [p. 14]." Therefore, the the gulf gulf between proponents of rules and activism is not nearly so great as as it might at at first appear. The optimal control aggressive studies have helped to emphasize the potential usefulness of aggressive policy action when initial conditions are far away from targets and the limited potential usefulness of activist policy in response to smaller displacements. This lesson is is perhaps one on on which proponents of rules rules and activism can agree. CONCLUSION 1 As As the ‘7Os 70s began, the monetarist—income monetarist-income expenditure controversy controversy in macroeconomics. was aa dominant theme in Particularly after the MPS and Particularly other large scale scale models began churning out large values for monetary the controversy controversy focused in on the size of fiscal policy multipliers, the multipliers, particularly the fiscal multipliers on nominal GNP. The econometric evidence of the ‘7Os '7Os has has not fully resolved this this issue, i.e., there are those who continue continue to be persuaded persuaded by the St. Louis equation results. And while while this evidence evidence questioning the the reliability reliability of the fiscal multipliers in the St. St. Louis equation undoubtedly undoubtedly has rereinforced the views of the skeptics, skeptics, it it has not necessarily necessarily shaken the confidence of of the equation’s equation's supporters. —97— -97- '70s began, the orthodoxy orthodoxy of aa Phillips curve embodying aa As the ‘lOs stable trade—off trade-off was was under under an attack it it did not survive. After aa trantran- sitional period, evidence evidence mounted mounted in in support of aa vertical long-run Phillips curve. Thereafter, the issues contested contested have have been the nature and sources of short—run trade-off of any short-run trade-off and the implications for the output output loss loss of of eradicating eradicating inflation. inflation. The The econometric econometric evidence from from aa wide range of and models models suggests monetary deceleration deceleration wide range of sources sources and suggests that that monetary can eradicate inflation, but not quickly quickly and not not without large costs in in terms of cumulative output loss. loss. The major major unresolved issue is the significance significance of of the credibility effect effect and the the degree degree of of overestimation overestimation in the cumulative cumulative output loss due to the the failure to take in the output loss due to failure to take into into account account the effect of recent policy actions and and expected expected policy of recent policy actions policy actions actions on on ininthe effect flation expectations. While fine-tuning may have few advocates, advocates, the evidence from model simulations are likely likely to simulations suggests suggests there there are to be be considerable considerable gains gains to to activism activism when the economy is far away from targets and in response to very large shocks, shocks. Rules or alor activism remains remains an important issue al- though the case against activism has been been broadened by the development of rational rational expectations expectations market clearing models. -98- REFERENCES Allen, S. and T. Seaks, (forthcoming), “Nixon ''Nixon and Ford Ford as Ex—Post Ex-Post Keynesians,” Journall of Macroeconomics. Keynes i ans, ~ourna Macroe~s,_n_om~. ll L. (1973), “The "The State of the Monetarist Debate,” Debate," Federal Andersen, L. Reserve Bank of St. St. Louis Review f<e_view (September). Andersen, L. and K. Carlson (1970), (1970), “A ''A Monetarist Model for Economic Stabilization,'' Federal Reserve Reserve Bank Bank of of St. Stabilization,” St. Louis Review (April), 7-25. 7—25. “Monetary and Fiscal Andersen, L. L. and J. L. Jordan (1968), "Monetary Fiscal Actions: AA Test of of their Relative Importance Importance in in Economic Stabilization,” Stabilization,!! Federal Reserve Bank of 11-24. of St. Louis Review (November), (November), 11-24. “Rational Nonrational Andersen, P. ((1979), 1979) , "Ra ti ona 1 Expectations Forecasts Forecasts from Non rat i ona 1 Models,” Models," Journal of of Monetary Economics (January), 67-80. Barro, R. (1977), "Unanticipated Barro, “Unanticipated Monetary Growth and Unemployment in States,” American Economic Review (March), the United States," (March), 101-115. Barro, R. (1978), "Unanticipated Level in in Barro, “Unanticipated Money, Output, and the Price Level States," Journal of Political Economy Economy (Auqust), (Auoust), the United States,” 549-580. 549-580, “The Reduction of Inflation and Cagan, p. P. (1977), ''The and the Magnitude Magnitude of Unemp 1 oyment" in Wi 11 i am Fellner Fe 11 ner ed. ,, Contemporary_Economic Contemporary Economic Unemployment” William Problems 1977, Washinqton, D.C., 1977, American American Enterprise Institute, Institute, vJashinqton, 41-SO. 41-50. by Slack Slack Demand” Demand" in Cagan, P. (1978), "The “The Reduction of Inflation by ed. , Contemporary Economic Problems 1978, William Fellner ed., 1978, American Enterprise Enterprise Institute, Washington WashTn-gton D.C., D. C., 13-45. Carlson, K. (1978), “Does "Does the the St. St. Louis Equation Now Now Believe in Fiscal Fiscal Carison, K. Policy?” Policy?" Federal Reserve Reserve Bank Bank of St. St. Louis Louis Reyiew Review (February), (February), 13-19. 1319. Chow, 0. “How Much Could Be Gained by Optimal Stochastic G. (1972), "How Stochastic Control Policies?'' Annals of Economic and Social Measurement Measurement (October), Policies?” 391-406. Chow, G. (1977), (1977), “Usefulness "Usefulness of Tmperfect Imperfect Models for the Formulation of Stabilization Policies," Policies,” Annals of Economic Economic and Social Social Measurement (Spring). ~ Christ, C. (1975), ( 1975), “Judging "Judging the Performance of of Econometric Models of International Economic Review (February). the U.S. Economy,” Economy,u Jnternational (February). Congressional mdc tandjn Fiscal Congressional Budget Office (1978), (1978), Understanding Fiscal Policy. —99— -99- Cooper, J. J. P. and S. Fischer (1975), “A ''A Method of Stochastic Stochastic Control of Nonlinear Econometric Models Models and an Application,” Application/ 1 Econometrica Econometrica (January), 147-162. (January), 147-162. Cooper, J. P. and S. Fischer (1974), (1974), “Monetary "Monetary and Fiscal Fiscal Policy in aa Fully Stochastic St. St. Louis Econometric Model Model," Journal of of Money, ,“ Journal Credit and Banking Banking (February), 1—22. 1-22. Cooper, J. P. and S. Fischer (1972), (1972), “Simulations "Simulations of of Monetary Rules on the the FRB-MIT—Penn FRB-MIT-Penn Model,” Model," Journal of Money, Credit and Banking (May), 384-396. 384-396. (May), Cooper, J. J. P. and S. S. Fischer (1972), ''Stochastic Mone“Stochastic Simulation of Monetary Rules in Two Macroeconomic Models,” Models," Journal of American American Statistical Association (December), (December), 750-60. Statistical_Association Corrigan, E. C. (1970), (1970), “Measurement "Measurement and and Importance Importance of Fiscal Policy Changes,” Federal Reserve Bank MonthjL3~yIeyL(June) Changes," Bank of New York Monthly Review (June), 133-45. Crane, R., A. Havenner, Havenner, and J. J. Berry (1978), (1978), “Fixed "Fixed Rules versus Activism in the Conduct of Monetary Monetary Policy,” Policy," American_Economic American Economic Review (December), 769-783. Crane, R. R.,, A. Havenner, and P. Tinsley (1976), “Optimal "Optimal Macroeconomic Control Policies,” Annals of Control Policies,u of Economic and Social Measurement 191—204. (Spring), 191-204. “The Channels of Policy,” deLeeuw, F. and E. Gramlich (1969), (1969), "The of Monetary Monetary Policy," Federal Federal Reserve Bulletin, Bulletin, 472-91. EconodeLeeuw, F. and E. Gramlich (1968), (1968), “The "The Federal Reserve—MIT Reserve-MIT Econometric Model,” Model," Federal_Reserve Federal Reserve Bulletin (January), 11-46. 11-46. (1969), "Monetary “Monetary and Fiscal Actions: deLeeuw, F. and J. Kalchbrenner (1969), A Test of A of Their Relative Relative Importance Importance in Economic Stabilization Stabilization -— -Conmient,” Comment," Federal Federal Reserve Reserve Bank of St. St. Louis Review (April), (April), 6-11 6-11. Dewald, W. and M. Marchon “A Modified Marchan (1978), "A Modified Federal Federal Reserve of St. Equation for Canada, France, Germany, Italy, the the Louis Spending Equation United add the United States,” United Kingdom, and States," Kreditund_Kapital. Kredit und Kapital. Eckstein, D. 0. (1978), The Great_Recession, Great Recession, North Holland, Amsterdam. L. R. R. Klein (1968), The Wharton Econometric Econometric ForeEvans, M. K. and L. Fore- casting Model, Model, Economics Research UnTf, Unit, University University of Pennsyl— Pennsylvania, vania, 2nd ed. Feldstein, M. (1979), "The of Permanent Permanent Inflation Inflation and Feldstein, “The Welfare Cost of Optimal Short-Run Economic Policy,” Policy," Journal of Political Political Economy (August), 749-767. (August), -100-‘Do- "The Credibility Effect and Rational Expectations Fellner, W. (1979), “The — Implications of the Gramlich Study,” Study," Brookings_Papers Brookings Papers on on Economic Activity, Activity, 167-178. · · (1978), "On “On Activist Fischer, 5. S. (1978), Activist Policy With Rational Expectations,” Expectations," National Bureau of Economic Research Research Working Paper. Fischer, S. and F. Modigliani (1978), “Towards "Towards an an Understanding of the Real Effects and Costs of Inflation,” Inflation," Weltwirtschaftliches Archiv, Archiv, 810-833. Friedman, 8. "Even the St. St. Louis Model Now Now Believes in Fiscal Fiscal B. (1977), “Even Policy,” Policy," Journal of Money, Credit Credit and Banking Banking (May), (May), 365—67. 365-67. Fromm, Fromm, G. and L. Klein (1973), (1973), " AA Comparison of of Eleven Eleven Econometric Models of the United United States,” States," American American Economic Economic Reyj~P~~pg4 Review Paper and Proceedings (May). “ Fromm, G. and L. L. R. Klein (1976), (1976), “The "The NBER/NSF NBER/NSF Model Comparison Comparison Seminar: An An Analysis Analysis of of Results,” Results," Annals of Economic Economic and Social Social Measurement, 1-28. 7-28. “Discretion in of Macroeconomic Garbade, K. (1975), "Discretion in the the Choice of Macroeconomic PoliPolicies,” Annals of Economic cies, Economic and Social Socia 1 Measurement Measurement (Spring), 21 5-238. 215-238. !I Goldfeld, S. and A. Blinder Blinder (1972), "Some Endogenous “Some Implications of Endoqenous Stabilization Stabilization Policy,” Policy," Brookings Papers on Economic Activity, Activity, 585-640. Gordon, R. J. (1976), "Comments Ando" in J. J. Stein, “Comments on Modigliani and Ando” Monetarism, North Holland, Amsterdam, 52—66. 52-66. “New Evidence that Fully Anticipated Monetary Gordon, R. J. (1979), "New Changes Influence Real Output After All," All ,“ National Bureau of Economic Research Working Paper Paper #361. #361. Gordon, R. J. "What Can Stabilization Policy Achieve,” Achieve," American American J. (1978), “What 335-347. Economic Review (May), 335-341. (1979), “Macro "Macro Policy Response to Price Shocks," Brookings Gramlich, E. (1979), Shocks,” pmYijj9~ Activit, 125—166. Papers ~gn on Espnomic Economic Activity, 125-166. Gramlich, E. ((1971), 1971 ) , “The "The Usefulness of Monetary and and Fiscal Fi seal PolicieS Policies as Discretionary Discretionary Stabilization Tools,” Tools," ~ Journal of Money, Credit and Banking (May), (May), 506-532. Hymans, S. H. and H. T. Shapiro (1970), "The The DHL—III DHL-III Quarterly Quarterly Model “ Economy,” Research Seminar of the U.S. Economy, Seminar in Quantitative Quantitative Economics, University of Michigan. 11 “Comentary on on the Current State of the Monetarist Klein, L. L. R. R. (1973), "Commentary Debate," Federal Federal Reserve Bank of St. St. Louis Review (September). Debate,” -101—~0~— R. E. (1975), "Econometric Policy Evaluations: AA Critique” Critique" in in IC, K. Lucas, R, “Econonetric Policy and A. Meltzer, The Phillips Curve and Public Policy, Brunner and North Holland, Amsterdam. Mayer, T. (1975), "The “The Structure of Monetarism I,” I," Kredit_and_Kapital. Kredit and Kapital. McCallum, Policy Mc Ca 11 um, B. (1979), ( 1979), “The "The Current State of the Po 1 icy Ineffectiveness nebate," American American Economic Economic Review Review (May), (May), 240-245. 240-245. Debate,” (1971), “Monetary "Monetary Policy and Consumption” Consumption" in Consumer Modigliani, F. (1971), ~ Spending and Monetary Policy: the Linkages, Federal Reserve Bank of Boston, Boston, Boston. Modigliani, F. (1977), (1977), “The "The Monetarist Controversy or Should We We Forsake Policies,” American Economic Review (March), 1-19. Stabilization Policies," 1-19. (1976), “Impacts ''Impacts of of Fiscal Fiscal Actions on Modigliani, F. and A. Ando (1976), Aggregate Income and the the Monetarist Monetarist Controversy” Controversy" in in J. Stein ed. ed., Monetarism, North Holland, Amsterdam. Amsterdam. Disinflationary Policies,” Policies," American American Economic Okun, A., A.,”!IEfficient Efficient Disinflationary (May), 348-52. Review (May), Okun, A. (1972), (1972), “Fiscal-Monetary "Fiscal-Monetary Activism: Some Analytical Issues,” Issues," Brookings on Economic Activity, 123-163. 123-163. ~rgpj gs Papers Pap op_Economic_Activity, disserPapademos, L. (1977), (1977), Optimal Optimal Aggregate Aggregate Employment Employment Policy, Ph.D. dissertation, Massachusetts Institute Institute of Technology. Perry, 0. G. (1978), “SlOwly "Slowly the Wage Wage Price Spiral: AA Macroeconomic View,” Brookings_Papers View," Brookings Papers on Economic Activity, Activit):', pp. z59—29l. t59-291. A Classical Macroeconomic Model of the United Sargent, T. (1976), (1976), “"A States,” Political Economy States," Journal of of Political Economy_ 207—237. 207-237. “Rational Expectations and Sargent, T. and N. Wallace (1976), "Rational and the Policy,” Journal of_Monetary_Economics. Theory of Economy Policy," of Monetary Economics. “The Almon Schmidt, P. P. and R. Waud ((1973), 1973), "The Al man Lag Lag Technique Technique and the MoneMonetary versus versus Fiscal Fi seal Policy Debate,” Debate," Journal of of the American American Statistical Societ):' Society (March), 77-19. 11—19. Silber, W. (1971), (1977), “The "The St. Louis Equation: 'Democratic' and ‘Democratic’ 1 ‘Republican’ Republican 1 Versions and Other Other Experiments,” Experiments,!! Review_of Review of Economics (November), 362—367, 362-367. ·· and Statistics (November), $mall, 0. Barro’ss Unanticipated Small, D. (forthcoming), “A "A Comment Comment on Robert Barro' Money Growth and Unemployment in the United States,” States," American American Economic Review. Journal_of Stein, J. J. (1978), (7978), “Inflation, "Inflation, Employment and Stagflation,” Stagflation," Journal of Monetary_Economics. Monetary Economics. -102— -102- DISCUSSION OF OF THE MEYER-RASCHE AND TAYLOR TAYLOR PAPERS Neil Wa 11 ace Wallace For us at l970s constitute ten years of at this conference, the the 197Os of additional data data and some theoretical theoretical developments developments that that suggest new new ways ways of interpreting those and earlier data. The two papers papers presented this morning -- in part, because because of of the assignments given the authors --- —— 1970s. contain very very different different views about the lessons of of the 197Os. II will come to still still aa third view and, as it happens, one that does not reprerepresent aa compromise between them. As II understand it, it, Meyer-Rasche accepted the task task of summarizing l97Os, while while Taylor accepted the task of of lessons from the data of the 197Os, surmiarizing l970s. summarizing lessons from the the theoretical developments of the 197Os. That division of labor did not turn out well; it it encouraged Meyer-Rasche to proceed as if one could learn lessons from data data without invoking theory. basis of the preliminary draft of the Meyer—Rasche Meyer-Rasche paper On the basis made available to me and on the basis basis of their their oral oral remarks this this morning, II am left somewhat in the the dark about the point point of view of of the Meyer-Rasche paper. paper. II know what Pm not sure what they did, but I'm sure what their message is. is. Based on what what they did, one might might infer that for Meyer-Rasche, the l970s represent no more the 197Os more than than ten years years of additional data. They Neil is Professor Neil Wallace is Professor of Economics at the University of Minnesota and and Advisor, Advisor, Research Research Department, Federal Reserve Reserve Bank Bank of of Minneapolis. -103-103- use those data and earlier data in the same way that most economists economists ten ten years ago used used the data available to to them. In particular, both their so—called structural reduced—form models their so-called structural models models and and their their reduced-form models consist consist of regression equations that in form are the same as those most econoeconomists used in the 1960s. l960s. Moreover, Meyer—Rasche Meyer-Rasche extrapolate extrapolate from those for the effects of different policies regression equations equations for the effects of different policies in in the the same same way that that many economists in the l960s their estimates. estimates. way many economists in the 1960s extrapolated extrapolated from from their That that for Meyer—Rasche, That is why I say that Meyer-Rasche, the 197Os 1970s seem to to represent represent no no more than ten years of additional data. data. Even at the level of pure pure empiricism, aa different different lesson lesson can can be drawn. Meyer—Rasche extrapolation procedure applied in the late The Meyer-Rasche l970s. 1960s did badly predicting the 1970s. Why, then, believe believe that those same procedures applied now will do well predicting the l980s? 1980s? Happily, though, we do do not have to decide on the the basis of pure empiricism. l97Os -- many of The theoretical theoretical developments of the 1970s of which which -- are described described in Taylor’s paper convincing arguments are in Taylor's paper -- provide provide convincing arguments why why -— we should should not take seriously seriously as as “multipliers” "multipliers" the correlation coefficoeffi- cients or the the functions of them presented in the Meyer-Rasche paper. paper. Meyer—Rasche criticism of the the multiplier multiplier interinterMeyer-Rasche are are aware of the criticism pretation of of their estimates. In effect, they acknowledge the criticriti- cism and say that they are unwilling to to defend such an an interpretation. That, though, is what leaves me confused about their message. Nor does it it help help to to suggest, as Meyer seemed to to in in his his oral remarks, that their estimates of Phillips curve trade—offs trade-offs provide provide upper bounds on on the unfavorableness of this trade-off. quires aa supporting argument. argument. ing, or not interesting. such aa claim claim also reLogically, such re- Moreover, upper upper bounds bounds can be be interestinterest- All of of GNP is an an upper upper bound on on the the output -104- loss loss that accompanies a one percent cut in the inflation rate, but it is not an interesting upper upper bound. Meyer-Rasche Meyer-Rasche must convince us us that their their estimates are interesting interesting upper bounds bounds if, if, in in fact, they are upper upper bounds at all. Such Such convincing convincing must take the form of aa theoretitheoreti- cal cal argument that says why why it is legitimate to to extrapolate in in particpartic- ular ways from particular particular correlations. In the the 1960s, 1960s, many economists thought that their their policy extrapoextrapolations from the kinds of models used used by Meyer-Rasche Meyer-Rasche were were legitimized leqitimized by existing theory. The The theoretical theoretical developments of of the 1970s 1970s have have convinced many of of us that that is not not so. Although Taylor's Taylor’s paper paper some of those those developments, his paper stops short of dedescribes some describing in full generality why we were were led astray badly by the kind of of theorizing that was used. used. Since that kind of theorizing still still perper- sists, it it is worthwhile summarizing in in a general way way what is wrong with it. Whether we are talking talking about most textbooks in macroeconomics macroeconomics or most macroeconometric macroeconometric models, models, the the models from from which which policy oolicy implications implications are drawn drawn consist consist of aa set of relationships relationships -- a consumption function, —- money demand function, and so so on. an investment function, aa money Let us label these these M M1,, M M ,, MM , ... ,MN MN (M (M for macroecolabel for model). model). The The style style of of macroeco1 22 33 nomics implinomics textbooks is to to present the complete model and its policy implications and also also to separate chapters consumption, cations and to present present separate chapters -- one one on on consumption, —- one on on investment, one on on money demand, and so on -- that are meant to —- justify one by one the relationships of the complete model, the M~. Mi. When builders builders of macroeconometric models try to justify their models, they also proceed in in this this way. In order to get at at what is wrong with this this kind of theorizing, theorizing, we must describe the logical relationship -105—105— between these justifying chapters and the macroeconomic or or macroecomacroecoMN. nometric model consisting of MM 11,, MM 22 , ... ,MN. Each justifying chapter consists of aa set of assumptions. Let Let us label label these sets of assumptions ~l’~2 s ,s , ... , SN (S for story), where where for i Si is said to justify M~ each i, justify Mi. . 1 2 The most extravagant extravagant claim made Si and and f\ M~is about the the relationship between between Si following: For each each i, is the follnwing: i' s. implies Mi. In particular, In particular, -1 impiiesM~. ~i is also true. true. In other other words, it it in is never never claimed that the the converse converse is claimed that general, Si and and Mi M~ are are not equivalent not equivalent M~. This nonequivalence and more is implied by by Si than than just just Mi. nonequivalence has two consequences. First, it it implies implies that consistency among the M~ Mi does does not imply consistency among the the Si. Si. consistency among If inconsistent, then If the the Si Si are are mutually mutually inconsistent, then it it cannot be claimed M~. Note, claimed that there is an underlying theory of the I\. Mote, in this regard, that consistency consistency among the Si is never checked checked and, as as II illustrate below, that that inconsistency is is easy easy to to demonstrate demonstrate for most macroeconomic macroeconomic models. Second, if if the the Si are mutually mutually consistent, nonequivalence nonequivalence between s7 and Mi implies we are are missing many of the implications of the the and M~ implies that that we missing many of the implications of underlying theory by limiting attention to the M~. Mi. Thus, for example, the Si often often contain at least hints of aa welfare analysis of inflation. As is is well known, the typical M~provide Mi provide no such such analysis. II will now briefly briefly defend the the nonequivalence nonequivalence claim claim and, at the same same time, argue that that inconsistencies are are present present in standard standard macro macro models. And, since this is St. St. Louis, II will begin by focusing on money demand. The usual usual way to defend the money money demand functions of most most macromacroeconomic models is is to to appeal appeal to the Baumol economic models to aa transaction transaction cost cost model model of of the Baumol —106-106- (1952), Tobin Tobin (1956), or or Miller-Orr (1966) variety, variety. Those models exex- plain money demand in the presence of default-free, default-free, higher-yielding securities -- Treasury Treasury bills, say -- by transactions transactions costs, for example, —— trips to to the bank. function. —— But But the the models imply imply more more than aa money demand They public’s means of They imply that if the ratio of the public's of paypay- its holdings of interest—bearing interest-bearing assets changes as aa result, ments to to its of open-market operations, operations, then then there there is aa change in the amount of say, of resources resources used up up in transactions. transactions. But But such such aa change contradicts contradicts the the usual resource-supply resource—supply assumptions of most macro models. models. Those make no of resources being used transallowance for an altered amount of used up up in transactions. For this and other reasons, the implications for open-market open-market operations of the the theory of interest in the inventory inventory models are very different from those of of most macro models, models, particularly monetarist monetarist models (see Bryant and Wallace 1979). 1979). It is also standard to assume that the money money demand function that one one derives for aa closed economy economy holds holds with only only minor modifications for an open economy in aa world in which each of several countries issues its its own money. It is this this view view that lies lies behind the the attachment of) laissez—faire laissez-faire floating exchange exchange rates. rates. to (the viability of) But such aa claim claim is supported neither by by an an acceptable theory (see Wallace 1979), 1979), nor by recent recent experience. experience. nor by That that the That experience experience suggests suggests that the demand demand for for aa particular money in a world of many many monies may be very different different from the demand for aa single money in aa closed closed economy. In the 1970s, inconsistencies regarding expectation 1970s, of of course, inconsistencies expectation formation have received the most attention. Expectation Expectation formation formation is is important because macroeconomics is concerned primarily primarily with with aspects of behavior that depend upon upon views about the the future -- asset acquisition acquisition -- -107- versus current consumption, the composition of assets, or nominal wage wage determination determination in in those contracts that Taylor discusses at length length in his paper. paper. macroecoIt has been been argued convincingly that the M~of Mi of most most macroeco- nomic models contain, either either implicitly or or explicitly, forecasting schemes that are good schemes in some environments environments and not in others. (See, for example, Lucas 1976J 1976:) of the Moreover, careful examination of Si reveals reveals that that the the particular forecasting schemes imbedded in in the particular forecasting schemes imbedded the Mi were chosen because they were good schemes schemes in particular environments. environments. The inconsistency arises because the environment implied implied by all the Mi -- including various various specifications for policy policy -- may not correspond -- —— at all to that assumed in the various Si. This kind kind of inconsistency is avoided by using aa perfect perfect foresight (rational expectations) equiequilibrium concept. concept. By By using that concept, the economist avoids imposing on the individuals whose behavior is being modeled any fixed way of of on the extrapolating attrib— extrapolating from the past, and and ensures that that he he or she is not attributin~to uting to them views views about about the future that make no sense for the envienvironment they are in. Now having said that perfect foresight is an equilibrium concept, its merits or its it should be evident that it is misleading to discuss its implications policy implications in terms of a particular particular policy conclusion conclusion like like "policy (whatever matter.” (whatever that means) does not not matter." The perfect foresight equilibforesight equilib- rium concept has been been around around for aa long long time. It It would be surprising, indeed, if if that that concept alone implied aa result like “policy ''policy doesn’t doesn't matter.’ matter. 11 In genera general, 1, of course, by themselves equilibrium concepts imply very little. The importance importance of the the perfect foresight equilibrium equilibrium has nothing to to do do with the validity validity of some vague conclusion concept has -108- like “policy "policy does not matter.” matter." Why, Why, then, then, all the attention attention to “policy "policy 1 doesn’t doesn t matter” matter 11 in this this morning’s morning's papers? In 1975, there appeared a paper by by Tom Sargent Sargent and me in which which aa result of result of that that sort sort was was obtained. obtained. We M1, MM We took took aa particular particular Ml' MN' 22 , ... , MN. one that we argued resembled in many many respects standard macro models, and and replaced aa fixed forecasting scheme, one of of the M~ Mi, by perfect , foresight. We argued that the replacement made aa great difference difference for the implications of the the model. In particular, under under perfect perfect foresight and certain other assumptions, all policies in in a certain certain class gave rise to the same equilibrium equilibrium values for real variables. did not follow under under the fixed forecasting scheme. This This result Our message was, was, of forecasting scheme imposed matters therefore, that the kind of matters greatly. greatly. Such aa message, though, is very different from one one that says says that that the perfect seriously as perfect foresight version should be taken seriously as aa model model of of this or any other other economy. From From the discussion above -- and from remarks in in -— our 1975 paper -- it should be be evident that the imposition imoosition of aa perfect perfect —— foresight equilibrium concept concept does not by itself turn aa hodgepodge of of indefensible relationships relationships into into aa coherent model. The Sargent-Wall Sargent—Wallace conace “policy—doesn’t-matter” "policy-doesn't-matter" result is is to to be con- trasted with aa neutrality neutrality result obtained obtained by by Lucas (1972). (1972). The Lucas result was obtained from aa model that is coherent in the sense that its conclusions are derived from aa mutually consistent (and defensible) set of of assumptions, aa single S. S. The Lucas neutrality result, however, applies only to to alternative deficits_consi~~j~~~of deficits consisting of money transfers that individuals know they will receive in proportion to hoj4jjiyipf to their holdings of money. This is neither monetary policy in the sense of open market operations -- there is, is, in fact, only one asset in the Lucas —— -109- model -- nor is it the kind of fiscal policy that any country ever mddel —- The Lucas model is important because it is the first coherent follows. model that implies anything like Phillips curve correlations. The The model implies implies that it is not legitimate to to extrapolate from these corcorrelations for the effects of of different different policies. What is new new about the 1970s and what what offers bright prospects for the 1980s 1980s is not so much much the view II have have set out about about the illogical illogical structure of standard macroeconomics. That view can, II think, be be found in in Leontief Leontief (1947) and and Koopmans (1947) and, II might might add, add, in in the the attiattitude of many nonmacroeconomists toward tude of many nonmacroeconomists toward macroeconomics. macroeconomics. What is is new and What new and exciting about the 1970s 1970s is is the progress we have made in devising dedefensible can explain explain aa wide fensible assumptions assumptions that that can wide range range of of macroeconomic macroeconomic phenomena. phenomena. Lucas (1972) ( 1972) is an outstanding example. In the the work on search search and matching matching models (see, in particular, Mortensen Mortensen 1979), we we see see the the beginnings of aa theory of unemployed resources. And, perhaps, perhaps, in in new work on money money (see, (see, for example, Kareken and Wallace 1979), there are are ideas about how to confront long-standing long-standing problems in monetary theory. theory. Although II think we are making rapid progress, the profession is very far from from having reached reached aa consensus. First, First, not not everyone, by any means, agrees that that we we must must completely abandon abandon the style of of macroeconomic theorizing and modeling modeling that II have described described above. above. For to do do that that is abandon macroeconomics. For many, many, to is to to abandon macroeconomics. This This is is right if macroeconomics is defined by aa style of modeling. But But macroeconomics is defined by if, instead, macroeconomics by the phenomena it it seeks seeks to exexplain and by the policies it seeks to analyze, then this is is not aa call for abandoning macroeconomics. It is a call call for abandoning aa fallacious fallacious style of reasoning that has evidently evidently gotten us nowhere. -110- Second, even imong those who agree that we must, as it were, start over in macroeco1mong macroecojomics and 1omics and monetary monetary theory, there is little agreement agreement about how how to pro— pro- :eed. For example, example, in my very brief listing of promising developments, For [ did not include disequilibrium theory. In my view, diso~uilibrium disequilibrium :heory is not not very promising, promising, but many many economists disagree. Given the lack lack of consensus on theory, it would be be surprising if :here were consensus on policy. And there is not. Academics, of ;ourse, thrive on controversy, which very naturally naturally accompanies the levelopment of of substantially new theories in aa field. :ontrast, seek consensus. Policymakers, in Since the economics profession is far from iaving reached consensus on envy the 1aving reached consensus on macroeconomic macroeconomic policy, policy, II do do not not envy the ;ask of policymakers in the 1980s. l980s. :ask The absence of professional concon- ensus leaves policymakers in the position of having to make up their ;ensus 1wn minds. ~n -111—111— REFERENCES REFERENCES “The Transactions Demand for Baumol, W. J. (1952) "The for Cash: An An Inventory Inventory Theoretic Approach,” Approach," Quarterly Journal of Economics 66, 66, 545-556. Bryant, N. Wallace (1979) "The of Interest-Bearing Interest-Bearing Bryant. J. and N. “The Inefficiency of National Debt,” Debt," Journal of ~ Political Economy 87, 87, 365-381. Kareken, 3. J. and and N. N. Wallace (1979) editors, editors, Models of Monetary Economies, Federal Federal Reserve Bank of Minneapolis. Koopnans, Theory,” Review of Economic Koopmans, T. (1947) ( 1947) “Measurement "Measurement Without Theory," Statistics 29, reprinted in R. A. Gordon and L. R. Klein, eds., ~ Reac!Tiiijs,n Business Cycles (Homewood: (Homewood: Irwin 1965). 1965). Leontief, W. (1947) "Postulates: “Postulates: Keynes’ Keynes' General Theory and the Classicists," in The New New Economics, ed. by bys. Classicists,” S. Harris (New York: Knopf). R. E. E. Jr. (1972) "Expectations Neutrality of Money," Lucas, R. “Expectations and the Neutrality Money,” 103—124. Journal of of Economic Theory 4, 103-124. (1976) “Econometric "Econometric Policy Evaluation: Evaluation: AA Critique,” Critique," in in K. ---,Brunner Brunner and A. Meltzer, eds., eds. , The Phillips Curve and Labor Markets, Volume 1l of the Carnegie—Rochester Conference the Carnegie-Rochester Conference on Public Public Policy, aa supplementary series to the Journal of of Monetary Economics (Amsterdam: 19—46. (Amsterdam: North Holland), Ho 11 and), 19-46. Miller, H. and Miller, M. H. and D. Orr (1966) “A "A Model of the Demand for Money Money by Firms,” Firms," The Quarterly Quarterly Journal Journal of of Economics 80, 413-435. Mortensen, Mortensen, D. (1979) “The "The Matching Process as a Non—Cooperative/BarNon-Cooperative/BarGame.” The Center for Mathematical Studies in Economics gaining Game." and Management Science, Northwestern University Discussion Paper No. 384. 384. Sargent, T. and N. Wallace (1975) “‘Rational '"Rational Expectations’ Expectations' the Optimal Monetary Instrument and the Optimal Money Money Supply Rule,” Rule," Journal of Political 83. 241—254. Political Economy 83, 241-254. Tobin, 3. J. (1956) (1956) “The "The Interest Elasticity of of Transactions Transactions Demand for Cash," Review of Economics and Statistics 38, 38, 241—247. 241-247. Cash,” Wallace, (1979) "Why in Foreign Exchange are Different Different From Wallace. N. (1979) “Why Markets in Other Markets," Reserve Bank Bank of Minneapolis Quarterly Other Markets,” Federal Reserve Fall. Review, Fall. -112—112— DISCUSSION OF OF THE TAYLOR TAYLOR PAPER DISCUSSION Hyman Minsky For For this conference John John B. B. Taylor has prepared prepared aa survey paper paper titled "Recent “Recent Developments in the Theory Theory of of Stabilization Policy.” Policy." Such aa survey is useful useful as it it develops the critical issues in in the field, indicates indicates what progress has has been made, defines the questions on on the research frontier and serves as guide through an important literaliterature. Its usefulness depends upon the competence, competence, taste and vision vision of of the author. John B. B. Taylor Taylor holds aa position and has the credentials that that bebespeak of speak of competence. competence. The paper us is academic exercise The paper before before us is an an academic exercise that that author’s command over a literature which is sometimes illustrates the author's sometimes technically demanding. The paper also shows that he is able to ignore the developments in economic theory and the economy which especially which are especially relevant for stabilization policy and the theory thereof. Hence in reading Taylor’s Taylor's paper II was led to question the taste and vision that guides him and the literature literature he surveys. The theory of stabilization policy is important only as as it it serves as aa guide to action action in in an an unstable unstable world. world. as guide to The The topics topics and and the the literalitera- ture ture that Taylor has has chosen chosen to to cover are are not not useful to to anyone anyone seriously seriously involved in stabilization policy; one cannot derive any guide for action with respect to to the serious issues of of stabilization policy from Dr. Or. Minsky is Professor of Economics at Washington University in St. Louis. -113- this survey or from the underlying papers. papers. Therefore the paper serves 1 showcase for Taylor’s Taylor s talents. talents. no useful purpose aside from being aa showcase In aa similar vein, the underlying literature may be be best interpreted as the products products of aa game game played for academic advancement. In selecting what to discuss Taylor ignores the literature which quite clearly demonstrates that nee-classical neo-classical aggregate aggregate economics, which focuses on price or wage rigidities and which introduces introduces money as an exogneous variable, will not not do. The The literature he he focuses on on looks to refining and making more precise the very neo-classical formulations whose whose logical consistency and empirical relevance relevance has been been demolished by developments in in theory theory in recent years. However one rule of the the game and the authors of the reviewed literature play is game Taylor Taylor and is that rere11 search is to be be carried on on “within” within 11 the neo-classical nee-classical model; thus taste taste search and vision conspire to rule out the relevant and the serious serious because because it it is unorthodox. The most important developments deve 1opments for the theory of sstabilization tabi 1i zation l970s were not policy during the late late 1960s 1960s and 1970s not in the literature literature but in the "world." “world.” The The observations that theory has to explain explain and the stabilization policy has to to contend developments in the economy that stabilization with with changed changed radically in the mid 196Os. 1960s. In particular, stabilization stabilization policy now has to finanto deal with threats and and partial realizations realizations of of financial instability as well as with stepwise stepwise increasing unemployment and aa stepwise acceleration of inflation. For For aall 11 who take “our "our economy” economy" rather than the literature 1i terature about economics derived from the neo-classical (monetarist (monetarist and pseudo— pseudo- Keynesian) research program as as the subject matter of their their research, the world underwent aa marked change of of state around 1965. -114- The The Instability instability that policymakers have to contend with after 1965 is of aa different order of magnitude magnitude and the the potential potential consequences of mismanmisman- agement of of stabilization policy are much more more serious serious than than earlier in the post-war period. The The financial system and practices evolved from 1945 to 1965 1945 1965 so so that the system, which had been been virtually impervious to financial instability, became highly susceptible. Between 1945 1945 and 1965 there were no threats of aa financial crisis crisis of of the scale which which could usher in aa deep depression; depression; in the years years that have followed there have been been at least three such such threats within the United United States, States, as well as aa number of threats to the stability of of the international financial system. Whenever Whenever financial instability threatens to trigger trigger aa debt- deflation process, policy interventions by both the government government and the make aa difference in the path of the economy economy central bank can really make through calendar time. time. Nothing in an apin the paper before us exhibits an ap- preciation of the change in the the character character of the the “stabilization ''stabilization probproblem” lem11 over the years years surveyed. Once the potential consequences of the the mismanagement mismanagement of policy becomes so so much more serious, serious, the importance of of economic economic theorizing about stabilization policy increases. In In particular, economic theory needs to be relevant in the sense that the critical critical situations -- - in this case financial instability and the way way in which which financial varivariables affect aggregate aggregate demand -- are well defined defined within the theory. -- If theory is based upon upon misspecifications misspecifications of of the the economic economic process and the problems faced by policymakers, then theory cannot be be relevant: garbage in -- garbage out applies to to theory construction construction as well as as to garbage —— computer modeling. —115— -115- The problems of the economy have been exacerbated because policymakers have been guided by insights and conclusions drawn from neoneo- classical theory. Neo—classical for Neo-classical theory is an an inappropriate tool for dealing with instability, instability, for financial or any other instability is foreign to this theory. In neo-classical theory any deviation from equilibrium must be due to exogenous developments and any sustaining of a disequilibrium must be be due to “barriers.” barriers. 11 11 Neo-classical theory is able to explain instability only by postulating the existence of of one one or more devils, be they trade trade unions, unions, OPEC, monopoly, the the central bank, government or democracy. Because economic policy policy advising over the past decades has been largely monopolized by practitioners of neoneoclassical theorizing our current economic malaise is in good measure iatrogenic. The physicians, including our hosts, have served to to make the disease worse. worse. AA theory of stabilization policy is needed if and only if if the economy is unstable. There is no sense whatsoever to to the concept ‘stabilization if the is stable. stable. When "stabilization policy’ policy" if the beast beast is When Wallace, Wallace, Sergent, Sergent, al play their games et al. games by positing aa system system whose behavior is deterdeter. mined by by elements that are independent of the variables that, in in their specification, stabilization policy directly affects, then the propoproposition that policy does does not sition that policy not matter matter is is true true not not by by demonstration demonstration but but by by assumption. As the instability that is so evident in the world cannot occur within their models, the games they play only serve to show that their models and the empirical tests that they perform perform are irrelevant for our economy. In my view the strong proposition that emerges from one literature literature surveyed by Taylor, is that this large body body of work is irrelevant for for the world in which we live. live. -116- If economics is to be anything more than an academic nit—pick, nit-pick, theory and theorizing has to go in other directions than those represented in the literature literature Taylor surveys. If economic theory is to be relevant for stabilization policy in “why and in be addressed are "why in what our economy, the questions that must be way is our economy unstable?” unstable? 11 Note the phrase “our our economy.” economy." 11 subThe sub- ject matter of any theory that aims to to be relevant is not an abstract economy devoid of institutional detail but rather an economy economy that is rich in specific institutional detail and which exists at aa particular time and has aa special history. The problem of of economic theory is to select the essential details of the institutional framework framework to to model: the aim of the theorizing is to show causal connections that lead to the observed instability. instability. The hope is that by showing how instability is generated the theory will indicate policy interventions which can atatif not eliminate tenuate if eliminate instability. Although the lines of argument examined by by Taylor Taylor are largely irrelevant to the topic of this conference, “Stabilization "Stabilization Policy: Policy: Lessons l970s and Implications from the 1980s," 1980s,” there were Lessons from the 1970s developments in theory over the past decade that are relevant to stabistabilization policy: Taylor either is ignorant of these developments or chose to ignore then. them. The developments in economic theory in recent years that are relevant to the theory of stabilization policy policy are: 1) l) Progress in general equilibrium theory 2) The two-Cambridge two—Cambridge debate 3) 3) The The recovery of the “lost” "lost" financial elements elements in in Keynes. Keynes. Because II am writing a comment rather than a survey article II will just just devote one paragraph to each of these developments. -117—117— years progress progress in general equilibrium theory theory made During recent years be satisfied for the key propositions of the conditions that need to be this theory to be be valid precise. One conclusion of these developments equiis that the coherence and coherence—seeking coherence-seeking theorems of general equilibrium theory are not unconditionally valid for aa decentralized set of of institutions markets with capital assets, money, banking and financial institutions such as we have. have. An implication of of this conclusion conclusion is that the introintro- duction of of money as an "exogenously “exogenously determined" determined” instrument designed to facilitate trade into into a general equilibrium model in which relative prices determine consumption and production decisions throws no no light whatsoever on the behavior of aa capitalist economy with with aa “money” "money" that that is created in aa banking process. There is no established nicroeconomics microeconomics that can serve as a basis for aa macroeconomic or monetary theory that is relevant to micro— to stabilization policy as long as as the results results in microeconomics depend upon highly artificial constructions to explain the 1 existence of and changes in money.’ money. 1Of the general equilibrium theorists, perhaps F. H. Hahn has 1of the general equilibrium theorists, perhaps F. H. Hahn has l imi tat i ans of theory. See F. H. Hahn: “On "On been most open about the limitations Some Problems of Proving the Existence of an Equilibrium in in aa Monetary Economy," in R. Clower (ed.), Monetary Theory Theory (Penguin, (Penguin, 1969). 7969). Economy,” Professor Friedman Money," Economica, EconOmica, February 1971, 38 “Professor Friedman’ss Views on Money,” 61—80. (149), pp. 67-80. On the Notions of Equilibrium in Economics (Cambridge: (Cambridge: Cambridge Cambridge University Press7i~7~T~~ University Press, 7973). 11 1 Also see K. Arrow Arrow and F. H. Hahn, General Competitive Analysis, (San Francisco Holder Day, 7977), 1971), especially Chapter 14, 74, The The Keynesian Model, pp. 347note that in their earlier earlier 369. In introducing their discussion they note proof that aa temporary equilibrium always exists they they"“...supposed ... supposed that that at the moment moment an an equilibrium was shown shovm to exist, economic agents agents had no capital assets as we know capital assets. It is interestinq to to note note that Arrow and Hahn head head Chapter Chapter 14 74 with aa quotation quotation from H. W. B. Yeats, Yeats, The Second Coning, hold.” Coming, “Things "Things fall fall apart, the centre does does not hold." -118—118— The two-Cambridge debate, ostensibly about capital theory, was really about the validity of the integration of Keynesian theory with the earlier neo-classical theory. The critical issue that the debate clarified centered around the pricing of of capital assets. economy is characterized characterized by two tv,o price systems. systems. capitalist AA capitalist One is is the price system of current output, output, the second is the price system system of capital assets. The price system of current output largely depends upon wages and markmarkupon current ups, whereas the price system of capital assets depends upon estimates of future expected profits, current estimates of of the the unceruncer- tainties involved over various horizons, and current capitalization rates of of profit profit streams. streams. In an economy monetary, banking and In an economy with with the the monetary, banking and financial systems that characterizes capitalist capitalist economies the capitalcapitalnmonetary" phenomena and the two price systems can ization rate is a “monetary” and do do vary relative to each other. Inasmuch as the ratio of the capitalized values of expected future profits profits to the supply price of investment investment output is aa determinant of of investment demand, demand, aggregate aggregate dedemand is sensitive to to the ratio ratio of of these two two sets sets of prices. prices. The The two— two- Cambridge debate is of vital importance for the the theory of stabilization stabilization conclusion that if the ongoing processes policy because it leads to the conclusion of an economy affect this ratio it it will lead lead to to endogenous endogenous change in the performance performance of of the economy: i.e., variations in in the ratio of employed to available resources will result. The two—Canbridoe two-Cambridae debate made it it clear that the “proofs’ "proofs" in the literature that aa growth grm;th equilibriequilibrium of an investing capitalist economy exists depend upon the the assumpassumption that the present present value of future profits profits always equals the perperpetual inventory valuation of of capital assets. assets. But the the equality equality of of the two valuations valuations of capital capital assets in in an an attribute of of equilibrium. -119—119— The "proofs" of the coherence of an investing capitalist economy does not “proofs” coherhold; the proofs depend upon first assuming that aa condition of coher. t s. 2 ence ex1s exists. The third theme in economic theory in the 1970s l970s that is relevant to to stabilization policy is the recovery of the financial and monetary monetary aspects of Keynes' aspects Keynes’ revolution in economic theory. There is is something There very queer about the standard interpretation of Keynes as embodied in in the various IS-LM models. This essentially non-monetary non-monetary view of of the economy is paraded as aa representation of the theory of the major ecoecolife’s work was almost entirely on nomic theorist whose whose life's on money and finance. In the recovery of what lost in Hicks-Hansen—Kleinin the Hicks-Hansen-Klein- Modi gl i ani-Pati nkin tradition it became clear cl ear that underlying Keynes’ Keynes' Modigliani-Patimkin theory was the premise that to understand capitalism it is necessary to model capitalism. This means that that it is necessary to model the way positions in capital assets and investment are financed, the dependence banking and financial system, and the effects of this financing upon the banking of financing relations first upon upon investment investment and then on on income, employemploy- ment and prices. In this analysis, in aa capitalist capitalist economy unemployIn unemploy- ment exists when the long run expectation of profits profits by by business men together with capitalization capitalization rates that reflect portfolio preferences in an uncertain world lead to to demand prices for capital assets that that are "too low" relative to the supply prices of investment investment output. output. “too low” The dede- of investment mand price for capital assets as well as the supply price of 2This is the outcome of the two-Cambridge debate on Capital 2 This is the outcome of the two-Cambridge debate on Capital Theory, although the standard discussion and summary of the debate, 11 6. C. Harcourt, “Some Cambridge Controversies in the Theory of Capital:’ G. Some Cambridqe Capital." (Cambridge, (Cambridge, England: this clear. The Cambridge University Press) does not make -120- output depend upon financing terms. te;y;s. terms. which cannot Financing terms, fu1ly be captured by by aa single interest or not fully be interest rate, reflect whether or term expected behavior of the recent and near terrn the economy lead lead to to suffisufficient realized and expected profits profits that almost all a 11 of of the payment commitments on outstanding obligations are expected to be be fulfilled. By integrating money, finance, expected profitability and the supply price of of current output into into aa theory of effective demand, Keynes dedeveloped the basis for aa theory of the economic processes of aa capitalcapital- ist economy that explained why such “so given to such an economy is is "so to fluctufluctuations.” ations." Instability is an inherent characteristic of of aa capitalist Keynes’ theory. economy in Keynes' Furthermore, Keynes’ Keynes 1 theory is rich, for even though it does not lead lead to a set of policies which eliminate instability, it does lead lead to policy moves ((fiscal fi seal policy) which offset 3 the effects of instability upon employment employment and aggregate income. 3 l970s natured, As the 1970s matured, history advanced the argument from the simple question of “why "why is our economy unstable?” unstable?" The question that economic theory had to address if if it was to be be relevant relevant to to stabilistabili11 zation policy po 1 icy became “why why is it that our economy is so much much more unun- l9SOs?” stable in the ll970s 970s than in in the 1950s ?" The issues that theory had to 3Miong works” in the reintegration of money are: Among the “key key works Joan Robinson, Economic Heresies, (London: MacMillan, MacMillan, 1971). 1971). P. Davidson, Money and the Real Real World, (New York: John Wiley && Sons, 1972). J. A. Kregal Kregal, The Reconstruction of Political Economy, (London: l97~)7 MacMillan, 1973 . S. Weintraub, ~j~y~nesian ~nesian Theory of Employment, Growth and Income Distribution, (Phil adelphia, Chilt~5~FW66JT (Philadelphia, Chilton 1966). Victoria Chick, The Theory Theory of Monetary Policy, (London: Gray-Mills Publishing Ltd., 1973). i973). H. P. Minsky, John~y~nard~ynes John _Maynard Keynes (New (New York: Columbia Columbia University Press, 1975). 11 11 , —121-121- address can be made even more precise by by dividing the the question question Into into two parts: “Why is it "Why it that that our financial system seemingly is more more unstable, more vulnerable to to threats and partial partial realizations of of financial crises (both “Why is (both domestic and international) international) since the middle middle l960s?” 1960s?" and "Why it 1g705 progressed?” it that that inflation became more serious as the 1970s proqressed?' 1 Once economic economic theory moves from the study study of an an economy to to the the of the instability of economy and once the the various faces of of study of of our economy our our economy are taken as the problems theory must address address then the the need to to model money, banking and capital-asset pricing moves to to the the foreforeground. ground. In Taylor’s Taylor's survey, which presumably presumably deals deals with with stabilization policy, banks and banking are nowhere discussed. discussed. We l✓ e all know know that in our economy money money is created created by by the actions actions of profit seeking banks and other financial institutions, institutions, that the assets acquired acquired and and liabilities liabilities issued by banks evolve evolve in to profit and that issued by banks in response response to profit opportunities, opportunities, and that the ~ix nix and activities of of financial institutions also also evolve. This This implies that am economic theory applicable app 1i cable to our our economy will wi 11 intei ntet:1at an grate banking and financing markets into the determination capital determination of of caoital and the determination of asset prices, investment decisions und of the domain of of stability stability of the the economy. economy. ignoring it. You You cannot cannot understand understand something something by by The literature discussed by Taylor’s Taylor's paper paper iqnores ignores his selection selection of of the the literature to discuss, apapbanking and Taylor, by his parently believes you can understand and and give guidance for stabilization policy for our economy by ignoring banks and banking.44 policy for our economy by ignoring banks and banking. 4 It would today’s economists were acquainted would be useful if today's acquainted with H. H. Simon’s Simon's “Rules nRules vs. Authorities in Monetary Policy,” Policy,u Journal of Political l~37. Political Economy. Economy, 1937. —122— -122- It has has long been argued that the instability instability of the economy economy is related related to the the structure structure of liabilities by which which positions positions in in capital assets are financed.55 Experience during the l97Os lends substance to assets are financed. this argument. Experience during the 1970s lends substance to The The relation between between the debt financing of of capital assets positions and the need to cormiitments on maturing debt to fulfill commitments by rolling over debts - by by issuing new debts - is aa critical deterdeter— - minant of the stability stability of of an an economy with sophisticated finance. As As aa the maturing of the the flow of of funds data (poorly designed as result of the the the set of of accounts may be) it is possible to relate the evident evident instainsta- bility of our economy to the growth of the debt structure structure relative to to income income and the increased complexity of financial relations. In In order order to answer questions about why our economy is unstable it necessary it is necessary to fully integrate the monetary monetary mechanism mechanism with system behavior. The literature Taylor “silent” on Taylor surveys is “vague” "vague 11 or "silent" on the processes by which capita] assets are financed. which positions in capital One striking characteristic of our economy that became evident evident in in the 1970s l97Os is the the link between financial instability and accelerating accelerating iinflation. nfl a ti on. Since l96Os whenever Si nee the the mid 1960s v,henever the Federal Federa 1 Reserve follows fo 11 ows the rules for monetary monetary policy to constrain inflation that were were develdeveloped on on the basis of of the experience of the l94Os 1940s and ‘SOs, '50s, aa financial crisis develops; when the Federal Reserve and the government succeed in containing the crisis so so no deep and long recession follows, the the finanfinancial base base is laid down for inflation at at aa higher rate. Since the Since middle 1960s 1t.'e have had three “cycles” llcyclesu of inflation, inflation, constraint, middle we have 5 1. Fisher, ~I. “The Debt-Oeflectiun Great Depressions,” Fisher, "The Debt-Deflection Theory Theory of of Great Depressions," Econometri cas 1933. Econometrica;;_, , —123— -123- Incipient incipient financial crises, lender of last resort resort intervention, federal renewed expansion, financial innovation and accelerated infladeficits, renewed infla- tion. "sequence" took took four to to five years years to work In each cycle this “sequence” its way way through the system. Any theory that is useful useful for stabilization policy policy will need to explain explain why the economy reacted to variations in in the rate of of growth of the reserve base, or or in one manner in the years years prior to 1965 in 1965 and in another manner in the years since 7965. 1965. For an an economic economic theory theory to do this it need need contain aa sub—theory sub-theory of of “financial "financial stability and instai nstability.” bi l ity." Nowhere in Taylor’s survey or in Taylor's or in in the literature he he surveys is this aspect of the stabilization problem addressed. Any theory of the capitalist process needs to focus in in the decideci- sions to own own capital assets, the techniques techniques used used to to finance control over capital-assets and and the investment and investment financing propro- cesses. Obviously Obviously aa theory, theory, if if it it is is not merely mechanistic, which decisions today that are based upon future revenues and costs explains decisions will include aa theory of of expectation formation. The fundamental problem in the making of of decisions today that involve revenues and time horizon is that costs over over a significant time that the future is uncertain; the future cannot be represented by by aa set of nice stable probability functions over well-defined outcomes. The need to make decisions in an uncertain world world leads to to one how does one behave rationally in an irrational irrational world?” world? 11 question, 11“how An "irrational in which what happens is not explained explained with “irrational world" world” is one in precision by by the accepted theory. theory. the requisite precision As As long as theory does not explain a phenomena phenomena with with the exactness required required for decision, then irrational. the world of that phenomena is irrational. -124-124— If, for aa capitalist economy, the world conforms to expectations derived derived from standard theory aa large part of the (instathe time, even as it behaves in aa manner (instability) inconsistent with this theory aa part part of of the the time, then decision making formulas that use the accepted theory will not determine determine the rational man. behavior of aa rational behavior In aa world where diverse types of behavior behavior excan occur, occur, theory is effective effective as as aa guide to to decision and policy policy ex- actly as it yields information as to to which of of the diverse diverse types of behavior of the economy is is likely to rule. If If economic theory is to be an an ingredient in the formation of expectations expectations by by aa rational man, it it needs to relate the expected expected behavior of the economy to to history history and the evolving evolving institutional arrangements. The Franklin National bankruptcy of 1974 1974 and what what followed is aa concrete situation in which which policy actions truly affected concrete example of a situation the the behavior of the economy. In May of 1974 the Federal Federal Reserve, under under Arthur National so so Arthur Burns, opened the discount window wide to Franklin National that all of Franklin National’s National 's overseas and money money market liabilities were validated. The Federal Reserve by by this this action action aborted aa wave of of withdrawals from the international banks “world of banks and assured the "world international finance'' finance” that the offshore liabilities of of large, large, if not respectable, American American banking banking institutions institutions were implicit implicit contingent liabilities of the Federal Reserve. Reserve. This and related related interventions by the cooperating institutions institutions in in 1974-75 1974-75 together the Federal Federal Reserve Reserve and and cooperating together with massive government deficits made it virtually virtually certain that that the recession of 1974-75 would be contained contained and that the subsequent rerecovery would lead lead to serious balance of payments difficulties and ininflation at an accelerated accelerated rate. rate. Policy Policy may not always matter, but but -125—125— there are junctures junctures in in the history of of an economy when policy really matters: 1974-75 was one such such juncture. juncture. It is the duty of economists economists who who parade as as knowing something issues. about stabilization policy to be aware of such issues. the Neither the literature Taylor discusses nor Taylor in in his his paper paper seem to to be aware of problems. these problems. policy needs Theory that is useful for stabilization policy to offer offer guidance to central bankers and other other policynakers policymakers when when they to situation such as ruled in in 1974—75. 1974-75. are faced with the need to act in aa situation Taylor’s paper nor the literature he By this criteria, neither neither Taylor's he chose to report on are useful. -126-126— THE CASE FOR GRADUALISM IN POLICIES TO REDUCE REDUCE INFLATION INFLATION Allan H. Meltzer Inflation is usually defined as aa sustained rate of of increase increase in aa broadly of prices. prices. broadly based index index of Whatever meaning meaning one gives gives to to the the imim- precise term !!sustained,!! sustained, the past fifteen years seem to to meet meet the standard. Both the all—item all-item consumer price index and the implicit GNP deflator have increased in every every quarter since late 1965, and neither seems likely to reach reach aa zero rate rate of change in the near future. Sustained Sustained inflation inflation at the rates of recent years is rare, even even if not unique, in the histories histories of developed economies. economies. It seems seems useful, at aa conference summarizing the lessons of the seventies and and drawing to look back on the path we we have travtravimplications for the eighties to elled elled and to to explore explore the path we might take to restore restore price stability. stability. II shall use the the opportunity opportunity to to discuss discuss some of what what has been been learned learned about about monetary policy. The list is is aa long one, particularly particularly if we inin- clude known but later clude propositions that once were "known" later forgotten or rerejected jected in the years of Keynesian Keynesian orthodoxy, so II shall not attempt attempt to be complete. Any gain from ending inflation depends on Any long-term gain on aa negative negative relation between inflation and real output. The The most common reason for suspecting that aa gain will occur is the the observed association between inflation and and changes in relative prices. See Cukierman (1979). (1979). The Or. Meltzer is Professor of Economics and Social Sciences Dr. Sciences at at CarnegieCarnegieMellon University. University. The author is is grateful grateful to to Alex Alex Cukierman Cukierman and and Jerry L. Jordan for for helpful comments on draft. Jerry L. Jordan helpful comments on an an earlier earlier draft. —127— -127- principal problem for monetary policy at present is to achieve this gain gain by ending inflation at minimum minimum transitional loss of output. Every six months, II join join with my colleagues on the the Shadow Open Market Committee Cammi ttee in recommending recommending aa policy po 1icy of pre-announced, gradual, gradua 1 , sustained sustained reductions in the growth of money as aa means of restoring price stabilstability. ity. A clear A clear statement of of the the reasons reasons for for aa policy policy of this this kind kind often called gradualism gradual ism -- has has not been provided. provided. —- II will will try to parpar- tially fill that gap and to to relate the case for gradualism to to some of the lessons we have learned from recent experience with with sustained sustained ininflation. The history of recent inflation is is surrounded surrounded by myths that that obobits persistence. scure the origins origins of the inflation and the reasons for its persistII begin with an account of the origin and an explanation of persist- ence. Much of of the case for gradualism depends depends on on the way in which which inin- dividuals dividuals form anticipations anticipations of the future. I present one view of rational rational expectations, expectations, in in the the sense of of Muth (1961), and and use use this this model model of of expectations to show how Federal Reserve policy policy procedures can conconvert real shocks into permanent permanent changes in the rate of of price change. change. Then II present the the case for gradualism in aa world world in which persistent and transitory transitory changes in monetary monetary policy cannot cannot be identified quickly. THE THE ORIGIN AND PERSISTENCE OF CURRENT CURRENT INFLATION The most most enduring myth about the origins of the current current inflainflation is that the inflation started during the the Vietnam war. viar. According According to of history, President President Johnson rejected the the recomrecomto aa standard version version of 11 mendations of his advisers by refusing to to choose choose between between “guns guns and butter.” butter." The delayed asking Congress The President President de 1ayed asking Congress for for increased increased taxes taxes -128- (or for smaller expenditures for redistribution) and allowed the budget 1967. deficit to overstimulate the the economy in 1967. has been intractable. has Since 1967, inflation some estimates, or more years years According to some estimates, ten or of recession would be required to eliminate inflation inflation by monetary monetary and fiscal policies.11 The facts do not correspond to to this capsule history. The rate of increase of of consumer consumer prices prices reached the the 33 to to 4% range range at least aa year before the Vietnam deficits. deficits. Spending by the federal government in in doldol- 5% below the 1962 1962 level lars of constant purchasing purchasing power power remained remained 33 to 5% level during most most of 1965. Budget deficits and government spending spending did not start the inflation or or encourage the Federal Reserve to expand in in 1965 1965 1966. or 1966. 1966. The budget had had aa small surplus in in 1965, 1965, and aa small small deficit if in The Federal Federal Reserve slowed slowed the growth rate of the monetary monetary base late in in 1966 1966 in aa sudden burst burst of of concern concern about about rising rising inflation. inflation. The The 1967 deficit of more than $13 billion comes after these first steps steps to slow inflation and much too too late late to explain the start start of the inflation. inflation. A surtax was was added to SO the Vietnam defA to the income tax in in 1968, 1968, so deficit proved proved to to be temporary. By late 1968, the budget again was in 1969. surplus, and and the surplus persisted persisted in 1969. The 1969 1969 surplus surplus of of $8.5 billion is one of the largest of the past past thirty years in in real real as well as in nominal terms. terms. To To sustain sustain the thesis that the Vietnam deficits started the curcurrent inflation, inflation, one must not only ignore ignore the the problem of the timing of of 1See Perry more complete statement statement of this this view and see (1978) for aa more for an extreme form of the the argument that that inflation inflation is intractable. Perry’s Perry's Phillips curve implies implies that it costs $200 $200 billion billion dollars of of real output for each percentage point reduction in in the rate of of inflainflation. tion. —129— -129- corrmented earlier, but must accept the start of inflation, on which I conniented generdeficit generthe improbable improbable proposition that six quarters of wartime deficit irreversible. ated anticipations that were irreversible. Credulity Credulity is is strained strained comfurther when the 1967 1967 deficit is expressed in in constant constant dollars to compare pare with with the deficits in in earlier earlier and later later years. years. The The 1967 deficit is is almost identical to to the 1958 deficit when both both are expressed in dollars of the same purchasing power. power. of sustained inflation. sustained inflation. The 1958 deficit did not initiate years On the contrary, inflation fell from the 33 to to 4% range of 1956-57 range in in 1958—59 to less 1% 4% range of 1956-57 to to the the 11 to to 2% 2% range 1958-59 and and to less than than 1% by 1961. The 1975 1975 nominal budget deficit of $70 billion is four times larger than the deficits of 1958 1958 and and 1967 when when the the three are expressed expressed in dollars of comparable comparable purchasing purchasing power. power. The The 1975 deficit is not not balanced budget budget or aa surplus but but by sustained deficits. followed by a balanced Yet, most broad broad measures of the rate of price change declined in 1976. 1976. deflator rose by less The GNP deflater less than 4.5%, 4.5%, on average, for the first three quarters the year. less three quarters of of the year, and and the the consumer consumer price price index index rose rose by by less than 5% for the year as aa whole.22 The proximate proximate cause of the start of the current current inflation inflation is the monetary of the 196Os. monetary policy policy of the early early 1960s. Inflation persists persists because Inflation because policy policy of future inflation by producing to sustain sustain anticipations of continues to persistent inflation. Bursts of anti-inflation policy, and announceannounce- ments of firm congnitments inflation, are not followed by by corrmitments to reduce inflation, policies that reduce money growth. 2The The decline in in the rate of inflation affected more than just food prices prices as as is sometimes claimed. The wholesale price price indexes of of consumer finished goods rose by less than 2.5% for the year. —130— -130- > 0 CD 0 tO 0 (D~ —‘C <0 r —~ tOED — ~1 CD° o tAG) I 8 .0 CD d, o I Rate of Growth of the Monetary Base (3yr. moving avg.) - CD 0 Rote of Growth of the Monetary Bose ( 3 Year Moving Average) ojo 9.0 w CHART 1 c,, C: > 5 I —4 ~ 7.0 ..: >. r<) o I 6.0 Cl> f/) 0 Q) - 5.0 5 ...>Cl> C: 0 b I 0 4.0 o 3.0 ‘ (iJ o -... - -Is ~ ..c: 3 b I ro 0 e> 2.0 0 Cl> 1.0 b I CD Year 0 -a 1980 (0 CD -.4 c_fl -a o 1975 (0 —4 (0 1970 -a 1965 o — 1960 (0 1955 o L....l---'---'-...,_J........1---'---'-...,_J........1----'---'-...,_L....l__,__.__,_L....l__,__.__,__.J....J (0 0 01 01 0:: 0 0 Chart 11 uses aa twelve quarter moving average of the growth of the adjusted of the effect of adjusted monetary monetary base base as as aa measure measure of the long—term long-term effect of monetary monetary policy. Using this measure as an index of the sustained thrust thrust of policy. we can divide the monetary history monetary policy, history of the past twenty— twentyfive years into into five episodes. The The first, from 1955 to 1960, has has aa low average rate of monetary 1.1%. monetary growth, 1.1%. transition. The second is three—year is aa three-year The The twelve quarter quarter moving average rises steadily steadily toward the 5.5% range. In the third period, 1964-71, the growth of the base remains in the neighborhood of of 5.5%. 5.5%. The fourth period is a one—year one-year transition, transition, 1972, during which the maintained growth of the base moves from about 5.5% 5.5% to from about to 8.5%. 8.5%. Since 1973, of the base Since 1973, the the moving moving average average of the base has grown at aa maintained rate rate of about 8.5%. 8.5%. AA number number of of studies, studies, including including my my own own Meltzer Meltzer (1977), (1977), suggest suggest that inflation follows money growth growth with an average two-year lag. The mean of the three—year three-year moving average ending in in year t, shown in Chart 1, is is an an unweighted average average centered centered in in year year t—l t-1. - If If we we impose aa twotwo- year lag, inflation inflation in in year t+1 t+l is influenced by by the twelve quarter rate of of growth growth of the monetary base ending in in year t. t. To To measure perper- sistence, I have have computed computed the standard deviation of the percentage rates of change of the consumer price index the percentage rates of change of the consumer price index and and the percentage rate rate of of change of 1956-61, 1965-72 1965—72 and 1974-78 1974—78 that of money wages for the years 1956-61, correspond to to the two-year lag lag of of prices behind the maintained maintained growth of the monetary base. base.33 The data are shown in in Table Table 1. 1. 3 The rates of price and wage change are one-year one-year averages averages of of the all—item consumer price index for six-month six—month spans and average all-item average hourly earnings over six-month BCD. Wage data data are not available available six-month spans spans from BCO. before 1965. 1965. —132-132- TABLE 11 TABLE Mean Mean (p) (w) and Standard Deviations (cv) (o) Rate of of Price Change t+l Growth of Adjusted Adjusted Monetary Base in intt Years ((t) t) 1955-60 1964—71 1964-71 1973-78 Omitting 1974 1974 pµ cva pw 1.1 1.1 5.7 5.7 8.4 . 18 .18 .44 ..31 31 1. 9 1.9 4.0 7.5 1.00 1. 26 1.26 2.42 6.4 .85 of ,iage Rate of Wage Change t+l CT cv p ll N.A. 5.9 8.0 7.7 CT cv 1. 13 1.13 .79 .79 .36 The The data show aa tendency for the standard deviation of the the rates rates change of money and wages to to fall fall in recent years. of change Removing the standeffects of the oil shock, by by omitting 1974, further reduces the stand- ard deviations. deviations of of the rates of change of of The standard deviations wages and prices deviaprices are not startlingly different different from the standard deviations of the maintained growth of of the adjusted base. The The persistence persistence of rates of price change from year to to year appears to to be be related to to the persistence of maintained rates of money money growth. To To examine examine further the relation between between the persistence of money growth and the persistence of of inflation, Table 22 compares the two quarter average rates of growth of base base money to the quarterly averages averages of the rates of 1. of change of of prices and wages wages used in in Table Table 1. As As before, before, II imposed imposed aa two-year two-year lag of rates of of price change behind rates of money growth. The data the variability variab1lity of base money money growth data now suggest that the is of approximately the same magnitude magnitude as as the variability variability of the the rate rate is of wage change.44 The standard deviations deviations of of the rate of price change, 4The time periods for the base base differ from those in Table Table 11 bebecause Table 11 has has aa three-year moving average. II have have kept the periods for rates of price and wage change the same as in Table Table I. 1. -133—13 3— however, are not closely related to the standard deviations of rates of base money growth. Short-term variability of the rate of of price change reflects more than the variability of of monetary growth. growth. TABLE 22 Mean (v) (ci) and Standard Deviations (ci) (0) Two quarter moving average average of of growth of monetary base Period t 1954-59 1954—59 1963-70 1.1 1. 1 5,7 5.7 1972—76 1972-76 8.2 p ci a 0.87 1.10 1. l0 0.91 Period Period t+2 1956—6] 1956-61 1965-72 196572 1974—78 1974-78 Standard Deviations Deviations (ci) ( 0) Standard quarterly average rate of change over six—month six-month spans Consumer Money prices wages wages p µ ci a 1.55 1.61 1. 4.00 1. 1.34 4.00 34 8.2 2.61 2.61 p µ ci a 5.9 5.9 1.19 1.19 8.0 8.0 0.90 0.90 1953-70 and 1972-76 1972—76 include several periods The data for 1963-70 oeriods in which inflation was given "highest ‘highest priority priority" as as aa goal of public policy. Careful Careful inspection of the data data shows that periods of slower growth of of the base coincide coincide with these announcements in in 1966, 1966, 1969-70 1969-70 and 1974—75, 1974-75, periods of of slower growth is long long enough to have any but none of these periods marked effect on on the ~tandarddeviation deviation of of the the growth rate of base. marked effect the standard growth rate of the the base. Table 22 shows that the standard deviation deviation of of the two quarter moving moving growth rates is is independent of the rate of growth of the base and not very different different in in the three sample periods. The data suggest two reasons for the persistence persistence of inflation inflation and the slow slow response of inflation to changes changes in the growth rate rate of of money. First, short—term short-term rates rates of price change are relatively variable, so growth from people have difficulty separating the effects of money growth other influences on short-term short-term price changes. —134— -134- This This is is particularly the case for recent recent years, years, when announced changes in oil prices have had on measured rates considerable influence influence on rates of of price change and their variability. last. not last. Second, the commitment commitment to anti-inflation anti-inflation oolicies policies does unwilling to buy buy long-term long-term contracts based on the People are unwilling assumption that the slower rate of money growth will persist persist long enough to reduce the trend trend rate of inflation. inflation. In the next section, I relation offer an an explanation of the rel a ti on between between the variability of money growth and the persistence of inflation. 5 THE BASIC INFERENCE PROBLEM PROBLEMS Each week week the Federal Reserve reports reports the the growth grm,th rates of various monetary aggregates. Market participants try to infer the future course of of money money growth, interest rates, prices and exchange exchange rates from the announcement. Their problem, and ours as economists, is to sepasepaTheir rate transitory changes in money growth (or other variables) from perpersistent changes. II call this problem problem of separating permanent permanent or perper- sistent changes from ephemeral or transitory changes changes the basic inferinference problem because it it arises for most economic variables variables and is aa major major problem for people making decisions. illustrate in aa given week To il 1us tra te the problem, prob 1em, suppose that ·in week the the anannounced change in money money is large relative to past changes. Few obob- servers will use the observation for a single single week to to predict predict the growth path, and fewer still will predict an an equiproportionate equiproportionate change in the rate of inflation. Let the increased rate of money growth perper- sist, for aa month or two, and the balance of of opinion will start to 55This This section section owes debt to owes aa large large debt to Brunner, Brunner, Cukiernan Cukierman and and Meltzer Meltzer (1979). -135—135— change. More observers will infer that there has been aa persistent change in in the growth rate rate of money. The effect of the the first week’s week's observation on market prices, prices, interest rates and exchange rates differs from the effects of of aa change that is is perceived perceived to be be permanent. Although the change in money is rere- ported, and therefore is known, the correct inference to be be drawn from from the information is uncertain because because the content of the the information information is is uncertain. uncertain. AA rational investor investor who \\lho uses all available available information, must must first decide what he knows; that that is to say, say, he must decide how he has has observed can be expected to to persist. much of the changes he This view of the world in which which monetary and other policies operate differs in an important way way from the usual model of rational expectations developed developed by Lucas (1975) and others. There, people are uncertain about whether the changes they observe observe are the result of shocks shocks that change relative prices prices or or shocks that that change the the absolute absolute price level; once information becomes available, there is no no doubt about its meaning. Given the speed with which information becomes available, available, the be the prinprinconfusion between aggregative and relative changes cannot be cipal source of confusion. The The main aggregates in our our models -- money, -— debt and deficits or or GNP, prices and output -- are observed observed within aa -— month month or aa quarter. abOnce they are observed, the confusion between between ab- solute and relative changes disappears. The permanent-transitory confusion does not disappear when data are published. published. The principal principal uncertainty that individuals face arises, in this this model, from an an inability to to properly properly interpret interpret information, not from lack of of information. information. People observing observing the price index must decide decide —136-136- whether a reported increase or decrease in an aggregate is aa one-time change that will soon be be reversed or the the start of of aa higher or or lower of change. maintained rate of change. Expectations remain rational, but the use of all available available information does does not solve the inference problem and and does not eliminate eliminate error. A simple model A model brings out the source of the permanent-transitory permanent-transitory confusion. It is, is, of of course, only one of many ways in which the problem can can be formulated, but it is the way way that has been been used in in an application to the problem of stagflation where it it produces changes in prices and employment that resemble the aftermath of the oil shock.66 prices and employment that resemble the aftermath of the oil shock. An An observable variable variable Xt can be be divided into into two components, aa permanent component. X~, X~, and aa transitory component component X~. Xi. X~and Xi and AX~are "X~ are permanent component, normally distributed random variables with mean zero and known, concon2 andd J~q• 2 . stant ~ People cannot observe X~ or s t an t variances, axp an axq· Xi but must X~. infer the permanent value by observing current current and past values of Xt. xt = + The expectation of X~,conditional Xt, conditional on all information available available in . d t, 1s . X~. xPt· period is per10 The inability to to separate separate permanent permanent and transitory components makes the optimal forecast forecast of XX aa distributed distributed lag lag of of past past observations. observations. Contrary to much of the rational expectations literature,77 we find that 6Brunner, Cukierman and Meltzer (1979). (1979). This application conThis application considers the effects of real shocks. shocks. The role of the permanent-transitory confusion in the transmission of monetary shocks to real variables introduces additional problems. 7Benjamin Friedman (1979) is an an exception. exception. —137— -137- distributed lag 1ag of past observations is an optimal method of using a distributed forecasting. The reason is that repetitive observation observation of an an aggregate aggregate are required to to learn whether whether aa permanent change has occurred. If perIf per- manent changes are frequent, and transitory changes changes are infrequent, a l·ikely to be treated as as permanent soon soon after after it it change in XX is more likely occurs. At At the opposite extreme, transitory changes are frequent and permanent changes changes are rare, so so it is optimal optimal to observe aa relatively long series series of observations before concluding that aa permanent change 2 axp has occurred. In more technical technical terms, terms, the larger the ratio -2- the axq xq faster people correctly infer that aa permanent change change has occurred; the smaller the ratio, the larger is the number of of observations observations required to sustain sustain the inference that aa permanent change has occurred. II 11 We We can put more content into the terms tenns “frequent” frequent 11 or “infrequent” i nfrequent 11 by using the computed standard deviations for the two quarter and three- year moving averages in Tables 1l and 22 to estimate the relative varivariof permanent permanent and transitory components and to find the implied ance of length of the lag in reaching rational rational judgments judgments about permanent permanent shocks. The permanent permanentS variance van ance of the growth rate of the monetary base is set equal to the variance variance of the three-year growth rates. The two quarter moving average growth rates include both both permanent permanent and transitory comcomponents. ponents. We assume that permanent and transitory variances are indeinde- pendent and compute compute the the transitory variance by subtracting the variance of of the twelve quarter average from the variance variance of the two quarter average. Muth (1960, pp. pp. 302-4) shows that the best (minimum variance) Muth linear linear estimator of the the permanent value of aa variable can be computed from past actual values using the the variances of the permanent permanent and -138- transitory components. For the problem at hand, the calculations for the three periods of relatively relatively constant growth of the monetary base show that the relative variances of the growth rates of the base are: are: 1955—60 1955-60 1964-71 71 1964- 1973—78 1973-78 2 °Ji__ .04 2 . 19 .19 . 14 .14 oq These ratios imply very different lags in the adjustment of the expected growth of the base. base. In 1955—60, 1955-60, only only 55% of the adjustment of expectations occurs within three years. The The reason is that the very very low three—year average growth variance around the three-year growth of base money obscures the change in the maintained rate of growth, when it occurs. Rational individuals interpret most of the permanent change change as transitory and fail to adjust fully for several years. In In the two remaining remaining samples, the variance of the the permanent permanent component is higher higher relative to to the variance of the transitory component. variance Expectations adjust more quickly; more than 95% of the full adjustment occurs in the first three years.88 are related to the growth of of money money that Expectations of inflation are individuals expect to to be be maintained. maintained. The expected growth of of base money money can be reduced permanently only only if the actual growth of base base money is reduced. The speed of of adjustment of expected expected to actual actual growth growth can be growth rate rate of of the base is rerereduced, also, if the variability of the growth duced. For example, examp 1e, if if the Federa va ri anee of the For Federal1 Reserve reduces the variance to equal equal the variance variance of the twelve quarter two quarter growth rate to 8BtTransitory1 variances are moving uTransitory variances are computed computed from from two two quarter qua,~ter moving averages, so so two two quarters are used as as one period when computing computing the lags. 11 —139— -139- about the permanent growth rate, 85% of the adjustment of expectations about growth occurs in the first year. growth of inflation inflation respond Expectations of more rapidly to monetary policy; the length of of the lag of inflation inflation bebehind hind money growth declines. It is, no doubt, doubt, aa mistake mistake to use use these numbers as as precise precise estiestimates mates of the expected length of the lag. lag. Fortunately, the principal implications do not depend on the the precision with with which we measure the speed of adjustment of expectations. expectations. short—tern policies are If short-term are less less variable, the speed of of adjustment adjustment increases. Faster adjustment of exof ex- pectations pectations lowers lowers the length of time between changes changes in the growth rate rate of the the monetary base and changes in in the expected growth of the base and, and, therefore, in the expected rate of inflation. The The shorter shorter the lag, the smaller, jbus, is the persistence of inflation. smaller, ceteris pparibus, A related, A related, but distinct, implication implication explains why why short—term short-term changes in in the the growth growth rate rate of of the the base have little effect on on maintained maintained changes base have little effect inflation. inflation. The larger the transitory variance variance of the growth rate rate of of the The base, given the long-term or or permanent variance, the longer longer is the lag. Short-term rate of the base have have little little effect Short—tern reductions in the growth rate on on long-term expectations expectations if the short—tern short-term growth of the base is highly variable. The The real real costs of reducing inflation are higher, higher, under these circumstances. circumstances. rising rising unemployment. unemployment. The costs take the form of of recession and Recession encourages the Recession encourages the Federal Federal Reserve Reserve to to shift shift to aa policy of of monetary expansion expansion thereby reinforcing expectations expectations that that rate of the the base will will not not be reduced. the maintained average growth rate Chart 1, anti—inflation policy have, l, above, shows that past periods of of anti-inflation in fact, had little effect on on the maintained maintained growth rate of the base. base. -140- The calculations in Tables 1l and 22 imply that the lag in in the fifties. is shorter now than in the fifties. formation of of expectations expectations is The data that the the reason for the shorter lag is the increase suggest, however, that in the measured variance variance of the the permanent permanent component, not not aa reduction reduction in the measured variance variance of the transitory component. THE POLICY PROBLEM the short-term variance of of the The Federal Reserve can reduce the growth of the monetary base by adopting targets expressed in in terms of base. the base. currency, the uses uses of of the base, are approxiReserves and currency, approxi- sum of reserve bank bank credit and international reremately equal to the sun serves. serves. With floating (or adjustable) exchange exchange rates, the the Federal Reserve can control the two quarter quarter growth rate of the the base by controlcontrol- ling the stock of Reserve bank credit. credit. ling To control the base the Federal Reserve need not solve an impossible or even a difficult problem. All they must do is control the asset side of of their balance sheet. As is well-known, the Federal Reserve cannot control both both interinterest rates and the growth rate of the the base. By specifyinq specifying short—term short-term targets in terms of values (or ranges) of the Federal funds rate, the Comittee surrenders control of short-term short—tern changes Federal Open Market Committee in the base. base. The problem of separating permanent permanent and transitory changes helps to explain short—term control of the base explain how loss of of short-term contributes to persistent movements movements of the base even if if the the dominant dominant are real, not nominal shocks. shocks in the economy are To illustrate the problem, II use use the the three three equation, equation, equilibrium equilibrium To illustrate the problem. model based on on Brunner, Cukierman and Meltzer (1979). -141- All variables are natural natural logarithms. neoProduction or output, Yt, is given by aa neo~ classical production function (1) ~ = ut + with ~ lt, the number of of man hours of labor labor and ut aa productivity productivity shock; 6o is the elasticity elasticity of output with respect to to labor. Real aggregate aggregate spending is always equal equal to to output, ~ Yt, and depends on expected expected or per— pernanent y~,on agmanent incone, income, y~, on the real real rate of interest and on shocks to agdiffergregate demand, demand, c~. et. The anticipated anticipated rate rate of inflation is the differ- ence between between the iogarithns logarithms of of the price level level anticipated for next period ~t~t+i~ (tPt+ll and and today’s today's prices ~ (pt). The market market rate of interest is is i~. it. (2) ~ = a+by~ + c[it - ~ + b>0;ccO b > 0 ; c < 0 Equation (3) equates + 1t, to the equates the current stock for base base money, BB + denand demand for base money, where where ft is the shock to to the level of of nominal 9 ct. affects money balances. balances. 9 Some Some part part of the the shock to to spending, spending, et, affects the demand for money; the rest affects the demand demand for bonds and the supply supply of labor. Increases Increases in spending are financed by by reducing reducing the demand for money money so ae is is positive positive and and increases in c2 reduce the demand for money. (3) B B ++ y t = + + ~ + y~ + y ~ - Bet 8<0 B < 0 1> > y y,, 8 a> > 0 0 l 99The analysis can terms of The analysis can be be cast cast in in terms of growth growth rates rates of of money money by by naking adjustnents. making minor adjustments. -142-142— IS-LM model. The three equations form an augmented IS—LM The principal novelties are the distinction between permanent and current income and the introduction of of permanent and transitory shocks. soocks, The three shocks, ~ ~ ut, ct and 't' have permanent and transitory components, but people are not able to distinguish the permanent and transitory components when observing the shocks. For example, ut u~== u~ u~ with known variances ci~q~ Eu~and EAu~ o~P and o~ and Eou~ , normal distributions and expected values Eui 9 equal to zero. ci~ Substituting eq. (l) (1) into (2) and (3) and i~reduces and solving solving for it reduces the system to to two equilibrium relations. The The money market equilibrium equilibrium eq. (4) and the IS curve, curve, eq. (5) relate i~to it to the three three or LM, in eq. shocks, to to the price price level level and to other variables. variables. For For the the current 100 treat y~and i~as shocks.~ analysis, I treaty~ and lt as given and and independent independent of the shocks. (4) si~ Bit = 8B ++ 't - Pt - yu~++ Bet SEt - ~ yut y•lt - (l(l-y)y~ 1)y~ - a (5) cit + ut - c(tpt+1_pt) = - - Ct + - - ai~ b y~ a - - of its existence, the Federal Reserve used the market During most of interest rate (or some surrogate like the level of free reserves) as as the operating operating target. terest rate at terest rate at ii 0 Suppose the Federal Reserve sets the target inin- and supplies or to keep and supplies or absorbs absorbs base base money money to keep i~ it== i. i • 0 10A full solution is given in Brunner, Cukierman lOA Cukierman and Meltzer (1979) by specifying the labor market equations. The The additional additional detail detail would not alter the conclusions of of this discussion. The The principal difdifexferences that have been been neglected neglected are the dependence of y~ on the expected values of of the real shocks shocks and the dependence of 1lt on the actual actual values of the real shocks. shocks. The reader who who is disturbed disturbed ~y oy the partial solutions can substitute permanent permanent and actual values of of shocks -- real y~and shocks -- for y~ and ~ lt· For the analysis that follows what matters matters is that the responses responses of IS IS and LM to the shocks cause cause i~to it to differ from ii . 0 4 -— -- 0 -143- The stock of base money B B ++ rt changes only as required to maintain the ~ interest that the the stock interest rate rate at at ii,, which which is is to to say say that stock of of money money now now dedeo pends on the real real shocks. ~t (6) ~(et~ ut) Equations Equations (4) and (5) are shown as solid lines in Figure 1. slope of LM LM from eq. eq. (4) is positive in the i, pp plane. plane. IS is -1. The The price level is p. 0 The The The slope slope of of The policy policy of fixing interest rates, at i, temporarily at i o , makes the interest rate pre—determined pre-determined at i. i0 . MoneMone- tary policy keeps the interest rate constant by changing money. WhenWhen- ever there are real shocks to productivity or to spending and the demand for money, the Federal Reserve changes the stock of money enough to hold interest rates fixed until it decides that the shock is perpermanent. Consider the effect of aa negative negative productivity shock, shock, dut < 0. From (4) and (5) we compute the elasticities 0 and > tILM ~t dut = 1 c IS AA negative shock shifts both the LM LM curve and the IS JS curve to the right right in Figure 1. small, If y~ is sma 11 , the the demand for money changes very 1little, i tt 1e, and interest rates rise. rise. The The Federal Reserve offsets the rise in inin- terest rates by by increasing the money stock. dit 1 1 = - < B < 0 0 -144- ii ' \' \ '\. 12 i2 ‘1 i1. l io0 •••••• ,. /I2~12 ' / I ' L--~er::--~:--'-. \ \ I I is 2 Is I 'rs 1 p p0p1 p2 P 3 FIGURE 11 FIGURE -145- If the negative productivity shock is is transitory, Federal Reserve If policy eliminates any effect on interest rates rates but increases increases the price level by more than than the increase resulting from the transitory decline in productivity. The dotted lines 1s 151 and LM11 in Figure 1l show the 1 effect of the transitory change in u~. ut. Prices and and interest rates rates rise; p1 is of the price level level at at the the intersection intersection of IS 1 and and LM LM 1,, is the the log log of the price of IS 1 1 and i11 is the interest rate. rate. Federal Reserve policy shifts the LM curve further to to the right, shown by LM22,, restoring restoring the the interest interest rate i 0 and increasing the price level level to pp22; p22 - pp11 is the relative relative rate - of change in the price price level level resulting from Federal Federal Reserve policy, and pl - pp0 is is the the rate price increase increase caused by the in producrate of of price caused by the decline decline in produc— t iv i ty. tivity. — The mean values of of the transitory shocks shocks are zero so the effect of Federal Reserve’s Reserve's response to transitory shocks is is on on the variance variance of rates rates of of price change and not on their average average over time. AA policy of pegging interest rates increases increases the variability variability of the measured rates of price change resulting from transitory transitory shocks. Our earlier earlier finding that that the variance of the rate rate of of price price change rose rose during the shocks is consistent with this impliperiod in which there were oil shocks implication.11 cation 11 Suppose, however, that the negative productivity shock is is perpermanent, or or persistent, not not transitory. In this case, price level case, the price fluctuates around pp2 following the increase in in money to LM22.. 2 following the Because Because permanent be observed separately, or permanent and transitory shocks cannot be 11 ~There There are, of course, other causes of variability including including the shocks for money (Ct) (ct) and the Federal shocks to spending and the demand for Reserve’s Reserve's response response to these shocks. shocks. -145-146- separated reliably, reliably, people people must must decide decide whether whether the the observed observed rate rate of of separated price increase, p22 - pp,, the change in money, s’~, ~t' and other other changes 0 — have caused aa one-time price change or aa persistent change in in the rate of price change. If the inferences inferences drawn from available available information to believe that some part of the change in the measured lead people people to measured rates of price change and money are persistent persistent changes in the the rates of change, change, instead instead of one-time changes changes in level, the IS curve shifts shifts to the right. further to The size of the shift depends on the degree to The 12 12 which the anticipated rate of inflation, tPt+l - pt, rises. rises. ~ - ~ The Federal Reserve policy policy of fixing the interest rate at at ii 0 sustains the inference that the observed changes in prices and money reflect reflect aa persistent increase in rates of of change, change, not aa one-time one-time change change in levels. levels. The reason is is that, when when IS IS shifts to the right the policy policy of fixing interest rates rates requires the Federal Reserve to again increase increase the money stock, shifting shifting LM further to to the right. right. The additional changes changes in in money and prices reinforce reinforce beliefs about the persistence of the changes in in money money and prices. As As the perper- ceived and measured measured rates of of inflation rise, anticipated inflation rises, and there is aa further further rightward shift in IS. inAdditional in- creases creases in money are now required to to hold hold the market interest rate at i 0 Each Each increase in the stock of money money reinforces the belief that has been aa persistent change in in the rate of money money growth. there has Each increase in in the the equilibrium price level level reinforces reinforces the belief that the run of transitory, negative shocks to productivity produces a 12 A run of transitory, negative shocks to productivity produces a similar result. tPt+l is today’s period’s price. today's expectation of of next period's The rational expectation takes the form of indiof aa distributed lag, as indicated earlier, so expectations adjust gradually. cated earlier, so expectations adjust gradually. -147—147— rate of price change has increased. The Federal Federal Reserve’s Reserve's policy policy of The maintaining one—tine change maintaining the level level of interest rates converts aa one-time change in in the price level into into aa series series of of price changes that strengthen percepperceptions that there has been been aa change in the rate of change. Rational investors anticipainvestors “know” nknow the model, so they know that anticipa11 tions about the price level level adjust slowly because they and others others are unable to to separate persistent and transitory changes. holding the interest rates at ii 0 as long long as as the money stock grows. The policy of of implies that that the price level level will will rise ‘~ That is, is, as long long as as ~ tPt+l- Pt is positive, positive, the policy of of fixing interest rates will will require require the Federal Reserve to let the money money stock rise. rise. The Federal Reserve can eliminate eliminate the bulge in the money stock and in the measured measured rate of price change change by by raising raising the target rate of II have drawn aa dotted line at the intersection of of IS 1s22 and LM22 in Figure l1 to show the rise in interest rates required required to to keep keep the price level from exceeding pp33.. The dotted line shows that the required interest. interest rate is i i 22; ii 22—i -i11 is the additional increase in interest rates resulting from Federal Reserve policy. The increase ii2—i -i 11 is is temporary, not permanent. Once people recognize that the the money stock is constant, anticipations of rising prices decay; IS shifts to to the the left; left; the market rate of interest falls to i i 11; and the price level level falls between pp22 and p33.. (The precise level of prices is at the value of i11 on LM22.) .) The combination i1,, pp1 is the interest rate and price price level level comcom1 1 bination to which the econony economy moved following the permanent permanent loss of productivity. It is not an an accident accident that the economy eventually settles at the rate of interest interest i11 following the “anti—inflationary” "anti-inflationary" increase in interest interest rates rates to i i2; it is an an implication of the neutrality 2 it -148- of money. money. of allowed the money stock to rise, Monetary policy, at first, allowed then held the money stock constant, constant, eliminated the anticipation of rising prices and allowed the interest rate rate to decline. The The lasting effect of the interest interest rate policy is is aa higher price level. The amount of increase depends, of course, on the speed with which the Federal i~= i. interest rate target it~ i • Reserve abandons the interest 0 This discussion of policy has has neglected neglected many many complicating complicating feafeatures. The adjustment of of prices and interest interest rates has has been analyzed as if these changes occur occur without real effects. effects. The gradual adjustadjust- perment of of employment when rational individuals cannot distinguish distinguish per· manent and transitory productivity changes has not been emphasized. The case for fixing the level of of interest rates is not strengthened strengthened by by these omitted effects. A principal result of the policy of fixing market market interest interest rates A principal is that additional changes in prices prices (and output) are induced by by monemone- tary policy. policy. People People are forced forced to to decide how how much of the the observed observed and how much much is is transitory. change in money is persistent amd The The deterdeter- mination mew permanent price mination of the new price level level is is made made more more difficult. The permanent decline in in productivity productivity produces aa temporary temporary inincrease in unemployment unemployment and aa permanent loss loss of of real income. UnemployUnemploy- ment rises because people do do not recognize instantly instantly that that the shock is permanent. Hence, they do not instantly instant 1y adjust adjust their real rea 1 incomes ((and and real rea 1 wages) to the level 1eve 1 they eventually reach. Monetary Monetary policy po 1 icy can reduce this this cost of adjustment only if the monetary authority cam can sucsucceed in reducing real wages to their new, permanent permanent level without set~ set- ting off off anticipations of of rising prices. prices. The monetary monetary authority must have superior information information on the speed speed with which people recognize the th, 0 -149- permanent loss loss of of real real income income and and the the speed speed with with which which anticipations anticipations of of permanent price changes changes form and and decay. moneThere is no reason to believe that mone- tary authorities have information of this kind or are able to to set marmarket interest rates in in a way that minimizes minimizes the cost of adjusting adjustin9 to ket interest real shocks. shocks. On the contrary, contrary, monetary monetary policy produced persistently persistently higher rates of price change following the productivity productivity shocks shocks of this decade. THE CASE FOR GRADUALISM Reliance Reliance on market interest rates as as the operating target target of monetary policy produced high rates of of growth of the the monetary base and sustained inflation. sustained The The low variance of of the long-term long-term average growth of the base suggests that that the 8.5% growth rate of the the base is perceived perceived “permanent” rate of change. as aa "permanent" To end inflation the rate of of growth of the base must be reduced. If If expectations form and decay quickly in the presence of of new information, information, the problem of ending inflation is made easier. AA credicredi- ble expectations. ble policy policy to stop inflation inflation causes prompt revision of expectations. Revised expectations, expectations, and slower slower growth of of base money money bring inflation to to an an end. Rational Rational individuals recognize that that sunk costs costs or or contracts must be forgotten, so as contracts are revised, they enter into into agreeagree- ments or or commitments that reflect their revised expectations. Even in this case, there are benefits to gradualism if costs of adjustment can to learn about the new environment. be reduced by permitting people to The analysis in the preceding preceding section section suggests suggests some of the diffidiffiThe culties people face when when forming judgments about about the the persistent persistent rate of of change of money. Some of these difficulties cam can be be reduced if policy —‘SD— -150- V makers announce announce the intended rate of of money growth. makers not sufficient to to change anticipations permanently. Announcements are Announcements AA principal reason is that policymakers statements are not entirely entirely credible. credible. Past promises to slow money growth and and reduce reduce inflation inflation have been been followed within aa few quarters by renewed expansion. Consequently, rational rational exindividuals treat any initial reduction in money growth (or budget expenditures) penditures} as temporary, temporary, not permanent, permanent, changes. changes. An announced reducreduc- tion in the growth of money, initially, initially, will not be interpreted as as aa reduction in the maintained rate of money growth. Gradual reduction in money growth can reduce the cost of lowering the three ways. the rate rate of of inflation inflation in in three ways. First, of First, maintaining maintaining the the growth growth of the base at aa steady rate lowers lowers the variance of the transitory comcomponent and reduces the lag in in the formation of expectations. expectations. Second, the maintained average average rate of money growth falls gradually, so people have time to adjust future commitments to reflect revised expectations. Third, if if costs of adjusting to aa lower rate of of inflation inflation are are not proportional to the total adjustment but increase increase with with the rate per period, period, costs of adjustment are reduced by lowering lowering the rate per per period. If the rate of of adjustment of money growth is very low, low, the varivariance of the permanent component component is is low, ance of the permanent low, so so the the lag lag in in adjustment adjustment of of expectations expectations increases. If the rate of adjustment of money growth growth is component increases, rapid, the variance variance of of the transitory component increases, so so costs of adjustment adjustment rise. rise. The The optimum rate rate of of adjustment adjustment is achieved by inin- creasing component and reducing the varicreasing the variance variance of the permanent permanent component variance ance of the transitory component of of money growth. This This is is equivalent equivalent to to finding the minimum lag lag in the formation of of anticipations. -151— -151- pol icy of gradual, pre-announced pre-announced reductions In in money growth The policy advocated by the Shadow Open Market Committee Committee did not emerge as aa solsolution to the problem of finding an optimal lag. The choice of am an optimal policy depends on on information that is not not yet available. Our like most policies, depends more more on empirical judgments about proposal, like the length of lags lags and costs of adjustment than than on on hard evidence. I have no doubt that future research will find aa better path. path. SOME FINAL SPECULATIONS SOME The The chief difficulty in in the policy of gradualism is is the length length of time required to to reach the rate rate of of growth consistent consistent with mom-inflanon-inflationary growth growth in the economy. economy. If we use the long—run long-run growth of of real real output as aa guide, the rate rate of base money growth must must fall from the current rate rate of of 8% to to no more than 3%. 3%. If payments payments technology concon- tinues to improve, base base velocity velocity will will rise in the future as it has has for at least the past quarter century. The non—inflationary non-inflationary rate of base money growth is then no more than 11 or or 2%. in money growth Is a seven year program of sustained reductions in the best that can be done? I expect not. There is reason to believe believe that policymakers cam can increase their credibility by by meeting meeting pre— preannounced targets. targets. Increased credibility credi bi 1ity permits permits policymakers po 1 i cymakers to to lower 1ower the maintained growth rate rate while lowering the relative variance variance of the transitory component of of money growth. Credible announcenents mean Credible announcements mean that individuals distinguish permanent permanent changes closer closer to to the time they occur occur by using announcements of proposed changes as as a reliable indicator of of future money growth. -152—152— No one can be very certain about these issues. The evidence on from experience in Germany, Switzerland, the United which we rely comes from Kingdom and our own experience in the middle seventies. sevent1es. Each of these anticexperiences suggests that within two to three years at most, the anticipated rate of inflation declines. dee lines. The rate of price and wage change falls; long-term interest rates decline, and real output rises or accelerates. accelerates. Those who desire “incomes incomes policies” policies!! to reduce the lag for adjustadjust11 ment might might find pre—announced pre-announced monetary policies more attractive attractive than pnent either the failed incomes policies of the past or present, or complicated, inefficient programs to tax wage and price changes. Instead of announcing the rate of price and wage changes that the government favors, the government can announce the rates of monetary and fiscal expan.sion that the government intends to maintain. These announcements, expansion if they are credible, help individuals to form expectations about future rates of inflation. Analysis of the length of the lag lay in the adjustment adjustment of anticianticipermanent values pations relates these adjustments to the adjustment of pennanent maintuined rates of change. change, or maintained v-Je have is is neither neither inconinconThe evidence we expectati ans that I have sketched nor more sistent with the theory of expectations a;1y other o~her explanaion I have seen. This is not a strong consistent with any inclaim, but it is considerably better founded than the belief that In- flation is intractable. -153-153— REFERENCES REFERENCES (1g79) "Stagflation, Stagflatiom, Brunner, Karl, Alex Cukierman and and Allan H. Meltzer (1979) Persistent Unemployment and and the Permanence of Economic Shocks,” Shocks," multilithed, cC- ‘negie—Mellon ·negie-Mellon University. Relative Price Variability, Inflation Cukierman, Alex (1979) "Relative Inflation and the Allocative Efficiency of the Price System, System," multilithed, multilithed, CarneqieCarneoieMellon University. Equilibriun Model of the Business Cycle," Cycle,’ Lucas, Robert Robert E. E. (1975) (1975) “An "An Equilibrium Journal of Political Economy, 83 (Deceriber) (December) pp. 1113-44. Meltzer, Allan Allan H. Anticipated Inflation and Unanticipated Price H. (1977) (1977) ''Anticipated Change’ Journal of Money, Credit, and Banking, g9 pt. Change" pt. 22 (February), pp. 182—205. 182-205. Muth, John F. ‘Optimal Properties F. (1960) "Optimal Properties of Exponentially Weighted \leiqhted Forecasts,’ Statistical Association, 55 Forecasts/ Journal of American Statistical 55 (June) pp. 299-306. 299—306. 1 Muth, F. (1961) Expectations and Theory of Muth, John John F. (1961) “Rational "Rational Expectations and the the Theory of Price Movements,” Movements,'' Econometrica, 23 23 (July) pp. pp. 315—35. 315-35. ‘Slowing the Wage-Price Wage—Price Spiral: The MacroPerry, George L. (1978) "Slowing Macroeconomic View," View,” Chapter Chapter 22 in Okun Okun and Perry (eds.) (eds.) Curing economic Curi nq Chronic Inflation, Washington: Brookings Institution. —154-154- FEDERAL BUDGET POLICTES OF THE l97Os FEDERAL BUDGET POLICIES OF THE 197Os: l9SOs SOME LESSONS LESSONS FOR THE 198Os Michael E. Levy Levy Michael E. At the close 1970s, the public and the politicians alike close of the the 197Os, alike perceive inflation as as the foremost economic challenge of of the day. Other important econonic economic and and social social issues will will carry over over into the the l98Os; forsaken claims will be revived and new demands are bound to 198Os; surface. the But our effectiveness in in coping with all all these -- in fact the —- very survival of poof this this countrys country 1 s traditional economic, social, and political structures -- nay may well depend on on our ability ability to to contain and —- control inflation in the coming coning decade. There inflation control control nay may require There is a growing belief that inflation fiscal restraint, aa slowing of government spending, aa reduction in the size of the realized budget deficit. Yet, as as we approach the threshold of of the l9BOs, 198Os, II can think of at at least five major policy policy issues issues in in solutions, each each of which would place new claims on our fiscal search of solutions, resources~ resources. oo Half aa decade after the initial “energy llenergy crisis, crisis/' we are still in in search search of of an energy policy policy that generates widespread widespread public public and and political support support for economically viable solutions. oo to channel the hardcore unemployed into the mainmainOur efforts to stream of of our economy have yet to succeed. succeed. Michael E. Levy is Director, Director, Economic Policy Research at the Conference Board. Board. —155— -155- a.rms race -- even If if attained oo Success in slowing the nuclear antis —- through SALT SALT II II -- may have to be bought at the cost of accelerating -— defense spending for years to come. oo Welfare reform has been the subject of several aborted propospropos- it is bound to resurface as l980s. als of the 1970s; 7970s; it as a major major issue in the 1980s. unfulfilled social promoo National health insurance -- aa major unfulfilled prom-- ise of the 1970s 1970s -- is high on the public agenda of the coming decade. —- It It is all all too easy easy to add to this list list of enlarged public —claims -even at aa time when inflation inflation control control is our top priority priority and budget restraint restraint is promulgated. (Note that I have omitted any 11 mention of 11“safety’ safety" or “environmental envi ronmenta 1 issues.”) issues.") Such are the complexcomp 1 ex- ities iti es and contradictions contradictions of of budgetary policy which would would seem seem to place inflation inflation control practically practically beyond our reach. Yet my monetarist friends friends are able to collapse collapse the social and political conplexities of inflation political complexities inflation control control into the simple issue of “monetary "monetary integrity.’ integrity." To them, the deep—seated deep-seated inflation inflation of the last last decade-and-a-half is strictly strictly a monetary phenomenon. Its "cause" “cause” (like (like that of every inflation) inflation) was excessive monetary growth grov-,th reinforced, 11 perhaps, by aa few nasty “shocks,” oil price escalations escalations of shocks/ 1 such such as the oil 1973 and 1979. 1979. Its “cure” inflation) is secured "cure" (like (like that of every inflation) through aa persistent persistent slowdown in money growth. On aa purely technical level, all the answers. level, the monetarists have, of course, all In fact, some of my own econometric exercises have tended to reconfirm their their valuable, if somewhat simplistic, simplistic, generalizations.11 if 1Michael E. Levy, assisted by Steven Malin, International InfluInfl,!Michael ences on U.S. U.S. Inf1atiop~l97L-1976, Inflation 1971-1976, a study prepared prepared for the U.S. DeDepartment September 7, 1977 1977 (unpublished, available partment of Commerce, Commerce, September available from the author). -156—156- moneHowever, even if they were formally correct, these simple monetary propositions would tell the changes in social tell us nothing about the attitudes and national priorities which generated generated the political prespressures that bent the economic economic structure and drove the monetary printing press. They They provide no no clues as to how and why the economic and social social structure was changed and whether this this process process is is reversible reversible or . 2 2 cumu l at1ve. cumulative. 11 budgetary policy,” policy/' such as it it is, is, By contrast, analysis of “budgetary promises to shed some light light on on these unanswered questions, because the government budget is aa fulcrum of social and political change. UnforUnfor- tunately, it is difficult, at best, to chart aa course of fiscal and budgetary policy over years and decades. In fact, one may even question the existence of aa meaningful "course" than the drift drift “course” other than created by the complex and contradictory forces and events that shape the federal budget from year to year. if this "drift" “drift” were governed by a powerful current Obviously, if current 11 if “bends” bendsn in this current could be discerned, we should expect and if far—reaching economic implications, far-reaching implications, because the the federal budget powerpower- fully touches all all social groups, all segments of our our economy. I have interpreted my assignment as the search for such such bends in the current. 22For For more formal analyses that question the independent contricontribution bution of of money growth in 11“explaining” explaining 11 the inflationary process, see, see, example, Franco Modigliani and Lucas Papadenos, Papademos, "Targets Monefor example, “Targets for MoneYear," Brookings Papers on Economic Activity, Activity, tary Policy in the Coming Year,” ‘Slowing the Wage-Price Spiral: l1:1975, :1975, pp. 141-63; George L. Perry, "Slowing esp. pp. 45-46, Inflation, The Macroeconomic View,” View,' esp. 45-46, in Curing Chronic Inflation, Arthur N. M. Okun and George L. Perry, eds. eds., The Brookings Institution, Institution, 1978; also Martin Neil Baily, Baily, ibid., ibid., p. p. 58. Washington, D.C. 1978; 1 , —157— -157- VIETNAM: THE ORIGINS OF U.S. INFLATION There is widespread widespread agreement agreement that the the persistent persistent U.S. inflation of the last decade-and-a-half got under way ‘Keynesian’ exexway in in 1965 as "Keynesian" 3 cess expendirapidly escalating defense expendicess demand inflation. 3 In 1965, rapidly tures for the Vietnam War were superimposed on on aa full-employment economy that was was on the verge of aa private investment boom. Not only to enact timely tax increases (until the belated belated ten— tendid we fail to but our exuberant exuberant "guns butter" percent surcharge of 1968-1969), 1968—1969), but guns and butter’ guns and Great Society’) (or "guns Society") policy policy added new new and rapidly escalating civilian programs (Medicare, (Medicare, Medicaid, Medicaid, Food Stamps, Stamps, Job Corps, Model Cities). Cities). Vietnam War costs rose rapidly from about $100 million million in in fiscal 4 Total del969.~ de1965 to almost $29 billion at their peak, in fiscal 1969. 67 percent, during fense expenditures expenditures rose by nearly $32 billion, or 67 this period; and the the share of GNP devoted to national defense advanced from 7.2 percent in in fiscal fiscal 1965 1965 to 9.5 9.5 percent in fiscal 1968 -- its its —— high for the decades of the 1960s l960s and 1970s. Yet it it would be a mistake to attribute the persistence persistence of U.S. inflation first and foremost to to the the Vietnam Vietnam War -- even if one’s one's time —- preceding the oil oil crisis of late 1973. horizon is limited to the period preceding 3E.g., see Perry, bc. lac. cit., p. 23. Note, however, that some monetarists have pointed out that the onset of this inflation was prepreceded by about two years of what was considered at at that time rapid monetary monetary growth. 4These are "full-cost" full_cost estimates. estimates. For further details and for “incremental—cost” incremental-cost estimates, see Michael E. E. Levy with with Juan de Torres, Torres, 11 11 Delos R. Smith and Vincent Massaro, The Federal Budget: Budget: Its Impact Imoact on the Economy, fiscal 1973 1973 edition, The Conference Board, Board, iT~iYbFk, New York, 1972, 1972, ~p.-27. esp. pp. 26-27. -158- From fiscal 1969 through fiscal 1973 annual expenditures for Vietnam dropped by about $18 billion billion in current dollars -- the decline in r~I real -- terms tei:-_ms was, of course, much greater -- while total defense expenditures expenditures -- declined by by nearly $5 $5 billion. billion. The share of GNP devoted to national national defense dropped from its 1968 peak of of 9.5 percent to 66 percent in fiscal 1973 and continued to decline to 55 percent by fiscal 1979. 1979. Yet the large Vietnam “peace "peace dividend” dividend" of of the early 1970s 1970s brought no end to U.S. inflation. inflation. When the 1970 recession barely barely reduced the inflation inflation rate, rate, a ninety-day ninety-day wage and price price freeze was introduced on on August 15, 1971. 1971. It followed by four phases of wage and price controls that lasted through the third quarter of 1973. April 1974.) ended in April (The final decontrol phase Yet these controls controls brought, at best, a modest and inadequate respite, respite, before the quadrupling of OPEC oil oil prices pushed the economy into double-digit inflation inflation in 1974. 1974. “SHOCKS” AND THE INFLATION OF THE l97Os "SHOCKS" 1970s A significant if not a major A significant part, part, if major one, of the inflation inflation surge of 1973—1974 that resulted 1973-1974 resulted in double-digit double-digit inflation inflation has been been attributed attributed to special factors -- "shocks" “shocks” of aa largely largely international international nature. nature. —— Three Three distinct distinct inflationary inflationary influences influences deserve to be be distinguished: oo The depreciation of the external value of the dollar. dollar. (It (It got under way around mid-1970 and accelerated after after the closing of of the “gold window" window” on August 15, 1971, hitting "gold hitting bottom in July 1973.) oo The escalation of agricultural conriodity commodity prices, prices, particularly particularly grains, 1973. grains, from late late 1972 1972 through 1973. (It (It was caused largely by the prior depletion of U.S. disappearU.S. agricultural stocks, stocks, the temporary temporary disappearance of the Peruvian anchovies, anchovies, bad bad weather and poor crops crops in in many —159— -159- parts of of the world in 1972, the “Russian "Russian wheat deal” deal" of 1973, and the worldwide boom that raised consumption of high—protein high-protein foods.) oo The sharp rise in the prices of fuels and some industrial industrial commodities, but mainly the quadrupling of OPEC oil commodities, oil prices during the last quarter of 1973. Elsewhere II have described these special special events and reviewed the 5 best available evidence as to to their impact on on U.S. inflation. inflation. 5 This significant before combined inflationary inflationary impact seems not not to have been significant mid- or late 1972. It increased rapidly thereafter, thereafter, appears to have peaked during the second half half of 1974, 1974, and faded durinq the second half of 1975.6 1975. 6 On On the basis of econometric estimates, I concluded that “the "the joint major identifiable identifiable 'international acjoint impact of these major ‘international shocks' shocks’ accounted for about 5.5 percentage points -- or roughly 60 percent -- of —- —— the dramatic increase in the inflation inflation rate of the implicit implicit GNP deflator from about 3.5 percent (annual rate) in the second half of 1971 to around 12.5 percent in the second half half of 1974. 1974. The elimination of this shock—induced shock-induced inflation inflation during 1975 accounted for over 70 percent of the decline dee 1i ne in the inflation i nfl ati on rate rate of the GNP deflator defl a tor to to an average 7 l976.”~ of about 55 percent by by the second half of 1976." Research evidence developed more recently recently leads me to believe that these estimates of international international influences influences on U.S. inflation inflation may “shock effects.” well represent upper limits limits of these "shock effects." In any any case, the 5 5Michael E. Levy, assisted by Steven Malin, International InMichael E. Levy, assisted by Steven Malin, International Influences on 1971—1976, op. cit., on U.S. Inflation, Inflation, 1971-1976, cit., esp. chap. 1. 1. 6lbid., chap. 4, 1bid., esp. Table 10. 7lbid., p. 8. Ibid., -160- evidence suggests that U.S. inflation would have remained substantial throughout the first half of the l970s 197Os -- though well below the double—- digit digit level level -- even in absence of of these special special price-escalating price-escalating interinter—— national developments. In fact, aa convincing case could be made made that the “basic” "basic" inflation rate embedded in the U.S. economy was trending trendfog higher, irregularly but persistently, during the last decade-and—a—half decade-and-a-half and that that this this uptrend was masked mainly mainly by temporary deviation deviation caused by the controls of the the early b970s 797Os on the one hand, and by special other.88 Not even the 1974-1975 international shocks shocks on the other. 1974-1975 recession recession by far far the -- by the most most severe severe of of all all postwar postwar declines declines —— —— was was abbe able to to brake brake this bong—term long-term (1965-1979) uptrend of U.S. inflation inflation rates. “INFLATIONARY EXPECTATIONS” AND “INFLATION "INFLATIONARY EXPECTATIONS" "INFLATION INERTIA’ INERTIA" Most econometric models designed to explain this persistence persistence of 11 U.S. inflation inflation have assigned aa major robe role to to ‘inflationary inflationary expectations” expectations11 that infbuence influence future wage agreements agreements and pricing patterns, and to inincreased “inflation "inflation inertia” inertia" (a (a concept which which implies simply that the inflation persists, persists, the more persistent persistent it it becomes). longer inflation In the words of one beading expert “the one leading "the significance of ongoing inflation has has risen together with the rising rate of inflation.99 risen together with the rising rate of i nfl at ion." To the layman, this this may seem aa bit like bike aa dog chasing its own tail, but for the econometrician, the loop loop has has been been closed: 8 econometric econometric This uptrend is clearly illustrated by Perry, bc. cit., esp. 8This uptrend is clearly illustrated by Perry, loc. cit., esp. 1, when the two periods periods babebbed labelled “Controls "Controls (1972—73)” (1972-73)" and p. 24, Table l, “Food-fuel explosion (1974-75)" (1974—75)” are excluded. "Food-fuel explosion excluded. The The batest latest international international shocks came from the rapid slides slides in the value of the dollar in 1978 (until November) and in 1979 1979 (May (May through October), October), and from the 1979 round of OPEC oil oil price increases. 9Perry, loc. cit., P. p. 37~ 37. boc. cit., -161-161- requirements for aa technical “explanation” explanation have been satisfied. The 11 11 end result of these elaborate econometric exercises is a widely acacPerry ca calls it aa nmai “mainline cepted model mode 1 —— -- Perry 11 s it n1 i ne model” modeP —- that explains 15 15 inflation on years of accelerating U.S. inflation on the basis of aa few initial initial price escalations years of excess demand, aa few years of orice escalations caused by ‘inflationary expectations” special ‘shocks,” "shocks," and a lot lot of ;'inflationary expectations" and “infla"infla- tion inertia” inertia" designed to link and extend these inflationary spurts and to bridge all all the intervening years when inflation inflation should have subsided 10 but did not.10 -- II would like to propose aa somewhat different different approach: would like a search for fundamental changes in our economic and social system that appear mid-1960s and persisted -- if if not gained to have originated in the mid-196Os —- momentum -- during the past decade—and—a-half. decade-and-a-half. -— If If such structural structural inflationary changes could be identified, identified, and if if they carried strong inflationary implications, persistimplications, they would go go a long way toward explaining the persistinflationary expectations ence of inflationary expectations and the increase in inflation inflation inertia. inertia. Analysis of U.S. of the last two two decades Analysis of U.S. budgetary budgetary policies policies of the last decades be extremely extremely useful in this search. proves to be 10Leading lOLeading supporters of the “mainline umainline model” modelu are well well aware aware of “From 1975 through 1977, all this difficulty. difficulty. Thus, Perry notes: "From all available measures of tightness in either labor markets or product product marmarkets registered ample slack. slack. And no no large large upward movements movements have ococthe price level since the Organicurred in particular components of the Organioil prices in 1974. zation of Petroleum Exporting Countries increased increased oil Yet despite all all these disinflationary disinflationary developments, the rate of inflainflation, by by any broad measure, has continued at aa historically high rate and now shows signs of creeping still still further upward." upward.” -162— -162- U.S. BUDGETARY BUDGETARY POLICY: POLICY: U.S. LOOKING FOR TRENDS Analyses of budgetary budgetary policy often tend to be be too global in in apapproach, focusing mainly on what what is perceived to be be the overall expanexpan- sionary (or (or restrictive) impact of the the budget on the the economy. economy. Because Because of neo— of our narrow narrow preoccupation with with “fiscal !!fiscal policy” policy 11 as a major neoKeynesian tool for economic stimulation (or restraint), we we have tended to to lose lose sight sight of the more complex ways in which the the size, composition, and rate of of growth of of the federal budget may affect the economic system. tendency to to focus on short periods -- usually aa Moreover, the tendency —— single fiscal fiscal year or two -- and excessive reliance reliance on simple, rather rather —— inadequate, “fiscal impact” inadequate, measures of "fiscal impact" (such as the “full—employment "full-employment budget surplus") has compounded the the myopia of traditional fiscal surplus”) has analysis. Since II have have chosen U.S. inflation inflation as the focus for the present present review of federal budgetary policies, policies, II am am concerned concerned mainly mainly with longer—term longer-term trends and their their implications, implications, rather than with with short—term short-term fiscal impact. Such an an analysis analysis should pay pay special attention to those budget components that tend to create special inflationary inflationary pressures. pressures. It seems to me that national defense spending and transfer payments to individuals individuals deserve special attention attention in this context. context. Defense expenditures expenditures have an an inherent inflationary inflationary tendency. They create employment and income, but do not produce produce any “market "market goods,” goods," nor do do they yield the the kind of of “public "public benefits” benefits" that are perper- average consumer as an an immediate enhancement of of wellceived by the “average consumer” as imediate enhancement well11 11 care, education, education, or or police being (as, say, public spending for health care, and fire protection). This inflationary inflationary tendency of defense spending -163—163- pronounced in the case of war expendiexpendibecomes, of course, particularly pronounced tures. Among civilian civilian programs, transfer payments to individuals give rise rise to special inflationary pressures. Designed to into redistribute in- sector (often in favor of of the poor and the come within the private sector needy), transfer payments tend to increase short-tern short-term inflationary inflationary pressures if if the income gainers tend to to spend aa higher proportion proportion of of their “contributors” (as is usually their marginal income than the "contributors" usually the case). genMore important important for for the the present present analysis, analysis, these transfers transfers tend tend to to gen- erate longer-term inflationary pressures in at at least least two distinct distinct ways: ways: o They impair incentives to to work and to to invest among the “conllcon- 11 Reductions tributors,” gainers.~ tri bu tors, n if if not also al so among the income gainers. Reductions in productivity and in growth of GNP are the more inproductivity gains gains and in growth of real real GNP are the more obvious obvious inflationary consequences. o If the ucontributors “contributors” consider themselves reluctant losers 11 (rather than “voluntary donors”) -- as may often be the case -- they "voluntary donors") —— -- will strive to recapture what they consider their their “rightful” "rightful" (e.g., (e.g., traditional traditional or expected) share share of real income, or real growth. If If the “losers” losersl! are concentrated in the productive productive sector of the private private 11 economy, nonproducers, this economy, while the income gainers are mainly nonproducers, this attempt at ''recapturing “recapturing rightful itself in wage and price rightful shares” shares'' will manifest itself escalations. 11 ~The The list list of of theoretical studies studies and empirical research on on disdisincentive ; ncenti ve effects on on “income u income gainers” gainers from from unemployment unemployment insurance and 11 too extensive for review here. Lately, welfare payments is too Lately, additional additional evidence on this subject has become available from analyses of of various “negative negative income—tax income-tax experiments.” experiments. 11 11 -164- With these analytical considerations in mind. mind, II have reviewed trends in total federal well as national federal budget outlays as we11 national defense defense exex- 12 Denditures Jenditures and and transfers to individuals.12 in Chart 1l and Table 1. The results are summarized Unemployment compensation has been been excluded excluded from transfers to to individuals as shown there (but not from my own dedetailed analyses) because its large cyclical fluctuations tend to mask the trends that that concern us here. FOUR PHASES OF NATIONAL DEFENSE DEFENSE SPENDING National l960s and 1970s may National defense defense expenditures of the 1960s may be divided into into four distinct phases: (1) the “cold war” phase "cold war" phase preceding Vietnam; Vietnam; (2) the the escalation phase of the Vietnam War (fiscal 1966 1966 through 1968); 1968); (3) the de—escalation de-escalation phase until until the completion of the 1973; and (4) the the recent post-Vietnam troop withdrawal in February, 1973; Only during the escalation phase did defense spending grow grow much much phase. phase. faster than l960s, it than GNP; GNP; during the the pre-Vietnam pre-Vietnam phase of the early 1960s, it barely advanced, and during the deescalation phase it it declined rapidly (see (see Chart 11 and Table Table 1). 1). More recently, the the growth rate of defense spending has accelerated, but it it has remained below below the growth rate of GNP. latest uptrend continues continues (as is is suggested suggested by the current If this latest political political climate and initial initial congressional debates of the SALT SALT II II 12 For the analysis of of transfers to to individuals, individuals, unpublished tabtabulations from the Office of of Management and and Budget on direct and inindirect “payments payments for individuals” individuals were were used, rather than than federal transfer payments to individuals individuals as as tabulated for the national-income— national-incomeaccounts (NIA) budget. budget. The former data are more appropriate for the the analysis at hand, since they include, include, for example, both both Medicare and Medicaid, Medicaid, while the NIA data data treat Medicaid as a purchase of of health services services by state and local governments. 11 11 -165—165- cnart I.1 Chart FEDERAL BUDGET OUTLAYS BY MAJOR COMPONENTS, 1961-1979 COMPONENTS, FISCAL 1961-1979 Annual Growth Rates Outlays as aa Percent Percent of GNP GNP w, Total Budget Outlays total 30 , - - - - - - - - - - - ~ - - - - - - , Outlays Total Budget Outlays 25 , - - - - - - - - - - - - - - - - - - - , 25 25 20 20 20 ,0 5 Of---->---------------, Payments for Individuals * 35 , - - - - - - , - - - - - - - - , - - - - - , - - - - , Payments Payments for for Individuals* IC 30 25 20~ 20 15 10 "New Ec.oriom,cs·· Social Actv,an, Eoonom,cs I 11110, "'Social Activism·' Soc,al Act,v,sm I C 0 National Defense 20 15 10 0 -5 National Defense Defense 7Un “cod War” t,on Ceescalauion “Vietnam War” H ‘ Post Vietnam -1 0 ' - - ' - ' - ' - ' - ' - ' - ' - - ' - - ' - ' - ' - - - - ' - - ' - ' - - ' 1977 1965 1973 1969 1977 1961 1969 c “Cold War” ,951 ‘961 lion Oe°scaIation Deescalation , “V,etr,am War” "Vietnam War" 1965 965 1969 1969 Post’V,etnsm ?ost·Vietnam 1973 c~de,ei1a.q,~,,~,d,act h,~*aseac’,~a~ ha,* as !M ii,. ,a,o~ ** ,nckides a,I (l~ecl at><! •nd~ect ‘,,,,ie’ trnt1sier oay,,el,’s,.caaiL~,.mØoyrne,,~c,nen,ai,.,. payments ••ceot urn,mp<oymen, comoensatt<Jn, ·,,tnc~ was e.cSuded Mere maier cyc;’c* cyc,«:aa compOf'Ml 5o,,,casI ,dS,oeqet, Oo,ini,,Oa Board Sou,casc Onion,, Office of Ma,ia~ernei’t Management o ar'\d S~et ma The come,ence —166-166- IC 1977 1977 Table I1 Selected Data for Analysis of Federal Budget Policy, Fiscal 196l~1979 1961-1979 forAnalysis 7961-65 1961-.65 1966-79 19$6.79 7967-65 196).1;!5 Averaoe ~ Average Average Average Tome) Total Budget Outlays Payments lndiv,duaia’ Payments for Individuals" National National Defense 5,2 5.2 6,6 6.6 11 1.1 10.8 10,8 15.3 5.7 6.7 5.2 6.6 1.1 Productivity Productivity Real ONP GNP Inflation (Implicit GNP GNP Della tor) Intlelion (implicit Deflator) 3,2 3.2 4,2 4.2 1.5 1.5 1.6 2.9 2.9 5.9 5,9 3,2 3.2 4,2 4.2 1.5 Total Budget Outiays Tote) Outlays Payments Payments for individusls’ lndividua1s· National Defense Deticit Budget Deficit 19.1 19.1 4.3 8.5 5,5 0.8 21.0 7.1 6,8 6.8 1.7 19.1 Fiscal Thrust Thrust Expenditure Component Expenditure component comoonent Revenue Comoonent 1.4 1.4 i1.0 .0 0.4 2,0 2.0 2.1 0.0 0.0 '1.4 1.4 1965’Se ,-, t97o’73 1970-73 Average Avernge 1974’79 1974-79 Aiterste Averaoe 11.9 11,9 16.1 14.1 14.1 7.6 7-6 15.9 -1.6 —1.6 12.3 12,3 14-5 14.5 7.4 2.4 4.6 3.6 3.6 2.0 3.2 3.2 4.9 4.9 0,7 0.7 1.6 1.6 8.0 20.3 20.3 5,0 5.0 8,7 8.7 1.1 1.1 20.5 20,5 6.7 67 7.~ 7.1 1.5 21.7 21.7 8.8 5.4 5,4 2,3 2.3 ;_5 1.5 2.2 ‘—0.7 -0.7 18 1.8 1.7 0,2 0.2 2,5 2.5 2.3 2,3 0.1 01 Avervge Average Annual Annual Growth Growth Rates Percent o! GNP Percent ol GNP 4.3 8.5 8.5 0.8 0.8 10 1.0 0.4 0.4 ‘tnclucex elI dorect paymetttt, except unemployment compensauon, compentatlon, whicn wticfl Wee Ina,or cyclical com•includes all direct and and ,nditect ,ndirect Ireotter transter payments, was excluded e~cluded here sa as the the ma1or cycl!cal component. conference soatd. Sources: Oftice Office of of Manegement Management Soc anc euoget: 8uoget The The Conference Soard -167—167— be reached when the share off GNP devoted agreement), aa point may soon be 0 to national national defense will be rising again. with the exception exception of the early Vietnam Vietnam War escalation -- its But with -- contribution l960s was discontribution to to the the inflation of the second half of the 1960s discussed earlier earlier -- defense spending as aa percent of of GNP has has been declindeclin-- ing. The decline in the share of GNP devoted to national defense could have been expected to moderate moderate (rather than stimulate) inflationary l970s. pressures during the 1970s. TRANSFER PAYMENTS: PAYMENTS: THE BEND BEND IN IN THE TREND Transfers to individuals individuals present present aa drastically different different picture. Fiscal 1965 1965 marks aa clear dividing line between the moderate growth of these transfers during the first half half of the the decade and the much higher growth rates that began with fiscal 1966 grov1th 1966 and lasted at least through 1). fiscal 1977 (see Chart l). During fiscal years 1978 1978 and 1979, the growth grov1th of of transfers to individuals slowed significantly. The The share share of GNP redistributed through federal transfer programs programs rose rapidly and 4.2 percent persistently from 4.2 percent in fiscal 1965 1965 to 9.1 percent in in fiscal 1976 1977; it it declined slightly during during fiscal fiscal years l978 and 1979. 1976 and and 1977; declined slightly years 1978 and 1979. Clearly, it is much too early to Clearly, it to tell whether fiscal 1977 1977 marked the end of the rapid-growth rapid-grov1th phase of these these transfers transfers and the beginning of of aa ,ihether it it represents simply simply aa new phase of relative containment, or whether brief "pause." “pause.” Whether pause pause or change, this is the first noticeable 13 downward deflection 1966.13 deflection in aa trend that started in in fiscal 1966. 13 Note that payments for individuals grew gre,i at at an average annual rate 1966-1979, compare rate of 15.3 percent during fiscal 1966-1979, comoare with v1ith 6.1 percent during fiscal 1961-1965. As As aa percent of GNP, these payments averaged 4.3 percent in fiscal 1961—1965, 5 5 percent fiscal 1961-1965, percent in fiscal 1966-1969, 6.7 6.7 percent in 1970—1973, and 8.8 percent in 1974-1979 1974-1979 (see (see Table 1). in fiscal 1970-1973, -168— -168- Clearly, fiscal 1965 marked a watershed for transfer programs: it was “New Economics” mew “Social the end of the uNew Economics!! and the beginning of aa new !!Social Activism." Activism.” The relatively relatively moderate growth growth of transfers to individuals during first half half of the 1960s 1g5os reflected the basic policy policy approach to the the first Kennedy Administrations “New Economics. Economics.” Administration's "New 11 The acceleration of real growth the reduction in the unemployment rate growth amd and the rate were to be be achieved through stimulation of the private sector, rather rather than through public public programs and an expansion of the government sector. sector. The The major policy tools were the liberalized liberalized depreciation of 1962, the investment tax tools credit credit of 1963, and the corporation and personal personal income tax cuts cuts of 1964 and 1965. 1965. The New Economics proved remarkably successful. During During 1961—1965, the unemployment rate declined gradually toward the fiscal 1961-1965, 4 percent full-employment target (as defined in the l96Os), 1960s), real GNP grew at at an an average annual rate of 4.2 percent and annual productivity productivity gains averaged 3.2 percent. All All these were far better performances 1970s, yet price price stability stability was preserved than those obtained during the 1970s, right up to the onset of the Vietnam War. The assassination of President Kennedy in 1963 and, in its its wake, Johmson, the passage of the Civil the assumption of power by Lyndon B. Johnson, Rights Act in 1964, and the burning burning of of the inner inner cities cities during the long, hot summer of 1965, ushered in aa new era of “Social nsocial Activism.” Activism." President Johnson -- one of the great parliamentarians of of this century -- and aa great admirer of President Roosevelt’s Roosevelt's New Deal -- secured the -- passage of far—reaching far-reaching new social and economic legislation; this inincluded the Economic Opportunity Act of 1964, the Permanent Food Food Stamp Act of 1964, the Social Security Amendment of 1965 which created —169-169- Medicare and “Medicaid,” Medicaid, and the Demonstration Cities and Metropolitan “Medicare” 11 11 11 11 Cities” Development Act of 1966 which established the new “Model "Model Cities" program. program. Many of of the the new new federal took the transfers to Many federal programs programs took the form form of of transfers to 1966—1968 individuals and and expanded expanded at aa very rapid pace even during the 1966-1968 expansion of the Vietnam War. expansion phase of In fiscal 1965, federal expendiexpendi- tures for Food Stamps, Stamps, Medicare and Medicaid were negligible; negligible; by fiscal 1968, they amounted to $0.2 billion, billion, billion, $5.3 billion, billion, and S2.0 $2.0 billion, 1978. the latest year for which respectively; and by fiscal 1978, which actual actual data (rather than estimates) estimates) are available, available, they had risen to $5.5 billion, $25.2 billion -- for aa combined total equal to 2.0 $25.2 billion, billion, and $10.7 billion —— percent of GNP. This rapid expansion expansion of social programs with with heavy reliance on on transfer payments extended from the second half of the 1960s 1960s through the the 1970s. After beneAfter repeated large adjustments in Social Security bene- fits far in excess of inflation, the entire Social Security Security program program was put under the umbrella of of aa cost—of-living cost-of-living escalator clause in in 1975, while real after-tax take-home take-home pay of many workers workers and real returns returns on on investment were lacking such protection and declined during aa major major part of of the 1970s. 1970s. the producing producing to the nonnonRapidly growing transfers, mainly from the retired, the disabled, the nonworking producing sectors (such as the retired, were financed in what would appear to to be highly inflationary inflationary poor), were ways: oo By frequent large increases in Social Security taxes which are, in the view view of of many economists, economists, among among the most most inflationary inflationary taxes. —170— -170- oo contributed to excessive money By large budget deficits that contributed 14 growth.14 oo By inflation itself which fattened the the federal federal government’s government's income—tax take, eroding real of take, while while eroding real after—tax after-tax purchasing purchasing power power of income-tax workers and real after-tax after-tax return on investment. The limited statistics available on on the subject subject tend to to confirm this erosion erosion of real real purchasing purchasing power power of of the the producing sector. sector. For For after—tax weekly earnings of nonfarm production workers example, real after-tax workers -- the the best measure measure available available from from the the Bureau Bureau of Labor Labor Statistics --- —— grew at rate of of 22 percent grew at an an average average annual annual rate percent during during 1948—1965, 1948-1965, as as comcompared with 0.1 1966-1978 (see (see Chart Chart 2). pared with 0.1 percent percent during during 1966-1978 2). Even Even after after allowing for all the limitations of these data, the sharp erosion since since 14 ‘4While While there there is no simple, positive, short-term short-term relationship between budget (e.g., deficits deficits may may be between budget deficits deficits and and inflation inflation (e.g., be induced induced or or enlarged by by aa recession which also tends to to curtail inflation), perpersistent high budget deficits during relatively prosperous periods periods exert exert strong upward growth. This This linkage was illuminated strong upward pressure pressure on on money money growth. linkage was illuminated during 1979 testimony testimony of of Paul Volcker, Chairman of the during the the September September 5, 5, 1979 Paul Volcker, Chairman of the Federal Reserve Board, before the Rouse House Budget Committee. Representative "There are those who say say there Representative Simon: “There is no relationship between money money supply and the money supply 2 policies the Fed deficits .... How depolicies of of the Fed and and our our deficits? How do do you you describe it and what what kind kind of of relationship is there between that increase in the money supply and the deficits?" deficits?” Mr. Mr. Volcker: Volcker: “The "The degree degree to to which the budgetary budgetary defideficit puts on the Federal Reserve, cit puts pressure pressure on the Federal Reserve, puts puts pressure pressure on on the credit markets and and through the credit markets pressure pressure on the the Federal Reserve to increase the money supply, depends deoends great deal on what else is is going deal on what else going on. on. And And the the relationship relationship aa great becomes much more difficult in boom period than in rebecomes much more difficult in aa boom period than in aa recession period. period. But all things things equal, equal, over aa period of time, the the deficit means at the the very least that credit marmarkets will be be tighter than than they otherwise would have been with a constant constant Federal Reserve Reserve money—supply money-supply tarqet and that that the money-supply money-supply target will have to be be increased, increased, which in turn has inflationary inflationary repercussions.” repercussions. I! —171— -171- chart Chart 2. ANNUAL PERCENT ANNUAL PERCENT CHANGES IN REAL AFTER-TAX WEEKLY EARNINGS Private Nonfarm Production Workers 5 7 - :..... •. ' -- i 3 no Dependents Dependents Workers with with no -5 c__J 5 --, nfl rfl -5 —5 '50 ‘50 '55 ‘55 '60 ‘60 '65 ‘65 Sources: Surnau of Laoor Slat1st1cs: The Conference Conlerence Board. Soard. Sources: Bureau 0’ Lacor Stat,st,ca: Tb 5 —172— - 172- 70 ‘70 ~ ~ Workers with Three Three Dependents Dependents 1948 ~ 75 ‘75 1978 1965 is obvious. In Its its 1979 Annual Report, the Council Council of Economic Advisers discussed discussed the erosion of investment incentives and stressed 15 the need for stimulating investment. 15 After After reviewing four alternate measures of of profitability, the CEA concluded: “Of "Of the the four measures of profitability, only one, one, the rate rate of of return on stockholders’ stockholders' equity, has has regained the 1955—70 1955-70 average. The other three are well below below the 1955—70 average and still further below the average for 1962—66, 1955-70 1962-66, when when 16 16 investment outlays rose very strongly.” strongly." Not only were investment investment incentives eroded in the 1970s, but a investment had to be devoted to “nonlarge and increasing increasing amount of of investment ''nonproductive enviromental regulaproductive uses” uses in order to to meet new safety and envirornental regula11 tions. In In this setting setting of poor real after-tax gains for workers and low investment incentives, productivity productivity and real growth could be be exexpected to to suffer. In fact, average average productivity productivity gains have have been dede- clining steadily steadily since the first half of the 1960s of l960s and real growth of 1970s averaged well below that of the the previous previous decade. GNP during the 1970s (For details, see Table 1.) Thus, not only did the federal government redistribute redistribute aa steadily rising share of real real income -- mainly from the producers producers to nonpro— nonpro-- ducers -- but this this redistribution appears to have contributed to, and -— was in turn affected by, aa slowdown in real growth. Thus, workers to sizable real—income real-income conditioned during the l950s 1950s and early 1960s to 150p._cit., op. cit., pp. pp. 124-34. The CEA concluded: “If "If the investment 1983 is to be realized, policy needed to to reach our economic goals in 1983 actions are required required that will strengthen investment investment incentives and and rererisks” (p.130). It went on to duce investment costs and risks" to recommend recommend 11 “tax tax reductions reductions designed designed to to strengthen strengthen investment investment incentives.” incentives. 11 16 p. 129. Ibid., p. 129. —173— -173- gains were doubly disappointed as they received a smaller part of a more slowly growing pie. In such such an environment, attempts to restore real gains of of workers through higher wage demands, demands, and to to shore up profitability through price increases, could be be expected expected to to recur frefre- quently, since since they were were bound to fail against the power of of the federal government to enforce its its own priorities. 11 11 In the struggle to to recapture aa ‘fair fair share” share of real real income growth the patterns of an an earlier earlier and happier period), (probably based on the strongly positioned groups could be expected to do better than than those those in in relatively weaker weaker bargaining positions. Thus, highly paid skilled workers workers and strong unions unions would experience less erosion of real gains unskilled or unorganized labor. than unskilled evidence presented by Some recent evidence Perry indicates that this is precisely what happened in the the l970s. 1970s. He concludes that "for “for the eight years years as as aa whole whole (1970-77), union wages have risen an average of of 11 percent aa year faster [than average wages]. But But while while they they have have outpaced average average wages wages over this period, period, the the 1.7 percent percent average annual increase in real wages wages in the union sector during the 1970s just just maintained the average rate rate of real wage increase 77 of the previous decade.’ decade. ,,17 During the 1970s, the federal government -- unwilling unwilling to to adad—— just its own inflationary policies and priorities -- applied wage and —- price price freezes and controls intermittently. These "incomes “incomes policies” policies" were were intended to suppress suppress inflationary pressures from the the private private proproductive sector that had had been created, or at least intensified, by by the the ductive sector government’s government's own policies. In order to to minimize minimize the political 17Loc._cit., pp. 31-32. Loc. cit., pp. -174- pressures that arise from large and frequent tax increases (and that ultimately led to to the “taxpayers’ "taxpayers' revolt” revolt" of the late late 1970s), 1970s), the federal government Social Security taxes mainly on increases in Social government relied mainly (which (which are less “visible” "visible" and create less popular resistance than than income income—tax income taxes), on the inflationary feedback that swells income-tax receipts it erodes erodes real receipts as as it real after-tax after-tax buying buying power, power, and and on on deficit deficit financing. financing. During fiscal 1961—1965, annual annual federal During fiscal 1961-1965, federal budget budget deficits deficits as as percent of percent; this of GNP GNP averaged averaged 0.8 0.8 percent; this percentage percentage rose rose steadily steadily to to aa percent 19701966-1969; 1.5 percent during fiscal 19701.1 percent during fiscal 1966-1969; 1974—1979 (see 1973; and 2.3 percent during fiscal 1974-1979 (see Table Table 1). 1). FISCAL FISCAL POLICY: POLICY: THE “FISCAL THRUST" THRUST” OF OF THE 1970s THE EXPANSIONARY EXPANSIONARY "FISCAL THE 1970s sketched some some of the processes processes through II have have sketched of the through which which the the diversion diversion of an increasing share of GNP to to transfers transfers (mainly from the producing producing to the nonproducing sector) added inflationary inflationary pressures pressures after after 1965. 1965. Implicit Implicit in in this this analysis were the the following two propositions: propositions: oo Direct Direct and and indirect transfers to individuals, jointly with of fiscal growth growth over national defense speniding, spending, dominated the patterns of the last decade-and-a-half. decade—and—a—half. (But (But except except for the Vietnam Vietnam escalation phase, transfers were by far the most prominent component component shaping fisfiscal ca 1 growth.) growth.) oo The budgetary The budgetary policies policies and and processes processes described described here here resulted resulted in expansionary budgets budgets in in the l970s than had been in far far more more expansionary the 1970s than had been the the case case in in the the previous decade. Moreover, expansionary thrust Moreover, this increased expansionary originated from rapidly growing spending programs (mainly transfers), rather than from tax reductions. —175-175- The extent extent to which which the first proposition is true may be gleaned from Chart l. 1. To my knowledge, the second proposition proposition is is new and has, so so far, been unproven. Therefore, it calls calls for empirical investigation and evidence. Until recently, I had suspected but had had been unable to document document satisfactorily satisfactorily that, on the average, fiscal policy of the l970s 1970s had been been more expansionary. With the the cooperation of the the Bureau Bureau of Economic Analysis of of the Department of Commerce, II have been able to develop reasonably consistent (preliminary) quarterly quarterly and annual annual estimates of ‘fiscal thrust” "fiscal thrust'' back to fiscal 1959 1959 -- just just in time for this meeting meeting (see Table 2).18 “expenditure component" component” 2) . 18 This measure measure consists of of an "expenditure 19 which measures measures change in in autonomous autonomous government government expenditures, 19 and aa -- ‘revenue componentn conponent” which measures the initial !lrevenue initial revenue loss (expansion(expansionary (+)) or revenue gain (restrictive ctiirajchan9es (restrictive(-)) from a structural changes in in (+)) (-)) tax tax provisions (rates or base). base). Each component, as well well as as total “fiscal "fiscal thrust” thrust" (their sum) sum) is best measured as as a percent of of GNP, in 18 . “. 18 I coined the term fiscal thrust 1974 when II published 1 the "fiscal thrust" in 1974 published my first annual estimates estimates in in ~ The Federal Budget: Its Impact on the Economy, The Conference Board, New York, 1975 edition, p. 12. York, 1974, fiscal 1975 12. My quarterly estimates were published in 1976 1976 (op. cit., fiscal 1977 first quarterly 1977 edition, p. 11). 11). The measure course7d~FT~ed rom measure itself is, of of course, derived ffrom Keynesian macroeconomic analysis. analysis. Previous uses uses of similar similar measures may be found in vlilliam William H. Oakland, "Budgetary ‘Budgetary Measures of of Fiscal PerPerEconomic Journa 1 (April 1969), 348-58; E. formance," formance,” Southern Economic_Journal 1969), pp. 348—58; Gerald Gera 1d Corrigan, Corrigan, “The "The Measure Measure and Importance of of Fiscal Fi sea 1 Policy Po 1icy Change,” Change," Federal Reserve Bank Bank of New York ~p~jjjyReview Monthly Review (June 1970), pp. pp. 135—45; 135-45; “Federal Budget Discipline and National Priorities Paul W. McCracken, ''Federal of the the 1970s,” in Michael E. Levy, çonpmic Issues of 1970s," in Mi chae 1 E. Levy, editor, editor, ~Ma ior Economic Issues of of the the l970s, The Conference Board, 1970s, Board, New New York, 1973, 1973, esp. esp. p. p. 9. 19National-imcome-accounts (NIA) National-income-accounts budget data were used; induced expenditures (mainly regular unemployment compensation) are excluded; and long—lead long-lead defense expenditures are are adjusted from their “delivery "delivery basis” basis" to aa timing that reflects more more closely closely actual production. -176- Table Table 2 ·auarterly "Fiscal Thrust” Thrust" and Its Major Quarterly and Annual Estimates (Preliminary) of “Fiscal 1959-1980 1 Components, Fiscal 1959~198O’ (NIA budgetdata; budget data;$ at seasonally adjusted annual rates) (NIA $ billion at Expenditure Contribution’ Conmbution' ny 1959 1959 FY 3.5 111tI Iv IV —0.9 -0.9 5.3 1,6 1.6 —2.3 -2.3 —1.1 -1.1 I II 1.5 F? 1960 FY 1960 F? 1961 FY 10.2 10.2 F? 1962 1962 FY 7.1 -1-2 1.2 III Ill Iv IV I II F? FY 1963 1963 4.3 4.3 III /II I It II F? FY 1965 1965 IIII Ill III tV IV I ItII F? FY 1967 21.2 212 ( IIII 20,4 20.4 FY 1968 1968 III Ill Iv IV I IIII F? FY 1969 1969 F? FY 1970 1970 —0.3 -0.3 2,7 2.7 1,5 1.5 3.4 16.2 Ill Iv IV I IIII 2.74 15.2 15.2 2.46 2.46 0.25 0.68 0.39 0,39 1.10 0.81 -7.5 —7.5 —6,6 -6.6 1.7 -5.6 —5.6 3.0 22.0 22.0 3.6 —0.4 -0.4 3.1 -0.5 —0.5 0.94 0.62 0.82 0.89 0.30 2.0 2.0 5.8 -22 —2.2 9.6 —6,3 -6.3 —1.0 -1.0 -7.1 —7.1 ‘—0.4 -0.4 5.8 0.1 4.3 4.3 0.9 10.9 10.9 0.83 0.83 0.48 0.98 0.47 6.6 4.6 5.0 5.0 2,7 2.7 0.0 0.0 0.2 —5,5 -5.5 0.1 0,1 -14.8 —14.8 7.3 2.75 2.75 18.9 18.9 —5.2 -5.2 0.00 —0.11 -0.11 0.03 0.03 0.38 0,38 8.7 3.6 2.2 3.3 —0.5 -0.5 -0.2 —0.2 -1.9 —1.9 0.3 0.3 2.0 2.0 5.6 3.3 9.5 III Ill Iv IV I IIII 0.32 17.8 —2.3 -2.3 -0.17 —017 0.87 0.10 0.06 -0.1 —0.1 —0.7 -0.7 1.9 44 4.4 2.9 0.2 -5.0 —5.0 -02 —0.2 7.1 4.8 4.8 6.9 6.9 2,4 24 111 IV 0.86 5,5 5.5 —2.1 -2.1 0.38 0.38 0.33 0.33 -0.03 —0.03 0.07 0.07 —1.0 -1.0 54 5.4 5.2 5,2 5.8 -0.1 —0.1 0.0 1.7 1.8 5.8 5,8 3.4 7,2 7.2 3.5 0,23 0.23 0.18 018 0.98 0.98 -0.09 —0.09 0.75 0,75 15,4 15.4 3.4 19.9 FY 1966 1966 1.30 2.2 2.1 -2.7 —2.7 0.3 0,3 0.0 0.1 0.1 ,.6 4,6 5.4 0.0 —0.7 -0.7 0.2 02 2,6 2.6 0.38 0.36 0.40 0.53 0.71 0.71 1,2 1.2 1.0 5,7 5.7 -0."! —0.1 1.9 10.1 10.1 2.1 ( 7.8 0.0 0,2 0.2 —2.5 -2.5 -0.1 —0.1 —1.0 -1.0 5.3 5.3 0.6 0.6 0.4 OA lIt Ill IV 200 2,00 1.7 17 2.0 2.4 3.7 3,7 0.0 0.0 0.3 0.4 04 —2,4 -2.4 5.3 III 11: Iv IV I IIII 9.8 g.e 0.7 0.25 0.25 0.04 0.04 -0.10 —0.10 0.12 0.12 -2.5 —2.5 0.6 0.0 —0.3 -0.3 0.0 0,0 2.2 2.2 1.9 —0.2 -0.2 0.4 0.4 IV 0.30 1.2 -0.4 —0.4 —0.03 -0.03 0.30 0.16 016 0.37 1.69 3.7 3.9 4.0 10.4 10.4 -177—177— Asa •Jo of GNP Asa%oiOWP tat’cnan ge T,u·cnange Contribution Contnbut,on {S) 151 -0.18 —0.18 1,17 1.17 0.34 0.34 -0.48 —0.48 -0.22 —0.22 i_s —0.1 -0.1 1.0 5,4 54 —0.5 -0.5 ,,, 0.74 -1.1 —1.1 —0.4 -0.4 Exoendirure Expenditure Contribution Conrnbuuon (4J 5.6 5.5 1.6 -3.7 —3.7 -1.1 —1,1 0.0 -0.6 —0.6 -2.0 —2.0 0.0 1.8 18 2.0 2,7 2.7 3,7 3.7 /II IV I (( II ,., 2.8 -2.6 —2.6 0.2 0.2 -0.5 —0.5 0.6 0.6 I IIIt .",seal Fiscal Thrust ~ (3)=/li+/2! (3)stt)+12) 0.3 0.2 -1.4 —1.4 0.0 1.2 12 II} IV F? FY 1964 1964 '" Tau’change Tax·change Contribution’ Conmbut1on' (2} (2) 0.01 0.45 0.09 1.12 1.12 0,07 0.07 0.05 -0.29 —0.29 0.00 0.00 -0.52 —0.52 0.00 0.00 -0.12 —0,12 -0.39 —0,39 0.00 0.00 -0.07 —0.07 —0,02 -0.02 0.00 —0.06 -0.06 0.00 0.13 0.00 0.00 0.00 0.00 0.05 0.05 0.08 0.08 -0.42 —0,42 0.00 0.00 0.04 0.04 -0.44 —0.44 -0.02 —0.02 1.64 0.00 0.01 0,01 0.74 0,74 0.86 0.52 -0.01 —0.01 0.00 0,26 0.26 0.27 -0.29 —0.29 0.42 0.02 0.02 -0.68 —0.68 -0.G3 —0.03 —0.30 -0.30 -0.07 —0.07 -0.02 —0.02 -0.25 —0.25 0.04 -0.63 —0.63 0,00 0.00 0,03 0.03 -0.65 —0.85 0.01 0,01 -1.64 —1.64 —0.72 -0.72 -0.11 —0.11 -0-77 —0,77 —0.06 -0.05 0.60 0.80 0.38 -0.04 —0.04 0.32 -0.05 —0.05 Fiscal Theusi --1!!!.E!_ (4/e,,(5)+!6; 0.55 0.55 1,24 1.24 0.39 0.39 -0.77 —0.77 -0.22 —0,22 -0..22 —0.22 0.25 0.25 -0.08 —0.08 -0.49 ‘—0.49 0.12 U3 1.93 0.34 0.34 0,40 0.40 0.47 0.47 0.71 0.71 1.43 0.23 0.18 018 1.03 -0.01 —0,01 0.33 0.38 0.37 -0.47 —0.47 0.05 2.50 2.50 -0.17 —017 0.88 0.84 0.92 0.84 0.64 -0.01 —0.01 —0.11 -0.11 0.29 0.29 0.65 2.46 2,46 1.25 1.25 0.50 a.so 0.30 0.30 0.44 0.44 2.44 0.87 0.60 0.64 0.34 1.83 1.83 0.25 0.71 0,71 -0.26 —0,26 1.11 -0.83 —0.83 —0.75 -0.75 0.19 0.19 -0.61 —0.61 0.32 0.32 2.29 2.29 0.39 0.41 0.4, 0.41 1.07 1.07 Table 22 (continued) Table Thrust'' and Its Major Quarterly and Annual Estimates (Preliminary) of "Fiscal “Fiscal Thrust” Components, Fiscal 1959·1980' 1959~198O’ (NIA budget budget data~ data;$ seasonally adjusted adjusted annual annual rates) rates) jNIA $ billion billion at at seasonally Expenditure Expenditure Contribution' contribution’ /1) F? FY 1971 1971 14.9 7.9 —1,7 -1.7 III Ill lv IV I IIII 23.6 III Ill IV I ItII 18.3 111 III IV I IIII 31,0 31.0 III 111 IV I II 2.3 2.3 7.0 7,5 7.5 14.2 14,2 FY 1975 1975 F? 60.0 III Ill IV 9.0 13,0 13.0 4.4 —0,4 -0.4 5.9 46.9 46,9 F? 1977 1977 FY IV I IIII III 111 IV iv I IIII III 111 eat. est. sat. est. eat. est. eat. est. 39,8 39.8 48.8 —17.6 -17.5 3,8 3.8 3.9 3.9 1,2 5.2 0,17 0,17 0.52 0.52 0.55 1.01 0.98 0.98 0,23 0.23 1.64 1.64 1.25 1.25 1,60 1.60 22.7 1.09 1.09 0,07 Q,07 0.51 0.5~ 0.88 0.88 1,55 i.55 8,8 88 4.1 4,1 2.5 7.3 55.7 0.65 0.65 0.22 0.13 0.56 0.56 1.99 15.4 15.4 —2,4 -2.4 15.8 15.8 —1.7 -1.7 0,80 0.80 0,26 0.26 0,34 0.34 0.59 0.59 21,8 21.8 6,3 6.3 12,2 12.2 —1.9 -1.9 33.6 —6.8 -8.8 9,7 9.7 —0.6 -0.6 —0,4 -0.4 0.58 0.58 0.81 0.81 0,27 0.27 —0.02 -0.02 0.34 2.54 15.9 '.5.9 5.2 5.2 12,1 12.1 20,7 20.? —4.1 -4.1 —0,3 -0.3 —02 -0.2 —4.7 -4.7 —0.1 -0.1 10,8 10.8 7.1 7,1 4,8 48 11.0 1~,0 38.9 53.9 9.8 2.28 4.12 4,12 —3,2 -3.2 4.0 2.5 2.5 3,7 3.7 17.8 17,8 6.0 8.0 14.1 14.1 33.7 2.5 7.5 3.1 14.4 14,4 -0.4~ —0.41 1.75 1.75 0,12 0.12 0.02 Q.Q2 142 14.2 0.6 22,1 22.1 21.7 —0.6 -0.6 4,3 4.3 —0.7 -0.7 —9,3 -9.3 45.9 IV I IIII Ill pre!. prel. 27.5 —30.6 -30.6 12,9 12.9 •.4 4,4 2.7 12.0 12.0 F? 1979 FY 1.48 58.6 Sa.6 7.0 32-0 32.0 IV I IIII III Ill F? 1980 FY 1980 12.9 12.9 0.13 0.13 0,34 0.34 1.04 1.04 0.59 0.59 -3.0 —3.0 21.8 -5.9 —5,9 0.6 0,2 0.2 —2.8 -2.8 —1.8 -i.8 3.0 3.0 19.1 1.2 9.6 9.6 17,0 17.0 F? 1978 FY 1978 0.2 0.2 0,5 0.5 —4.4 -4.<: 0,2 0.2 —1.4 -1.4 26.0 III Ill IV I IIII Ill Ill TO. T.Q. 2.12 2.12 13.5 1,36 1.36 2,0 2.0 16,8 16.8 4.2 4.2 10.6 10.6 As 3% ofCNP Asa¼ofGNP rxu-cnange Tax•cnange Contribution Conlribulion (Si /Si 0.77 0.77 —0.17 -0.17 0.36 0.65 0,60 0.60 4,0 4.0 1.4 1.4 3,7 3.7 9.4 1.9 1,9 0.4 04 —7.5 -7.5 0.4 04 14.0 14.0 3.4 3.4 23.9 18,7 18.7 II 18.5 —8.0 -8.0 2,6 2.6 —3.5 -3.5 (4) !') 1.46 1.46 2.6 2.6 —4.8 -4.8 Expenditure bpena,ture Contribution Contribution 5,0 5.0 3,3 3.3 8.5 6,0 6.0 —2-3 -2.3 -4.9 —4.9 21.4 1.6 0,2 0.2 F? 1974 1974 FY 22.8 22.8 —5.1 -5.1 1,4 1.4 3.7 3.7 11.7 11.7 6,8 6.8 F? FY 1973 FisCa) Flsc.ii Thrust ~ )3)s)t)+f2) i3)=f1i.,.i2) 6,7 6.7 —0,3 -0.3 1.8 1.8 —0,3 -0.3 3.6 3,6 6,7 6.7 6-3 6.3 F? FY 1972 1972 F? 1976 FY 1976 Tax-change Tax•change Conmtn//ion' Contnbulion’ (2) (2; 0.45 0,29 0.29 0,19 0.19 0.43 Fiscal Fiscal Thrust ~ (4)e(EJ+181 !4i,,,(S)+/6! 2.23 0.67 0.50 0.33 0.33 0.82 0.82 0.57 0.57 —0.03 -0.03 0.17 —0.03 -0.03 —0.46 -0.46 0,24 024 —0.21 -0.21 —0.71 -0.71 0,22 0.22 —0.39 -0.39 0.16 0.16 0,04 0.04 —0.59 -0.59 0,03 0.03 —0.28 -0.26 0.02 0.02 0,03 0.03 —0.32 -0.32 0.02 o.oz -0.10 —0,10 0.01 0.0. —0.19 -0.19 —0.12 -0.12 0.20 0.20 0,79 0.79 2.54 2.54 —1.91 -1.9" —0.04 -0.04 0.25 —0,04 -0.04 0.38 —0.18 -0.18 0.22 0.13 0.13 0,19 0.19 —0.45 -0.45 —0.20 -0.20 —0.02 -0.02 —0.01 -0.01 —022 -0.22 0,42 0.42 —0,11 -0.11 0.69 —0,07 -0.07 —0.08 -0.08 0-00 0.00 —0,37 -0.37 0,40 0.40 —0.02 -0.02 —0,02 -0.02 1.66 1.66 0.37 0.37 0.13 0.13 0.33 0.33 0.81 0.81 1.08 -0.25 —0,25 1,79 1.79 —0.47 -0.47 0.05 2.02 202 019 0.19 0.55 0.55 0,23 0.23 1.03 1.03 4.02 0.99 0.04 1.52 1.52 1.45 1.45 2.39 2.39 3.12 3.12 —1,10 -1.10 0.23 0,23 0.23 0,30 0.30 2.92 2.92 0,91 0.91 0.29 0,64 0.64 1.07 1.07 1,10 1.10 0,45 0.45 0.20 0,12 0.12 0.34 2.41 2.41 0.69 0,95 0.95 0.27 0.27 0.51 0.51 1,36 1.36 0.08 0.69 0,17 O.H 0,41 0.41 ‘Author’s derived Itory pubiianec and ‘01 'Author·s preuntinary prel1m,nary estimates esl!mates derived trom the the best available avsHable publ,sned and unpubt~uhedsources. unput,i she-c sovces. Dale Data revisions rev,s1ons and and retinement, refinements hare have not tmen cornpleteo. yet Deen complelec. ‘Increases I( —- icr i—I ln ents mdv0= subltaclions ci cvangea i,i,i ‘‘regular” 'Increases 1or reductions reductions 1-) !n “adlusled” "adJusted" NIA NIA budget expenditures. expenditucss. Adiuslm Ad1ustments 1nciuoe subtract,ons of changes "regular" unempicyunemploysent benell!s benetitsand “delense timely ment and oI of toe tne 14IA N!A "defense timely adiuslment adjustment" 'Initial (-) or reductions I( - is l ,n tax tax revenues resu!t,,.,g from sttuclural stre.Jctural Cflaii~e5 cnan~es it1n tao tax bases bases or on best best ouDIlshea ouol!shea and un• ‘Initial increases I—Icr resultrrgirorn cv rates. baseo bssec Ott anus,gublisneo estimates tom the Treasury Timing ci ncreases in inthe ;:,uo!isheo estimates from Treasury Department and and she the Bureau ot of Economic Ecoriomic Anaiysis. Analysis. T,ming oi the ettect eftect nt ,:,f the mcreases the tan tax base base on tile employee's emoioyee’e pan contrioutions to social secunty secunty has beet, chat ged by by author ccncenl,ale this increase mainly on the part ol of contnoutions to soc,al has been changed author to to concent,ate this ,ricrease ma,rHy it1n the the lastiwo last two calet’ ca!en• darquartars darquanem. TO —Transitional T.Q.Trans1t1onal quartet quarter Sources; Bureau oI Economic Analysis; The Conference Conlereece Board. Sources: of Econom,c -178- to permit permit historical comparisons comparisons and minimize inflation—induced inflation-induced order to 20 distortions of these measures distortions measures2° In short, fiscal thrust and its its components are designed designed to measure the initial initial expansionary impact originating originating from the federal budget to which the traditional traditional Keynesian multipliers multipliers could could be be applied aoplied (or which could trigger trigger fiscal fiscal simulations simulations in in econometric models.) models.) What concerns us for the present not so much the present analysis analysis are not quarterly, or even the annual, levels or or changes in fiscal thrust, but but rather the average degree of stimulation of of the budget over the broad broad longer time periods distinguished here. The suirmiarized in The results, results, summarized Table 1, confirm the proposition that, on balance, the budgets of of the 1970s were more expansionary expansionary than than those of the 1960s, largely as the result of much much faster spending growth. Fiscal 1961Fiscal thrust averaged 1.4 1.4 percent percent of of GNP during fiscal 1961- 1965, compared with a 2.0 percent average average for fiscal 1966-1979. Within the latter latter period, average average fiscal thrust rose from 1.5 l .5 percent of GNP during fiscal 1966-1969 1966-1969 to 1.8 1.8 percent during fiscal 1970—1973 1970-1973 and 2.5 percent 1974-1979. percent during fiscal 1974-1979. nant nant throughout. The expenditure expenditure component was domidomi- But tax cuts cuts provided significant stimulation during the period of of the “New New Economics; Economics tax increases provided belated and 11 11 ; restraint during the escalation phase of the Vietnam 7limited imited .r_estr_ii_i.'1.!. Vietnam War (par(parguns and Great Society” spending); tially off—setting off-setting the "guns Great Society" spending); and tax changes were nearly neutral over the course of of the l970s. 1970s. 20 be constructed (analogous to ZOAA “weighted ''weighted fiscal fiscal thrust thrust'' could could be constructed (analogous to the "weighted “weighted full-employment full-employ:;1ent budget surplus’), surplus"), but but the complications created by such such aa refinement are hardly hardly warranted in the light of of the use of any simple overall overall measures measures of fiscal impact impact and the crudeness crudeness o~the O' the basic estimates. —179— -179- THE NEW "SOCIAL REGULATION" “SOCIAL REGULATION’ Changes Changes in the composition and growth of of the federal budget and its components were not the only inflationary manifestations of what has been been termed here aa new “social social activism.” activism. 11 11 The same emphasis on social welfare and on the the consumer, rather than than on on real real growth and and the producer, gave regulation” in the mid— gave rise to to aa new wave •.,;ave of “social social regulation mid21 The 1960s and the early 1970s. l970s.21 1960s The impetus came from consumer consumer groups, groups, 11 11 environmentalists, labor unions, civil civil rights advocates and diverse diverse public public interest groups, who felt that the traditional traditional regulatory agencies were not achieving “social social goals,” goals, such as as product safety, safety, 11 11 clean air afr and water, equal equal employment opportunities, safer and healthhealthier working conditions. In response to these public pressures, twenty “social regulatwenty new "social regulation’ tion" agencies have been created since 1970. Among these, these, the most important important ones are the Consumer Consumer Product Safety Commission, the EnvironEnviron- mental Protection Agency, Agency, the Equal Employment Opportunity Commission, and the Occupational Safety and Health Administration. These These new agencies charged with social regulation regulation were among the most prominent prominent “growth l970s; their full—time industries" of the 1970s; full-time staff staff increased from "growth industries” 17,324 in fiscal 1970 to 17,324 to 69,258 69,258 in fiscal 1979 (86 percent of the regulatOry staff). federal governments government's total total regulatory The The administrative administrative and 21 21 For “social regulaFor further discussion of the evolution of of new "social regulation” tion11 and some cost estimates, see Michael E. E. Levy, Levy, assisted by Delos R. Smith and Steven Malin, The Federal Budget: Budget: Its Impact on the 1980 No. No. 2, pp. pp. 12-14. For For an an encompassing encompassing critical critical Economy, fiscal 1980 of the impact of government regulation, see Murray L. Weidenbaum, review of Business, Prentice-Hall, Inc., Englewood Englewood Business. Government, and the Public, Public. Prentice-Hall, 1977; also Murray Cliffs, N.J., 1977; Murray L. L. Weidenbaum, The Impacts of GovernGovernment ment Regulation, Working Paper Paper No. 32, 32, Center Center for the Study of of American Business, Business, Washington University, St. Louis, July 1978. -180-180- reporting costs imposed on businesses grew grew accordingly. More Important, important, business had to divert large and increasing increasing amounts of cash cash flow and capital into investments designed designed mainly to achieve achieve compliance with with new social regulation. A major part of of these investments -- regardless A major —— 11 of whatever their social social benefits -- was “unproductive” unproductive in terms of of our our 11 -— traditional measures of real output and productivity. In fact, accordaccord- estimates, productivity of the nonresidential ing to to the best best available estimates, nonresidential it would would business sector was 1.4 percentage points lower in 1975 than it have been under the regulatory conditions conditions of 1967.22 1967. 22 The tendency of the new “social "social activism” activism" to pursue socially dedeThe sirable goals without any proper regard for economic economic implications, implications, without due consideration consideration of benefit-cost has been been benefit—cost relationships, also has felt in in the regulatory area. Excessively short deadlines for meeting regulatory standards, detailed detailed prescriptions of specific specific technological solutions, absolute prohibition of the use of certain certain substances or processes have often raised marginal compliance compliance costs well well in excess of 23 marginal benefits. 23 Consequently, the new social social regulation —— regardless of whatever its its social social merits -- has has been highly inflationinflation—— ary. In its 1979 1979 Annual Report, the Council of of Economic Advisers 22 . ,, 22 Edward Denison, "Effects InstituEffects of Selected Changes in the Institutional and Human Environment upon Output per Unit of Input,” Input," Survey of Current Business. Business, January 1978, pp. 21—44. 21-44. 23 For For aa discussion of these problems, problems, see the section on "Regu“Regulatory Reform" Reform” in in the 1978 Annual Report of of the Council of Economic Economic 206—216); also the section Advisers (pp. (pp. 206-216); section on “Regulatory "Regulatory Policy” Policy" in the ~1979 Annual Report of the CEA (pp. 85-91). -181- described the dynamics of the inflationary process process induced by the new social regulation in in the following way: actions enter Once incurred, the costs of regulatory actions into wage— and price—setting into the the wageprice-setting mechanisms of the economy. economy. Most costs of of regulatory action show Most of of the the costs regulatory action show up up not not as as governmental budget expenditures, but as as increased increased costs to industry. relative to and industry. Acceptance Acceptance of of higher higher prices prices relative to wages wages and other money incomes is the way in which which society pays pays for the benefits social regulation. however, our ecobenefits of of social regulation. In In fact, fact, however, our economic institutions and and measures of of prices do not distinguish between between these these sources of of price increases increases and others. others. IndiIndividuals and groups try to to escape escape paying the costs of of reguregulation by by increasing wages and other forms of income to match match the the higher higher prices. prices. The The result is an additional round of increases. But regulation cannot of price price increases. But the the costs costs of of regulation cannot be be avoided, and widespread attempts to do so add to asd widespread attempts to do so simply 24 inflation.24 inflation. SOME SOME LESSONS FOR THE 1980s My journey along the inflation road road of the last last decade—and-a—half decade-and-a-half has ended with aa thesis, rather than with solid solid conclusions. The search for an explanation of of the largely unexplained aspects of of our ininflation (or of the "excessive" you will) -- its its duradura“excessive” money growth, if you -- tion, persistence persistence and steady escalation -- uncovered basic changes in in -— social social and political orientation and in our public policy. These These changes -- II referred to them as aa new social social activism -- originated originated in —- —— the mid-1960s mid-1960s and gained momentum momentum in the 1970s. 1970s. This social activism activism manifested itself in in increased increased reliance reliance on on the federal federal government government to achieve socially desirable goals through new, new, or enlarged, budgetary budgetary and regulatory regulatory programs. The consumer and and “social nsocial benefits” benefits 11 were stressed, stressed, often at the expense of of higher costs, slower real growth and lower productivity gains. lower productivity gains. 24 24 Among consumers -- many Among consumers many of of whom whom are, are, after after —— Op. cit., p. 87. op. -182-182- all, producers as well -- these new social benefits were often to be —- focused on the nonproducers (who tend tend to be perceived perceived as as nmore more needy' needy’ 1 11 s oci al benefits). benefi tstl). and, hence, more deserving of “social The ‘costs” costs of this new new social activism included increased increased disdis11 11 incentives to work and to to invest, slower growth of real GNP, GNP, amd and lower lower productivity gains. AA main result was was aa persistent persistent increase increase in’infla— in-infla- tionary pressures of our entire economic system. system. If this thesis has any merit, if it it contributes in any signifisignifi- cant way to the explanation of the ongoing U.S. inflation, inflation, the impli— implications catioms are clear: inflation control depends depends on removal of of Successful inflation the fundamental causes of U.S. U.S. inflation. inflation. Fiscal and monetary monetary policy policy Fiscal restraint, restraint, while while necessary, will not be sufficient. sufficient. New policies to encourage encourage greater productive productive efforts efforts and faster real growth will vii 11 be essential, if price stability stability is to be be restored in in the 1980s. -183- DISCUSSION OF THE LEVY AND MELTZER MELTZER PAPERS Poo·1e William Poole Michael Levy has has assumed the task task of of explaining explaining the persistence persistence of inflation. of confess, though, that II got off to to aa bad start at the II confess, very beginning of of his paper. His second sentence sentence reads: reads: “Monetarist "Monetarist explanations deep—seated inflation provide no insights as explanations of this deep-seated as to to its economic, economic, social, social , and political pol iti cal causes’ causes" (emphasis added). sentence towards the end of of his summary reads: And And aa “Fiscal and monetary "Fiscal monetary policy restraint will be necessary, but may not be be sufficient [to control contra 1 inflation].’ i nfl a ti on J." 1lieve i eve these claims. Fortunately, however, Levy Levy does not really bebeOn page two of his paper paper he says says that, “on "on aa anpurely technical level, the monetarists have, of course, all the answers. In fact, some of my own econometric econometric exercises have have tended to reconfirm reconfirm their valuable, if if somewhat simplistic, generalizations.” generalizations." If If we strip away the loaded words words such such as as “simplistic, "simplistic," then it clear that Levy accepts the basic argument that inflation inflation cannot cannot is clear occur in the absence of of excessive money money growth. Accepting this propopropo- sition, inflasition, Levy surely does not believe that successful successful control of of inflation would be possible possible without slowing money growth. Indeed, II cannot cannot believe believe that Levy would would claim claim that slowing money growth would would fail to reduce inflation. He simply does not in fact fact believe that monetarist explanations provide no insight into the economics economics of of inflation. Dr. Poole is Professor Professor of Economics at Brown Brown University. -184- Levy s paper is not about monetarist monetarist propositions linking money Levy’s 1 growth to inflation, excessive money growth. inflation, but about the causes of excessive This obviously important. This issue is obviously But the reasons reasons monetarists have not paid much attention attention to to this issue to date ~re, are, first that that it it was important to is indeed to gain gain agreement that inflation is indeed aa monetary phenomphenom- enon -- aa proposition not widely accepted thirty years ago -- and, secsec-— —— ond, that the methods of economic analysis may not provide great ininsight into the causes of of excessive excessive money growth. growth. Levy feels that the important issues issues concern and social important concern changes changes in in the the economic economic and social structure structure that have produced produced an an inflationary environment. let me me introduce introduce aa qualification to the Before commenting further let simple nionetarist monetarist view. Clearly, insofar as changes in the economic and social structure, in the the average tax rate, and and in the regulatory burden affect incentives and and productivity, productivity, the rate of productivity productivity growth nay may slow slow down. Reduction in the growth of real output, output, given the rate of money growth, will raise the the rate inflation. the rate of money growth, will raise rate of of inflation. As As aa first first approximation, what matters matters is the money money stock per unit of real GNP. But the slow-down in productivity productivity growth and therefore in in output growth can directly account account for only aa very small small part part of our inflation. We might be able to explain one to two percentage points of of the inflation in recent years by by the the slow-down in output growth. what all the the shouting is about. But that that is not If the current current rate rate of of inflation were only one or two percentage points above the rate of the early early sixties, sixties, then the subject subject of of this this conference would not be be inflation inflation but rather productivity or or some other other issue. If II understand Levy correctly, he he feels that sociological sociological and non-monetary factors have accelerated the rate of inflation inflation and and that -185- the monetary authorities have been dragged along -- forced to accoswnoaccommo-— infladate with money growth the more fundamental factors producing inflation. Even on on this view, however, however, Levy should be much much more interested Even than he is in what he calls the simplistic monetarist explanation. If the price level were very closely linked to to the money money stock, stock, with with a very small margin margin of of error, then it it would would be be absolutely absolutely clear clear that non-monetary non-monetary factors could work to to increase increase inflation only insofar as they operated operated quite directly directly on the Federal Reserve. The greater greater the amount of slack or imprecision in the money/price relation, the more credible credible Levy’s Levy's argument argument becomes. becomes. If If the relation relation is imprecise, in the the short run, there is much much room for non—monetary non-monetary facfacespecially in tors to produce an acceleration in the rate rate of inflation directly, directly, and and for the Federal Federal Reserve to be drawn into into monetary expansion expansion later by pressures to to sustain the ongoing inflation process. process. The very word "ac“ac- commodation" to an inflainflacommodation” has the flavor of the central bank responding to tion that that has already occurred occurred in order to prevent longer—run longer-run forces from reversing the inflation through aa process process involving unemployment. showing changes in defense spending, governgovernLevy presents data showing ment transfers, transfers, and so so forth. But But he presents presents no evidence whatsoever such as these that even bears on the validity of his claim that factors such are responsible for the inflation. Surely time series series evidence evidence on United States inflation relative to government spending spending would be relerele- vant. Also, cross cross section section evidence evidence relating the rate rate of inflation to the size size of the government budget or its its rate of growth in different countries would be relevant. Does Levy dispute the common finding that inflation follows rather than leads money growth, aa finding that seems inconsistent with the accommodation argument? -186- I'm forced to make a few Since Levy has presented no evidence, I’m comments based on casual casual empiricism empiricism and aa priori plausibility. plausibility. cormients One of Levy’s Levy's claims is that the erosion in in the growth of real incomes has has led led to seek higher wages and prices in an attempt to reworkers and firms to relost income growth. coup their lost If this this argument is true, why why did wages and prices fall sharply sharply as people became poorer between between 1929 and and 1933? Is short—run Phillips curve -- which shows that wage inflation Is the short-run inflation —— slows as people become poorer through unemployment -- consistent with -— Levy's proposition? Levy’s If growth in taxes has been aa major major factor in rere- ducing growth in disposable income, then why have we not seen more more acac- spending and and taxes rather than the activity tivity to reduce government spending claimed claimed by by Levy to raise nominal nominal wages and and prices? II may be be wrong, but but Proposition at the federal level seem awfully weak to to me me at level, Proposition 13 pressures seem and in in any event event seem to be a lagged lagged result of of the inflation inflation process rather than part of aa process that can explain explain the inflation. What What other other evidence beside nominal wage and price increases can we look at? What about strike activity, or union membership, or concon- centration centration ratios in industry? industry? All All of of these would would seem to to have some some possible possible connection to inflation, or at least as symptoms of of the procprocess ess Levy is talking about. My impression is that these factors all cut in the wrong direction direction in the United States. Most fundamentally, how can real factors, other other than through productivity effects and effects on Federal Reserve behavior, have anything to do with nominal magnimagnitudes? Levy seems to recognize recognize the importance of of explaining Federal Federal seventeen deals at some length Reserve behavior; his footnote on page seventeen -187- with the question of the role of budget deficits in explaining Federal Reserve money creation. I believe that aa number of factors, some some of which which are closely connected to the ones ones Levy has emphasized, emphasized, should be examined in terms connected of their effect effect on Federal Federal Reserve behavior. My list of important important items is this: l. 1. Since the mid—sixties mid-sixties there have have been consistent underestimates underestimates of of the natural rate of unemployment by by the Federal Reserve and by by the economics economics profession. These These underestimates underestimates have led led to money money growth that on on average has been been too too high, even accepting accepting the the view that monetary policy should aim for an unemployment rate close to the natural natural rate. 2. There There has been aa great over-emphasis on nominal interest interest rates and doesn’t really matter. aa view that short-run money money growth doesn't Although Although the Federal Reserve has has long recognized the importance of long-run long—run money growth, operating in growth, it it always seems to be operating in aa series of short short runs that never never add up up to aa long run. 3. The Federal Federal Reserve has has from time to time made political miscalcumiscalcuThe lations based on a combination of overly optimistic forecasts of of the effectiveness of fiscal actions and overly optimistic optimistic forecasts of when fiscal actions would occur. Probably the the best example of this point is the Fed’s Fed's delay in in tightening money in in 1967 1967 while while this pass aa tax increase. waiting for Congress to pass 4. The Federal Reserve’s short. Reserve's policy policy horizon horizon has been been too short. OrdinariOrdinari- ly, ly, the Fed Fed looks looks ahead long long enough to see significant significant impacts of monetary policy on employment and and output output but but not not long enough to see any important impact on prices. -188- 5. The Federal Reserve has used aa poor control control mechanism based on the federal funds rate. has produced produced aa procyclical procycl ical monmonThis mechanism has etary policy because it it makes persistent procyclical mistakes so easy. 6. 6. The Federal Reserve is obviously obviously responsive responsive to political po 1it i ca 1 pressures, especially from the administration. administration. These political political considerations considerations nay may have reflected concern, from time to time, over over reelection reelection of of aa President and over over reappointment President and reappointment of of aa Federal Federal Reserve Reserve Board Board ChairChairman. JI continue continue to to believe believe that Federal Reserve behavior behavior is is not not at all aa simple function of broad societal trends. Accidents of hishisAccidents tory such as as assassinations happen and tory such assassinations do do happen and are are important. important. While While II certainly rule out. out~the importance of research on general certainly would not rule principles of political behavior, II still still feel feel that neglecting the interplay of personalities personalities and events events is aa mistake. In an endeavor of this type, traditional traditional historical analysis can provide very very subsubstantial stantial insights. In summary, II believe that Levy interpretaLevy provides aa misleading interpretation tion of what monetarism monetarism is all about. Monetarism involves the economeconom- ics ics of of the relations between money, output, prices, and interest rates, and and the economic processes processes responsible for these relations. It It does not not pretend pretend to to offer offer an economic economic explanation explanation of of money money growth growth and should should not, therefore, be criticized criticized for not doing so. Now that monetarist monetarist propositions propositions -- at at least in in their their long—run long-run -— form -- are so so widely accepted, it it clearly makes sense to move move on to —— issues concerning why why the the monetary monetary authorities authorities behave behave the way way they they do. do. Levy has offered aa number of interesting interesting hypotheses on on this this question, but has not provided any evidence. To To my my taste, taste, his approach approach is is less -189-189- productive than it might be because he pays so little little attention to to the monetary authority authority itself. itself. Surely the Federal Reserve Reserve should be be the focal point of of the political and sociological analysis. caretaker function. functinn. far more than a caretaker The Fed Fed has has If If the factors Levy discusses are important, important, we need to to know how they impinge impinge on on the Federal Reserve in order to have much much confidence confidence in in the the argument. Now let to the Allan Meltzer. Meltzer. Now let ne me turn turn to the paper paper by by Allan II will will start start with with an outline of his his argument argument as II understand it. it. First, affect output. Fi rs t, Meltzer Meltzer believes believes that that expectational expectational errors errors affect output. The The expectational errors err"'ors that he stressea stressei are are those between the normal normal, or permanent plus plus the transitory components. components. or permanent He mentions in passing He that this of Lucas. this view is different different from that of While it is true that Lucas Lucas uses a spatial rather .-a_ther than aa temporal model, II think that it it really comes cones to much the sane same thing. Additional output can be be obtained obtained in the Lucas model rnode1 only if labor is willing to substitute hours inter— inter- temporarily. In any event, the Meltzer Meltzer view is that when prices are viewed as temporarily high high the level of output is expanded, expanded, and when when prices are temporarily low the level of output is contracted. Actually, it it is probably probably better better for me to state Meltzer’s Meltzer's proposition aa bit differentdifferently: the permanent level of prices depends depends on the permanent level of the money stock and and it is is deviations the actual from the money stock deviations of of the actual money money stock stock from the level that are most clearly clearly related to deviations of of output permanent level from normal full employment output. Since deviations of output from potential output are related to errors, it is important to investigate investigate the formation of of expectational errors, these expectations. To To illustrate the basic idea, Meltzer uses aa -190- simple model model from the statistics literature in which which aa time series has has known properties consisting of permanent permanent and transitory variations. inference problem is to use the past data to make the best guess as The inference component in the next period. period. to the the permanent permanent component to to the The solution to problem requires knowledge knowledge of the permanent permanent and transitory variances. Given that information, information, the next—period next-period forecast depends depends on on aa distribdistributed lag of the past observations of of the series, series, with with the distributed distributed lag weights depending on the pernanent permanent and transitory variances. variances. lag This be generalized generalized easily -- although the technical problems basic idea can be -— may not be solved easily easily -- by considering more complicated time series series -- models including multivariate frameworks. frameworks. However, However, the basic basic idea idea comes through quite clearly in the univariate model analyzed analyzed by Meltzer. His tables 11 and and 22 provide the flavor of how the means His and permanent and and transitory variances might might be be extracted extracted from the data for different periods. periods. Now let me make make an an important distinction that does not seem seem very clear clear in in Meltzer’s Meltzer's paper. When we we examine aa policy of gradualism there are two analytically considerations. analytically distinct distinct considerations. One concerns the time series of agents’ agents' forecasts of permanent permanent values and the magnitudes of expectational expectational errors under the the assumption that agents’ agents' estimates of the permanent permanent and transitory variances remain fixed and given an assumed money growth path. Here, Here, it it is clear that if money growth slows sharpsharp- ly, ly, then then the market will interpret the initial initial slow-down slow-down as being largely transitory; if the slow-down is is in in fact permanent, then there will be aa large and persistent persistent expectational error. sumptions, the case for gradualism is compelling. -191- Under these asasOnly with with aa gradual gradual decline in money growth would It it be possible to avoid large expectaexpectational errors and the accompanying losses in in output. An entirely entirely separate issue -- and one that II think is is at at the -- heart of the problem -- concerns the way in in which which agents agents change change their their -— estimates estimates of the permanent and and transitory variances variances over time. Meltzer’s Meltzer's discussion is much less helpful on on this this issue. issue. If the Federal Federal Reserve could convince convince agents that the money money growth process had changed and could convince agents that it would slow money growth sharply, then then forecasts of the permanent permanent money stock would not be dedeon past observations. observations. termined by the old distributed lag on Under these these slow money growth abruptly without producing assumptions, the Fed Fed could slow producing expectational errors and there there would be no no case case for gradualism. gradualism. Meltzer has not offered offered any formal analysis of how how agents learn learn permanent and transitransifrom experience to change their estimates of the permanent tory variances. Nor has Meltzer offered an analysis of of how agents might might be led to change their estimates of these variances by the Fed introducing introducing aa new policy, policy, aa process which would would not require any learning from past money stock observations at all. My My comment on on this this to reflect reflect aa criticism criticism of Meltzer’s Meltzer's paper; paper; II do do not point is not meant to know know of any interesting models of learning and II do not have the fogfoggiest idea of how to to go about modeling this process. process. My point is simsim- ply that it it is important important to separate the issue of of calculating permanent permanent values qj~y~p of how agents given estimates of the variances from the issue of form new estimates estimates of these variances variances over over time. The only constructive thought II can offer is that prescriptions the best path for the money stock in the future might might be be based in as to the part part on on an analysis of of the effects effects of of reducing transitory variance. variance. -192— -192- Money growth has been high high in in the recent past; if the actual rate of Money money growth is brought down only slowly from this this high initial starting point and if is compressed by making if the transitory variance is this slow—down smooth and in this slow-down in accordance with announced intentions, then it is possible that the initial effects would actually actually be to to raise rctise agents'estimates of the permanent rate of money growth for the next few agents’estimates of periods. This result would occur This occur if aa significant significant part of the the recent high money growth had been been regarded by by agents agents as transitory and and theretherefore had not been been built into their estimates estimates of the permanent part of money growth. The likelihood of the perverse result could be inVestiinvesti- on money gated by by examining examining the effect of aa reduced reduced transitory transitory variance variance on growth growth expectations expectations for next year in a time series model applied applied to to money growth over the past past few years. actual money actual II have two final comments. First, First, as as John Taylor has emphasized, there is considerable uncertainty about the relative relative validity of of purely expectational expectational theories of the business cycle and and theories theories that that stress lagged adjustment due to contracts and similar similar types of institutions. institutions. As Meltzer has noted but not emphasized, the case for gradual reduction As of of money growth is considerably strengthened by this uncertainty uncertainty bebe- cause insofar as the the contract contract view has validity, aa sharp reduction in money money growth -- even if if fully anticipated anticipated -- would would produce produce a sharp dede—- -- cline in output. Secondly, although although we have concentrated on economic economic factors, II think it is worth mentioning political processes. It It is is not obvious obvious to me that maintainance maintainance over aa long period of time of aa gradual reduction of the money money stock is politically politically feasible. It is certainly It certainly conceivconceiv- able able that aa quick quick and and dirty reduction of of money growth, accepting accepting the -193-193— output effects that would occur, is the only solution that is severe output politically politically viable. II am not not sure whether whether or or not II believe that a quick purging of inflation inflation would be better politically, politically, and even if I did know what II believe II would not have have any idea of why II believed believed it. it. Nevertheless, this issue is surely important for aa full policy analysis of winding winding down inflation. inflation. An economic analysis of the minimum cost method of reducing inflation is obviously obviously important, but unfortunately cost—benefit calculation that it is not at at all clear that the cost-benefit that governs governs the political process is is very closely connected connected to the economic economic costs and benefits, however firmly we we may may establish them. -194- DISCUSSION OF OF THE LEVY AND AND MELTZER MELTZER PAPERS Albert E. Burger What did the experience of of the last last half of of the the l96Os 1960s and and the the l970s teach us about the effects of monetary decade of the 1970s monetary policy policy actions? It did not teach us anything "new." new.’ It only gave It only gave us us another another long-standing proposition proposition set of empirical observations to support the long—standing 11 generate an acceleraexcessive growth of money wi that aa maintained excessive will generate an acceleration tion in inflation and and will will raise raise inflationary expectations. The policy price stability stability in actions that engineered the move from from price in the the first first half half of the 1960s to aa 66 percent rate rate of maintained inflation by 1973 were were of l96Os to an accelerated rate of purchase of government securities by the Federal Federal Reserve which resulted in in aa faster growth of monetary base and bank reserves and, hence, aa rise in the trend growth of money from 1—2 l-2 perpercent cent to to 66 percent. nid-l96Os there already existed Prior to the mid-l960s existed aa very very large amount of evidence evidence that this would be the expected expected result of of these types of of policy actions. Indeed, one does not have to use highly sophisticated sophisticated methods of analysis to come to this this conclusion. conclusion. Simply Simply aa close look at the the data data should be be enough to convince most most people of of this strong rere- between the growth of money and inflation. lationship between The experience since 1973 1973 has reminded us us that price theory can be useful in analyzing macroeconomic macroeconomic developments. Severe supply Albert E. Burger is Assistant Vice-President Vice—President and Economist, Economist, Federal Federal Reserve Bank of St. Louis. —195— -195- shocks raise the level of prices and, hence, hence, contribute contribute to the measured rate rate of inflation. 1975—1976 illustrate, these effects However, as as 1975-1976 effects do not result in sustained inflation. Both Mel Meltzer and Levy point point out that sustained sustained inflation inflation is a Both tzer and phenomenon. monetary phenomenon. They differ with with respect to whether monetary monetary actions are the “fundamental” "fundamental" cause of iinflation. nfl at ion. Meltzer puts the blame of the blame for inflation and its acceleration acceleration directly directly at the door of Federal Reserve. Reserve. He rejects the assertion that the Vietnam War, deVietnam War, de- ficits, and government spending of the mid—l96Os mid-1960s were were the origin origin of of ininflation or were the motivating force causing the Fed to to expand expand money. II agree with Meltzer that the Fed Fed must accept the blame for starting and maintaining inflation. The The money stock grew at steadily more rapid rates because the Fed allowed allowed it to do so by providing providing the necessary bank reserves. If the Fed had not supplied more reserves, money growth would would not have accelerated accelerated and, hence, inflation inflation would would not have have accelaccel- erated. The Fed can make excuses excuses about why why it followed such a policy, but the fact remains that it did follow such such aa policy. of why policy moved from one one Levy raises the interesting question of inthat underwrote price stability stability to one that underwrote underwrote accelerating accelerating inflation. His His conclusion is that, in in the mid—l96Os, mid-l960s, there were were major political and social changes changes that led led to greater social social activism on on the part of the government (such as aa shift toward increased “nonproductive” unonproductiven transfer payments payments and and regulation) that that reduced productivity productivity and and set off the inflationary spiral. I would interpret interpret his conclusion as as meaning the Federal Federal Reserve was caught up in in this process and and essenessentially pulled pulled along the path path it followed by forces over which it it had no control. -196-196— idea that the There is a growing body of evidence supporting the Idea factors Levy discusses operated operated to lower potential real real output output growth. However, if moneif these factors had not been been accompanied by aa surge in in monetary is considerable considerable doubt have had tary expansion, expansion, there there is doubt we we would would have had the the acceleration in inflation that we experienced. leaves open the question of of why, despite repeated This still This still leaves statements inflation, the Federal Reserve statements of policy intent to halt inflation, allowed its policy actions actions to feed inflation. If the Fed had actually actually planned an acceleration in in inflation, inflation, it it could not have followed aa that was better grounded in theory and supported by by empirical program that evidence. I have difficulty difficulty accepting the explanation that the Federal Federal was simply pulled along by the tide of expansionary sentiment. Reserve was To some extent, that that may have been the case. Especially, one can point to the repeated failure of certain certain members of Congress Congress to to accept the interest rate consequences of their deficit spending. However, the basic cause of the high and rising interest rates that have charactercharacter- ized the last last 15 years has been been the the inflation generated generated by by Federal Reserve actions and the resulting rise in inflationary expectations. I would would ascribe the failure of monetary policy to achieve its obob- jective of stable overall prices to aa failure to accept and remain comcommitted mitted to a few very basic principles. These are: (1) the primary job of aa central bank is to to prevent prevent an acceleration in in the basic rate of inflation and monetary policy cannot fine tune real output; (2) excessexcess- ive money growth means an acceleration in inflation; (3) money grows at sustained, faster rate only when the central bank moneaa sustained, bank provides more more mone- tary base; (4) if there is aa surge in government demand for credit or private demands for credit surge in private demands for credit or or aa surge in measured measured inflation, inflation, short-term short-term —197— -197- interest rates will rise and Federal Reserve attempts to prevent this rise only ensure ensure that at these rise will will only that interest interest rates rates remain remain at these higher higher levels; levels; (5) the Federal Reserve can control control the trend growth growth of of money; and (6) although in theory, money growth can be controlled by operating on the federal funds rate, in in practice this is aa very unsatisfactory procedure. procedure. If the Federal Federal Reserve had remained committed committed to to these six basic principles, it policy would basic principles, it seems seems very very unlikely unlikely that that monetary monetary policy would have followed the path that characterized characterized the last 15 years. Of the above above six principles, the last •two two have been the the hardest for the Federal Reserve to accept: flaws in aa federal funds target. ability to control control money money and the More More than anything else, these these two items have contributed to the failure to achieve achieve policy objectives. objectives. Too “can the Fed Too often the question of "can Fed control money?” money?" has has gotten mixed up with the question of "should money?" up “should the Fed control money?” central If the central bank can control the growth of the monetary base, it can control control the supply of money. This should be be aa lesson that that is learned in an an introintro- ductory money and banking banking course. During the past past 15 years the Federal Reserve has tried to control the federal funds rate, not growth growth of monetary monetary base and bank bank reserves. reserves. Hence, the Federal Federal Reserve has has not “control led” money. controlled 11 11 This is is why the most important important aspect of of the policy actions actions announced by the Feds 6, 1979, announced by the Fed's Open Open Market Market Committee Committee on on October October 6, 1979, was was the part announcing aa change in operating operating procedures. procedures. Primary emphasis was shifted from the federal funds rate to growth of aa reserve aggreaggregate. gate. If the Federal Reserve remains committed committed to to this this change, monemone- tary actions actions may may start start to monetary policy. tary to match match the the intent intent of of monetary policy. 198-198- - we got into our current predica— predicaIt is much easier to analyze how we ment than it is to to state how to to get out of of it. it. Obviously, to lower the trend rate of of inflation, inflation, the growth rates of the monetary base and money money must be be reduced. However, the objective of monetary policy policy is so with not just to to slow inflation, but to do so with aa minimum loss loss of of real output. As other other papers at this conference have emphasized, there is is aa “slowing” great deal of of uncertainty about the effects of alternative "slowing" policies policies on real output and employment as well as their short—tern short-term effects on on the financial markets. Traditional macroeconomic macroeconomic models models usually assign a fairly large and prolonged prolonged real output effect effect to anti- inflationary monetary policy. However, as as Taylor points points out in his paper, recent developments in in economic theory raise serious questions about implications of traditional models. Despite our of the effects our uncertainty about the exact magnitude of on real output, it is becoming generally generally accepted that the the less the dedethe less effect gree of uncertainty about the path of monetary actions the these actions will have on real output and the larger and quicker their effect on on inflation. Meltzer discusses this this issue under the heading of of 11 the “basic basic inference problem.” problem. 11 He shows that, that, if transitory changes in the the growth of of money money are frequent, it it is optimal to observe aa relatively long long series of observations before before concluding that aa permanent change has has occurred. The The past behavior of the Federal Reserve with respect respect to the growth money has has made this aa good rule to follow. growth of money The Federal Reserve has announced monetary monetary targets targets and then repeatedly repeatedly failed to to hit these targets. The Federal Federal Reserve Reserve has announced announced major policy policy actions designed to money growth, as it November 1978, 1978, and actions designed to slow slow money growth, as it did did in in November and then actually substantially reduced money growth for five months. -199—199— !.~i transitory change in money growth, However, this was apparently only aa transitory as the last six months have completely reversed the pattern of slow money growth. lesson that the Federal Reserve Reserve has learned learned is is Hopefully, one lesson that it it must make its policy policy announcements credible to the public. Credible means taking actions, and maintaining maintaining those those actions that are consistent with its stated policy intent, intent. Also, when the Federal Federal major change in in its method of implementing policy, it Reserve makes aa major should should clearly clearly explain this this new procedure. procedure. The immediate case case in point is the October October 66 announcement of aa move toward aa reserve reserve targeting targeting procedure. To minimize disturbances in financial markets and to have a on inflationary expectations, the Federal Federal Reserve should maximum effect on clearly clearly explain the new rules of the the game. game. How much much more short-rum short-run flexibility does the Fed Fed plan plan to to allow allow in in the federal federal funds rate? Exactly which reserve aggregate is going to to be the new target? What is the Federal Reserve's Reserves growth target for this this reserve aggregate? How How is the Federal Reserve going to project the the relationship between the reserve aggregate and money? An improved monetary policy policy for the 1980s An must include answers to to these questions. questions. -200- FLEXIBLE EXCHANGE RATES IN THE l970s 1970s Jacob A. Frenkel INTRODUCTION Our Our recent experience experience with with aa system system of flexible exchange rates iad led 1ad led to to aa renewed interest in the the operations of foreign exchange exchange narkets and and in in studying the the principal principal determinants of of exchange rates. rhe 1970s l97Os witnessed evolution of the international fhe witnessed the dramatic evolution international monemone- tary system from aa regime of pegged exchange exchange rates which which prevailed prevailed for about aa quarter of a century since since the Bretton Woods conference into into aa regime of flexible (though managed) rates. The emergence emergence of the new legal legal and economic system confronted confronted traders, national national governments governments and and international international organizations organizations with new economic problems, choices and instruments. instruments. During the 1970s 1970s exchange rates have fluctuated widely widely and inflation rates accelerated. The international monetary monetary system had had to accommodate extraordinarily extraordinarily large oil related shocks shocks which affected trade flows in goods and assets. Huge oil payments had to be recycled. Uncertainties concerning future developments in international international politics reached new heights and the prospects prospects for the world economy got gloomier. on These developments have placed unprecedented pressures on the markets for foreign exchange as well as on other other asset markets. Or. Frenkel is Professor Professor of Economics Economics at the University of of Chicago and Dr. Research Associate at the National National Bureau of Economic Research. Research. The author is indebted to to Lauren Feinstone Feinstone for efficient research assisassisNational Science Foundation for financial support. support. In In tance and to the National revising the paper he would like having benefited from like to to acknowledge acknowledge havinq useful comments by Sebastian Edwards, Edwards, Stanley Stanley Fischer, Craig S. S. Hakkio, Paul Nussa and Paul Krugman, Krugman, Nichael Michael L. L. Mussa and Nasser Nasser Saidi. Saidi. -201-201- They have been associated associated with aa large slide in in the value of the U.S. dollar, dollar, and have resulted in in speeding up up the creation of new instituinstitu- tions like the European Monetary System which which provides the formal framework for the management of of exchange exchange rates rates among among members. The inin- creased interdependence among countries and the the recognition that exex- change rate policies by one national government government exert influence influence on other economies have also induced aa legal legal response from international organizations. For example, in in late April April 1977, 1977, the Executive Board of the International Monetary Monetary Fund Fund approved the details of of the second amendment amendment to Article IV IV of the the amended Articles Articles of of Agreement Agreement dealing procedures for surveillance surveillance of member member countries' with the principles and procedures countries’ exchange rate policies. These developments provide the background for this paper which is intended to present aa brief survey of key issues and and lessons from the 197Os. experience with floating rates during the 1970s. The main orientation orientation of the paper is is empirical and the analysis is based on the experience of three exchange exchange rates: the Dollar/DM. the Dollar/OM. the Dollar/Pound, Do 11 ar /Pound, the Dollar/French Do 11 ar / French Franc and In the second second section II analyze the efficiency of of the foreign exchange markets by examining the relationship between spot and context I also examine examine and and interpret forward exchange rates; in that context the extent of exchange exchange rate rate volatility. volatility. The The analysis of of the foreign exchange quesexchange markets markets is important important because it sheds light on on several questions like: l) 1) have exchange rates fluctuated "excessively?" ‘excessively? 2) is there evidence evidence that speculation in in the the foreign exchange markets is is dedestabilizing? 3) is there evidence evidence that there is “insufficient” insufficient specuspecu11 lation lation in the foreign exchange markets? 4) is there evidence for a market are unexploited profit market failure in the sense that there are —202-202- 11 opportunities? These Issues issues are relevant for assessing the performance opportunities? of floating rates as well as for discussing whether there is aa case for government intervention in the foreign exchange markets. The analytical framework that is used for interpreting interoretinq the the volatility volatility of exchange exchange rates and the association between spot amd and forward rates is the modern modern theory of of exchange rate determination. Within this perspective perspective exex- change rates are viewed as the prices of assets that are traded traded in organized markets and, like the prices prices of of other assets, are are strongly influenced by expectations about future events. interest rates is The relationship between exchange rates and interest analyzed analyzed in in the third section from the perspective perspective of of the monetary monetary approach to the exchange exchange rate. rate. This releThis analysis is of of particular rele- vance in view of of the new policies of the Federal Federal Reserve Board, Board, which which were announced on 1979, that are on October 6, 1979, are intended intended to curb inflation and to support the the dollar. One of the key issues issues that is is raised in in this section section is the distinction between anticipated anticipated and unanticipated changes in in rates rates of of interest. interest. tion is obvious. The The policy implication of this distincdistinc- As an an analytical matter this this distinction distinction is important because the modern approach to exchange rate rate determination implies that that exchange rates rates are strongly influenced by unpredicted. 11 news which by definition is news’11 which Therefore, unanticipated unanticipated rather rather than anticipated anticipated changes changes in interest on changes in interest rates should have a strong effect on in exchange rates. This This prediction is tested empirically. The fourth section analyzes the the relationship relationship between between exchange exchange rates and prices by examining examining the patterns patterns of of deviation from purchasing power parities. parities. This This examination examination is relevant relevant for assessing assessing whether the flexible exchange exchange rate rate system was successful in insulating insulating national -203— -203- economies from foreign shocks, shocks, and whether It it provided policyinakers policymakers an added instrument for the conduct of macroeconomic macroeconomic policy. with an The purchasing power evidence on on deviations deviations from purchasing power parities is also relevant for the discussion of of whether there is aa case case for managed float. The fifth section concludes concludes the paper with some concluding remarks. THE EFFICIENCY THE FOREIGN MARKET THE EFFICIENCY OF OF THE FOREIGN EXCHANGE EXCHANGE MARKET AND THE MOVEMENT ANO THE MOVEMENT OF OF EXCHANGE EXCHANGE RATES RATES In this section I analyze the principal characteristics of the relationship spot and relationship between between spot and forward forward exchange exchange rates rates which which seem seem to to l97Os. emerge from the the experience of the 1970s. Following an analysis of of the the efficiency of the foreign exchange exchange market market I discuss the more more general general issues underlying the relationships between spot spot and forward rates and their volatility. cy of the Foreign Forein Exch an e Ma rket The Efficiency Exchange Market One of the central insights of the monetary monetary (or the asset asset market) exchange rate is the notion that the exchange rate, rate, approach to the exchange being aa relative relative price of two assets, is is determined in aa manner similar to the determination determination of other other asset prices and that expectations expectations conconcurcerning future course of events play aa central role in affecting cur1 rent exchange exchange rates) rates. exchange market is efficient efficient and if the exchange exchange If the foreign exchange rate is is determined in a fashion similar similar to to the determination determination of other asset prices, we we should expect current prices to reflect all currently currently 1‘For For collections of articles articles summarizing summar1z1ng this this approach see the Scandinavian ,Journal of. E.c.orulmii;_,_, no. no. 2, 1976, and Frenkel and Johnson Johnson (1978). -204- available information. Expectations concerning future exchange rates Expectations should be be incorporated incorporated and reflected in forward exchange exchange rates. Thus, of the market, II first regress the logarithm to examine the efficiency of of the current current spot exchange rate, an tn St, on the logarithm logarithm of of the oneoneof month forward exchange rate Fti rate prevailing at the previous month, an £n Ft-l, as in in equation (l).2 (1). 2 (1) an = a + b an Fti + If the market market for foreign exchange is efficient and if the forforward exchange exchange rate is an unbiased forecast of the future spot spot exchange exchange rate, then then we expect that: l) the constant term in equation equation (1) should 1) not differ differ significantly from zero,33 2) the slope coefficient coefficient should not differ differ significantly significantly from unity and, 3) 3) the residuals residuals should be serially serially uncorrelated. II examine three three exchange rates: Pound, the the Dollar/Franc Dollar/Franc and the Dollar/OM. Pound, Dollar/DM. the Dollar! Dollar/ Equation (1) (1) was estimated Equation estimated period June 1973 1973 —- July July 1979. 1979. using monthly data for the period The The beginning beginning of the the period was was determined by the attempt to concentrate on exon the experience of of the current exchange exchange rate regime (following the initial post Bretton-Woods Bretton-Woods transition period). period). The resulting ordinary leastleast- 22For an efFor application of the same methodology methodology in analyzing analyzing the efficiency properties of the foreign exchange market market during the German 1921—1923 see Frenkel hyperinflation of of 1921-1923 Frenkel (1976, (1976, 1977, 1977, 1979). For an application to other exchange rates during the the 1920s, Frenkel and l920s, see Fremkel Clements l92Os and the 1970s, l970s, see Clements (1980), for an application to the 1920s Krugman (1977); for an interesting analysis using time—series time-series and crosssection data, data, see Bilson (1979), for an an analysis of market market efficiency using novel econometric techniques, see Hakkio (1979a), and Hansen and Hodrick (1980), 1979).. ( 1980), and for surveys, see Levich Levi ch (1978, ( 1978, 1979) 3More precisely, if if (assuming risk neutrality) the forward rate spot rate measures the expected value of the future soot rate, then then the constant constant 0~5 2 term in the logarithmic equation (l) (1)should °~~ see should be be --0.5ag; see Frenkel (1979). -205- squares estimates are reported in Table 1. l. Also reported in Table 1 are additional regressions which will be be analyzed shortly. are As may may be be seen for the Dollar/DM Dollar/OM exchange rate, the hypotheses that (at the 95 confidence level) the constant term term does not differ signifisignifipercent confidence cantly from zero and that the slope coefficient does not differ signifisignificantly from unity cannot be rejected. These hypotheses are are rejected for the Dollar/Franc Dollar/Franc exchange rate and are rejected rejected (marginally) for the Dollar/Pound exchange rate. Dollar/Pound rate. The joint joint hypotheses, however, that the reconstant is zero and that the slope slope coefficient is unity cannot be rejected at at the 95 percent for the Dollar/Pound and the Dollar/OM Dollar/DM exexchange rates and at at the the 99 percent for the the Dollar/Franc Dollar/Franc exchange exchange rate. The test test statistics for testing the joint joint hypotheses are reported in in Table 1. the column headed by by FFin l. an efficient efficient market, expectations It was argued above that in an concerning future exchange exchange rates are are reflected reflected in in forward rates, and that that spot spot exchange rates reflect reflect all all currently available available information. If forward exchange rates prevailing at at period t-l summarize summarize all relerelevant information available at that period, they should also also contain the the information that is summarized in data corresponding to period t-2. It thus thus follows that including additional lagged values of the forward (1) should rates in equation (l) should not greatly affect the coefficients coefficients of determination signifidetermination and should not yield yield coefficients that that differ significantly from zero. The results results reported in Table l1 are consistent with an Ft this hypothesis; in all cases the coefficients of in Ft-Z do not differ 2 do of the additional significantly from zero zero and the inclusion of additional lagged variables does does not not improve improve the the fit. fit. Furthermore, in in all all cases cases the Durbin-Watson statistics are consistent with the hypothesis hypothesis of the —206— -206- Foreign Exchange Markets Mllrket:s Efficiency of Foreign Monthly Data: June June 1973 July 1979 19)3 —- July 1979 (standard errors (standard errors in parentheses) parentheses) Depend,cnt Dependent ·variable: Variable: tu in S St L Dollar/Pound DoJlar/l'ound Es r tarn Lion Estimation l'k,thod Method Constant constant in tF, t-1 th r—l OLS .033 (.017) (.017) .956 .956 (. O24) (.024) OLS OhS .031 (.038) (.018) 1.047 1.047 ((.116) .116) tv lV .035 (.018) (.018) OLS OLS in F F~ 9.n /F2 s.c. . s.e D,W. 0.8. .96 .027 . 027 1.72 1. 72 .96 .027 .027 1.94 1. 94 ..953 953 (.024) (.024) .96 .027 1.72 1. 72 —.237 -.237 ( .078) (.078) .843 (. 051) (.051) ..79 79 .029 2.23 2.23 OLS OLS -.225 —.225 (. 082) (.082) .706 (.117) (.117) ..79 79 .029 1.90 IV rv —.219 -.219 (.079) ( .079) .855 (.052) (.052) ..79 79 .029 2.25 -.023 .023 (.027) (.027) .971 ( .032) (.032) .93 .93 .032 2.12 OhS OJ.S -—.019 .019 ( .028) (.028) .913 ( .119) (.119) . 93 .91 .032 1.96 IV rv — .022 -.022 ..972 972 (.033) ( .033) .93 .93 .032 2.12 t-22 —.088 -.088 (.113) (.113) F m m 1.86 .90 Dollar/Franc Dollar/Franc ' N 0 ~ ' .146 (. 117) (.117) 4.83 2.73 2. 73 Do tIn r/DM DolL.ir/DM OLS OhS — (.028) (.028) No to: ~,Tote: .063 (. 122) (.122) ..51 51 .02 2 s.c. is the c:;taudard standard error ont of determination; s.o.c. i.s Ll1(c error of the equation i:.quat~on and F R is the coeffici coeffl~_ient determination; in in the case of i.nslrurnvntal was computed as l-Var(ut)/Var(£n The FF statistic. tests the ins trustee taT variables variables estimation estimation the R. R was 1—var (ut) /Var( In St).. statistic rests joint restrict ion that the constant equals zero and the slope equals nil ity. statistic ts joint r(.'strictiou unlt:y. The rest test statistlc i.s distributed distributed as F(2, Critical F(2,7l) (95 percent) percent). The instrumental as F(2, 71). Criti.cal values for F(2, 71) are 3.13 (95 percent) and 4.92 (99 (99 percent). The instrumental variable tisrntion method is used in order to allow possibility of errors variables arising arising variable; (iv) (lV) es estimation allow for for the possibility errors in In variables from using ‘sing Kn in F F~ rate; the instruments instruments are a constant constant and ourhin a _ as a proxy proxy for the expected future spot rate; and Durhin's 1 2 T~en~statis tic which tests for the in variables dist rihuted XX with with rank variable. variablt~._ tThe rn--statistic the absence of errors in variables is distributed 22 2 degrees of of freedom. critical value for X~(2) x( ) is 22 dcgrel'.s The critical :is 5.99 (95 percent). percent). absence of first—order first-order autocorrelated residuals and an examination of higher order correlations (up to 12 12 lags) shows that no no correlation of of any order is significant. To To further examine examine the relationship relationship between the various exchange exchange rates we we note note that one of the the assumptions underlying equation (1) was the notion that the forward exchange exchange rate measures the unobservable value spot exchange rate. value of of the expected expected future spot This This assumption propro- vided the justification for using equation (1) (1) instead of the more fundamental relationship that is is embodied embodied in equation (2): (2) ( 2) an St in = = an(S~ a ++ bb rn (S~ It - 1) 1) ++ st - where (S~ (S~ I t—l) spot exchange t-1) denotes the expected expected spot exchange rate for period t available at period t - 1. based on the information available 1. - If, however, the If, at tt - 11 is a "noisy" forward exchange exchange rate at noisy’ proxy for the expected - future value of of the spot rate, (i.e., it measures measures it with a random random error) then we would obtain that error) (3) an Ft1 = am(S~ t - 1) + vt,; E(vt) = 0 and substituting equation (3) into equation equation (2) yields: (4) an St = a + b an Ft1 + (ct - bv~i). In this case the the error error tern term in equation (1) would be be ut ut = = st - bvt-l' — and the assumption that the covariance between between an £n Ft, Ft-l and lit ut is is zero would entail aa specification specification error, error, and and the application of the ordinary ordinary least-squares (OLS) procedure procedure would yield inconsistent inconsistent estimates estimates due to least—sguares (OLS) the classical errors in variables bias. -208- In order to examine the possibility that the OLS estimates might be subject to to the errors in variables bias, one needs to to test the hyhypothesis that cov(ut. cov(ut, In in Ft_ 1 ) = 0. This test follows the specifispecifi4 (l978).~ To perform the test cation test test outlined by by Hausman (1978). equation (1) was estimated by applying the OLS procedure as well as by using an instrumental variables (IV) estimation method. Under Under the null—hypothesis of no misspecification the OLS coefficients vector bb0 null-hypothesis 0 is an efficient and an unbiased estimate of of the true coefficient vector. Under the alternative hypothesis of misspecification misspecification the vecvec- . . tor bb0 is biased and an unbiased coefficient vector b can be obtained 0 1 estimation procedure. The test— testby applying an instrumental variables estimation statistic relevant for testing the null—hypothesis null-hypothesis can be written as ( 5) (5) m = ~l - b0) (var b1 - var bY1(b - b0) where var(b11)) and var(b var(b00)) denote the variance-covariance variance-covariance matrices of . . is distributed b1 and b0,, respectively. respectively. Under the null-hypothesis mmis (in large samples) as x22 with two degrees degrees of freedom. Table Table 11 reports the results of estimating equation (1) by applying the instrumental variables estimation method. As may be seen for all exchange rates the two vectors of coefficients bb11 and b are very close close to each other. 0 For example, for the Dollar/Pound exchange rate the constants are .033 consequently, the resulting and .035 and the slopes are .956 and .953, consequently, m statistic is .90 which is well below 5.99 -- the critical value of m -- 4This test was recently applied by Obstfeld (1978) to the analyanalysis of l970s and by Frenkel of the foreign exchange market during the 1970s (1980a, 1980b) to the analysis of of the foreign exchange markets ~arkets during the 1920s. —209— -209- 2 (2) at percent confidence level. xx2(2) at the the 95 95 percent confidence level. Themm statistics correspondThe statistics correspond- ing to the other exchange exchange rates are also below this this critical critical value. value. It It is concluded, therefore, that the use of of the forward forward exchange exchange rate rate as as a proxy for expectations does does not introduce aa significant significant errors in in variables bias bias and thus the use of the OLS estimation procedure procedure seems appropriate. The efficiency of the foreign exchange exchange market and the the rationality rationality to measure expectations expectations can also of using data from the forward market to be analyzed from aa different angle. Consider equation equation (6): n (6) xt = a 0 + + ~ t +i=1 5~xt~+ ~t_~ Cll I + w~ i=l where xt denotes the percentage change in the spot exchange exchange rate where x~denotes (an 5~ ~t-ldenotes (in St - an zn St_ 1 ), rrt-l denotes the forward premium on on foreign exchange exchange (an an St_ ~tl~’ (in Ft Ft-l t denotes time, nn denotes denotes the number of lags, and 1 ), ~ 1 - in - ~ - ww denotes an error tern. term. If ~t-1summarizes If rrt-l summarizes all available available information concerning the future evolution evolution of of the exchange exchange rate, then qiven given the the value of of the forward premium nt-l, the past history of the percentage percentage ~ change of of the exchange rate should not "help" prediction (i.e., the change “help” the prediction past history should not be viewed as Granger-causing Granger-causing future changes), and the joint joint hypotheses hypotheses that a 1 and Si are zero should not be rejected. rejected. The results of applying these tests to the three exchange rates for various number of lags are reported in in Table 2. Also Also reported reported in Table 22 are the the results of testing the the joint hypotheses that aa11 and Si are zero and and that y, y, the coefficient coefficient of the forward premium, is unity. unity. The The relevant statistic for testing the null-hypothesis is an an F— F- statistic which is reported in Table 2. 2. As is evident in all cases cases the null-hypothesis cannot cannot be rejected at 95 percent confidence level null-hypothesis at the 95 -210- since the values of the various F—statistics F-statistics fall well below the the corcorsince responding critical critical values. It is concluded, therefore, that the the forward premium on on foreign exchange exchange may be viewed as aa rational expectations measure of of the percentage depreciation of the currency in that it it inincorporates corporates the available information that that is contained contained in the the series of past depreciations. that may may be be drawn from the previous previous The principal conclusions that discussion are that the behavior of the foreign behavior of foreiqn exchange market market during been broadly consistent consistent with with the general implications of the 1970s l970s has been the efficient market market hypothesis hypothesis and that the forward exchange exchange rate summarizes the relevant available available information concerning the future evolution of of the rate. Exchang_<e_Rate Movement: Volatility and Predictability E xc han ~ In this this section II analyze the volatility of exchange rates and the the extent to which this this volatility is predictable. To To set the stage quarterly perfor the the analysis, II present present in Figure l1 the daily and quarterly per- in the three exchange exchange rates. rates. centage changes in This figure indicates exchange rates have been been very volatile and that the that the various exchange degree of volatility of day-to-day changes in the exchange exchange rates have been extraordinarily extraordinarily high high and has been much smaller when averaged over over longer periods. Further, the standard errors of the regressions in indicate that the forecasts of future spot exchange rates rates based Table 1 1 indicate on the forward rates are imprecise: the standard errors of the equaequa- tions are about 33 percent per month. These characteristics of price changes are typical to auction and to organized asset markets. In such markets current prices prices reflect reflect —211— -211- Table Table 22 Test of Rationality Prediction of currency Rationality of Forward Forward Premium Premium Prediction Currency Depreciation Depreciation 1973 —- July Monthly Data, Data, June 1973 July 1979 1979 Dependent Variable Dependent 5 an S~-— an t—l in st in st-l Null Hypothesis Null Hypothesis Number of Lags Number of Lags F—statistic 3 F(4, 64) =• 1.680 1.680 F(4,64) 44 F(5,62) P6,62) 5 F(6,60) • 1.231 fl6,6o) 1.231 66 F(7, 58) P(7,58) 33 55 = = 1.555 1.555 P(6,62) F(6,62) =• 1.518 1.518 P0,60) F(7 ,60) • 1.207 1.207 66 P(8,58) F(8,58) 3 3 P0,64) F(4,64) 44 P3,62) F(5,62) =• 1.327 1.327 5 F(6,60) • 1.087 P(6,60) 1.087 P0,58) 1.014 F(7 ,58) 1.014 F-statistic Dollar /Pound Dollar/Pound a = 0, 13 • 0 a 1 = 0~Bj1 — 0 1 44 = o, r = 1 0, ~ 1 0.1"" 0, e1 a = 0, ~ 1 Dollar/Franc Dollar/Franc "1 o, ~i 0, $.. = = 00 66 a"1 1 0, 3 . • i 0, 5. o. y = 1 0, y 1 Dollar/DM Dollar/OM 0, 8.0• 0 1 "' 0, Sia o, 0, 6. Si = 0, y 1 = 1.131 1.131 1.175 1.175 P6,64) F(5,64) 1.436 1.436 P3,62) F(6,62) 1.519 1.519 5 66 5(7,60) F(7,60) 1.146 1.146 P(8,58) F(8,58) 1,063 1.063 3 P3,64) 44 P6,62) F(5,62) S5 66 P3,60) F(6,60) == 1.321 1.321 5(7,58) 1.342 F(7 ,58) = 1. 342 3 P(5,64) 44 5 P3,62) F(6,62) P0,60) 66 5(8,58) F(8,58) 5 —212— -212- P6,64) F(5,64) 33 3 aa "1 1.141 1.141 44 3 a 0.1 1.610 1. 610 F(4,64) = 1.262 1.262 1.183 F(5, 64) F(7 ,60) 1.123 1.123 1.183 = = = 1.287 1.287 1.403 1.403 1.525 1.525 Pigure Figure 11 ,HORT-RUN VARIABILITY IN EXCHANGE RATES HORT-RUN VARIABILITY IN EXCHANGE RATES APRIL 2, 1973 DECEMBER APRIL 2, 1973- DECEMBER fOuanerty pe,centage percentage changeal changes) lOuarterly Daily percentage changeel (Daily changes) t ~ [U~iTED~INGDOM UNITED KINGDOM tE- I~ ~1 • 11i!tt /. , rs 11,, 10 10 ~ ' 30 :UNITED KINGDOM I r~ '~1 j j ~ h~j 1 ,~,-~ 1 0 0 1 ' 7975 978 977 7979 1973 lif l~ .... ,. . . ~ ~ l, ,~i*LLLJ ~ SitLb~ 4 976 977 1978 ~ j 1975 978 1 4 FED.REP OF GERMANY ~11,,t l 978 0 -10 -20 ~ 973 1974 978 978 1977 978 30 - OF GERMANY 20 - A ~ 10 ~= -10 ~ 975 10 978 1977 20 0 '--,-es,--'-,,,,..........,ai,--'=a~---cc--::,,,.---'-L10 1973 1974 197S 1976 1!1n 19711 1974 1975 FRANCE I ~-10 I -! 973 1974 30 j' , 1974 0 ‘\j~y1i==~=~i 100 f-FRANCE FRANCE 1973 I 10 "j f 974 20 - .10 1*7~~ IM+ ½,,,=,~"',,,~.~~.,"-",.~....c,,,,.~~.-,,,-'-~,.9;,a~......J -10 973 IN TERMS TERMS OF OF U.S. U.S. DOLLARS, DOLLARS, IN 31, 1978 31, 1978 -20 1977 197$ 973 1973 Source: Artus and and Young (1979). source: (1979). —2 13— -213- 1974 1914 1973 1915 978 1976 1977 19n 1378 "" expectations concerning future course of events, and changes In in expectexpectations are immediately reflected in corresponding changes in prices. Periods Periods which are dominated dominated by uncertainties, mew new information, rumors and announcements announcements are likely to be be periods in which which changes in in expecexpec- tations are the the prime cause of fluctuations in asset prices. Further, Further, information which alters expectations must be new, the since the information resulting fluctuations in in price cannot be predicted predicted by lagged lagged forward . h are based base d on . f orma t·ion. 55 Therefore, Therefore, during exchange which information. on past pas t 1n h1c exc hange rates ra t es w such exchange rates to to exhibit exhibit large fluctufluctusuch periods, one should expect exchange ations and to be unbiased but imprecise forecasts of future spot rates. gain further insights To gain insights into the implications of this perspecperspective on on the relationship between predicted and realized changes in exexchange change rates, rates, II present present in Figures 2-4 plots of of predicted predicted and realized changes in exchange rates rates for the the three pairs of currencies where the predicted change is measured by the lagged forward premium. Also prepre- sented in these figures are the differemtials differentials in national national inflation inflation rates which are discussed discussed in the the fourth section. The key fact which emerges from these figures is that predicted predicted changes in exchange rates 6 account account for aa very small fraction of actual changes.6 55The The analysis of the role of “news "news" in determining determining current exexchange rates and in explaining forecast errors from the forward forward rate has been made (l976a, 1976b, l976b, 1977, 1979a). See made forcefully by by Mussa (1976a, Dormbusch (1978). The large degree of volatility is also also Dornbusch also analyzed by McKinnon (1976) who attributes it it to to insufficient insufficient speculation. speculation. 6These and the the following empirical regularities are analyzed analyzed in in interdetail in in Mussa (1979a). See also Frenkel and Mussa (1980). An interesting extension extension would would examine the relationship relationship between the variances of predicted and actual changes in exchange exchange rates rates in aa manner analogous that of of Shiller (1979). to that to Shiller (1979). -214— -214- P4 tO !( US/UK C — N —~--~ N ~ ~ I I I I 1,1 I I I1' lI II ,, I ,, 1I II I ,,Ii 1,,, i I I ',,1 ,,,, ,, 1, i I I - , ‘ 1977 N’~Ca 1976 to N. 1975 1978 1979 L4,a a) (- a) “C a) Ci 4,) 0--co P a) 01 —I I H a,J 0 U a) “‘—‘ Na Ci 00 ci Ca • ‘C CO 0 4’) N’ Ca Ca LI’) C’) -215- a)a) ,C 4,) (4,4 0 Ti 000. “’a 0 a) H “a I ‘-“ 0”,0 a) ~ Ci (00)04 ‘~‘), Pa H Monthly percentage changes of the U.S./U.K. consumer price i~d~ces [~(,;',n COLLS/COL;K)], of the $/5. excha~ge rate, (L.. ,,:.,n S ) , and the montklv forward premium; f"n(F /s t- )] t • t- 1 1 July 19/3 - July 1979. a’) Ci 0Ci ‘0Ci “a a) H a)0’, 0 a) p a) Figure 2: Li, 1974 — ~ C~’-, r— 2? 1973 : i ,, ,,,, ,,,, I I, 11 ':',l; i,,! I1 11 11 11 I 1 ·,,!\,. , , ,, I \ / I I I — — •_.—___———— — __ ~ .r-a’t.~.— I I I : == I,''1 I I ~=_.n— COLUK I '11 ) I .— /:,.ln ( _sl§ i' ~- “~Thgi o o 00 COL _.r..t—r.— I I = j I I 0 \j I '-.!I 1 — — II p : I ,, i'i I/ 1 I —— ' 1· — ..=•.— : — == F.... __= = II ' ..v- — 11 ,Cl ‘ (0 “‘~:::::::“~ — / —. l . I IJ ' ' · I I\ /\ . 1 l; I . : : \ 1'1 ~ ; .1 I I i; i t; I I I I ·· 1 !, I 1 I.. ; , , ' I I i\J ;/11 1I ;f ; ! } I I I 1, ' I I j,,:! 11lU \,I II:; !i :\ i 11'' I 1 I .I \: \' I I l ;J If , ; l f JI 1 ii I I I I 11 :; ~ . I I I II I I I ' \ I I ;, I : I I I I II I I I I \ ,v~ I I I I I 11 I I I I 11 I I II ,\ I 11 1 1 I — [':1' I — 't \. I Ii\ ' fi l..._ .J \ ! I -‘- I I!\ ,,. .{ i CO _. a)0~’) — .~“ :..._—~.- ll I I I 2? I : I~ C) 'I n, ' '.;-1, : I —. . . . 11 ~ :I I \ I ~ ;11r\ V\ I • I\ I I I I II I I I I I 1I I ii I I I I 11 1 11 I \ I 1\ : I I \\A 1'1 I I I 1,11,1 I I\ J· \_, I :f ···I .. I •. Y lJ :'·. ·II _.. ·. 'I ·.. .' } I . ' '' ;, .. ! ·. 1( I ·, I ... ,·•.1 _tt, I I I I —— : : i1 I I 1 ~.. —~ !: ' !I ii I ,1 ,1, I / I lI — " IflMVHI 1N1IU1A r I -‘-- I /": II ·, I I i i I I : ll I..' I ! ~\ 1 ,C :\ l ri ~ II II / II I I I I 1 I I I l I Ii 11 II I 1 I1 1 1 I I I I I I I II ,,, I III I I II I jl I I 1 1I II I ii I: I I 11 I I 11 11 1 1 I I /I I I I I 1 1 I 1 ' I I — — 1' II ,1 I I I ! 11 11 11 ____ i——— C b. lnSt i/ 1' II I II II ii II II Ii 11 price Monthl y percen tage change s of the U.S./F rance consum er rate, ge exchan . $/F.Fr the of , indice s, [6(£n COLU5 /COLF] )] (6 1n St), and the monthl y forwar d premiu m; [1n(Ft _ 1 /St_ 1 July 19/3 - July 1979. Ci Ci -216- ~ Ct ID C, —-cotlCa Ct 1979 CO ,.~ CO “i ~° 1978 “‘-Ca ID 0)11)41 — Ci 11) 0 Ci m Ci Pt Ci (-C I 00 ID -4 “ 1977 0’ ~ H. Pt Pt Ci Ca Ca coPt. ‘0Ot. Ci -~ 0) CO ~ __ 1976 n-rn ~ hI’ 0 414 o o ~1 “i ID H’t ~‘ ‘--i CO 1975 coCCi ‘-‘4~~ Coj~ CD ID CO ‘C 0’ ,—, 1974 Ca Ci Ci Ci’ 0-ri - ‘0cfiW’< -. an’Dn.fl..r ~_—_—_:t_-=.=fl:•• Cl. ~ ‘~ 7 COrn -~ — —. .. . .•• .: . ‘ 0 ~_% — — — — — — L — r•.. .,.• ~.. . .. ~ ...t—. ~_7 ——— .nr ‘~‘ —c N N “~-~t_. —a >1 —~~-rr_ “r,~. ,,,_~ PERCENT CHANGE :z 4.’ Figure 3: CO ....z "" = .... ,_ 0 .., 0-’Ci < :,:: Ci 1973 I ~ ....'-" N.) Pt Ca) C 0 = :t~. 3 —CA 0 0 ~rJ CA US/F ranc e l 11 11 Ii 11 11 0H “S ‘I) p ±J US /Ge rma ny — • 4*’-” _,— ..-“ t.~-~— — —• • ~ —— — ~.. ,—.. ‘c__F.. •, a.—~ — INIDHId —_•4•a.. — S’” InIJWII9 -c~~T~ r— ~ C!) <1a-~——~ c...: 7 ‘-a.----_—.— C I -~ .. yO C : . Germ an consu mer pric e Mont hly perc enta ge chan ges of the U.S./ exch ange rate , indi ces, [6(£n COLU 5 /COL ~)], of the $/DM lln(F t- 1 /s t- 1 )] ium; prem ard forw hly (6 Zn s ), and the mont July 19 3 - July 19 79. 7 0) ‘— to at — C” Ct — _ I”- Pa Ca .‘— -‘~ p00a)P04004~”) a) II a)Cl’ a)”-’Cl) 0) 0% P — (0 CiWNa r-. 000” III C) HCl,) 0-’ CO Ca a) a)$a) 2? toti 1H 2? ,~i”’-’ oo.,o.c C) 00a) ~4 (1) H 0’ H ~,, (001-P a) 000 0% 0 a) Ci 0’ci a)H 04<01 ciCJ’1$N’ Ci”) i Figu re 4: -217- 1979 1978 1977 1976 1975 1974 1973 This fact rate changes seem to be fact suggests that the bulk of exchange rate 11 11 due to “new information” which, by new information by definition, could could not have been been anticipated and reflected in the forward premium or discount discount which which preprevailed in the the previous period. In order to examine this hypothesis, hypothesis, II present in in Figures 5—7 5-7 plots of the spot and the contemporaneous contemporaneous forward exchange rates for the three pairs of currencies. Also Also presented presented are the ratios of national price levels which are discussed in the fourth section. section. If If the dominant factor underlying changes in rates rates is new information, which alters views about current and expected future exchange exchange rates by which approximately approximately the same amount, amount, then one should expect aa high high correlation correlation between rates, between movements movements of spot and forward rates. This fact is clearly demclearly dem- exonstrated by Figures Figures 5—7 5-7 where where it it is is seen seen that spot spot and forward exchange change rates rates tend tend to move move together together and by by approximately approximately the same ampliampli- tude (the vertical difference between the two rates correspond correspond to the forward premium of discount on foreign exchange). The high correlation between movements in spot and forward rates is expected expected since the two rates respond at the same time to the same same flow of mew new information. This characteristic is is typical to the foreign exchange exchange market market as as well as to other markets for stocks and durable assets. The recent pattern gold prices provides aa useful example of of gold of this this general principle. principle. and the the future price of of gold as as recorded recorded rereTable 3 3 reports the spot and on four recent consecutive cently in the the New York Commodity Exchange Exchange on days. The The two key facts which are illustrated by this this table are are the exex- tent of day-to-day day—to—day volatility in gold prices prices and the uniformity uniformity by which these changes are reflected in the price of gold for immediate immediate delivery delivery as well as as in the prices for the twelve future delivery delivery dates. -218- /··--.... __________'\ \......... . I /I I I I ~ I \I • ' I a)-, ' "' ;:: 0C < a: tnFt LI = 0- = C ‘F, -J 4— C = (0 C = LI >1 US/UK US/UK 1973 Figure Figure 5: 1974 1976 1975 ! 1976 1976 ! 1977 1978 1978 ! 1979 1979 Monthly Monthly observations observations of of the the Dollar/~spot Dollar/E spot (S-n (:n S ) and and Forward Forward (Zn F) F~) Exchange Ratio of the U.S./U.K, Cost (tn Exchange Rates Rates and and the the Ratio of the D.S~/U.K. Cost of Living Indices [Zn (CCL~J ICOL )(scaied to equal the of Liiing Indices [Zn (CO½J 5 !COL1TK)(scaled to equal the spot spot initia~monE~)]:June exchange rate at the initial month)]:June 1973 —- July July 1979. 1979. —219-219- !-... . \·-~~ ~ .... \ · · · •. I···.....\ i CCL ln( ~ \ CCLF ·•••·••··.·............··... —Inst .•·······.. I I /I I I I I / I I/ I V I I I "'lnf:t 1973 1973 Figure 6: 1974 US/France US/France 1976 1976 1975 1975 1977 1977 1978 1979 1979 Monthly observations of the Dollar/F.Fr. St) and Monthly observations of the Dollar/F.Fr. spot spot (Zn (£n S) and ForwarC Forwar< (Zn Ft) (in Ft) Exchange Exchange Rates Rates and and the the Ratio Ratio of of the the U-S./French U.S./Fr~nch Cost Cost of of Living Living Indices Indices [Zn(COLTTS/COLF)(scaled [tn(COLUS/COLF)(scaled to to equal equal the the spot spot exchange rate at the initial month)]:June 1973 -— July July 1979. 1979. -220- I LnFt In US/Germany US/Germany 1973 Figure 7: 1974 1975 ‘1976 1976 - ! 1977 - - 1978 1979 Monthly observations of the Dollar/PM Dollar/DM spot (Zn (tn SS)) and Forward (Zn F~) Exchange Exchange Rates Ratio of Cost (2..n F) Rates and and the the Ratio of the the U.SJGerman u.sJGerman Cost of Liiing Living Indices [Zn(C0L~ /COL~) (scaled to equal the spot Indices [2n(COLUS/COLG)(scaled to equal the spot 5 exchange rate at the initial month)]:June 1973 —- July 1979. 1979. —221— -221- Table 33 Table Futures Price of Gold Gold on Consecutive Days Futures Daily Data; Data: October October 1, 1979 —- October 4, 1979 ---~--------··-Price (per (per ounce) change from day Price ounce) and and change from previous previous day ___ Delivery Date __ Oct. 1, 79 Change Oct. 2, 79 Change Oct. 3, 79 3, 79 Change Oct. 4, 79 Change ,, 416.0 1;16.0 21 21 411.0 411.0 —5.0 -5.0 393.5 393. 5 17.5 -17. 5 369.7 369.7 23.8 -23.8 November 419.5 20 416.0 —3.55 -3. 397.7 397. 7 —18.3 -18.3 377.7 377. 7 —20.0 -20.0 December 424.5 20 20 421.00 421. —3.5 -3.5 402.5 —18.5 -18.5 382.5 —20.0 -20.0 1980 February 432.8 432 .8 20 429.6 —3.2 -3.2 410.7 18.9 -18.9 390.7 N ' 20.0 0J ~j -20.0 April 440.9 20 438.0 438 .0 —2.9 -2.9 418.8 418.8 —19.22 -19. 398.8 —20.0 -20.0 June June 448.6 20 41,6.0 446.0 -2.6 —2.6 1,26.6 426.6 -19.4 —19.4 406.6 -20.0 —20.0 August 456.3 20 20 454.0 —2.3 -2.3 434.4 —19.6 -19.6 Lfll, .4 414.4 —20.0 -20.0 October 464.0 20 462.00 1,62. —2.0 -2.0 1,42. 442.22 —19.8 -19.8 422.2 —20.0 -20.0 December 471.5 20 469.88 1,69, —1.7 -1.7 449.8 —20.0 -20.0 429.8 —20.0 -20.0 1981 February 20 20 20 477.5 477 .5 485.0 —1.4 -1.4 —1.1 -1.1 457.5 465.0 1,65 .0 -20.0 —20.0 —20.0 -20.0 437.5 437. 5 April 478.9 486.1 1,86.1 445.0 445.0 —20.0 -20.0 —20.0 -20.0 June 493.3 493.3 20 492.5 —0.8 -0.8 472.5 —20.0 -20.0 452.5 L152 .5 —20.0 -20.0 August August 500.5 20 500.0 —0.5 -0.5 480.0 —20.0 -20.0 460.0 —20.0 -20.0 October 1979 October Note: These prices are settlement prices at the Commodity Exchange, New York as reported in the Wall Street Journal October 2-5, 2—5, 1979 c’.J N ' Another feature feature which which is is revealed revealed by by Figures Figures 5—7 5-7 Is is that that the the Another approximately equal equal ontemporaneous spot and forward exchange rates are approximately hus indicating that the markets market's best best forecast of the future spot spot rate ss (approximately) the current spot rate. This phenomenon reflects reflects the act that, as an empirical matter, exchange rates have followed approximately) aa random walk process. For such aa process, current rices are indeed indeed the best forecasts forecasts of of future prices. To To the extent extent hat the exchange rate had some drift, drift, the above statement statement should be drift. nterpreted in reference to that drift. empirical phenomenon phenomenon This empirical exchange rates even even though eems to correspond to the actual paths of exchange t does not reflect aa theoretical theoretical necessity. characteristic of the foreign exchange market is deThe final characteristic de- cribed of currencies cri bed by Figures 8-10, 8—10, which plot for the three pairs of he spot exchange rate and the forward premium on on forward exchange. ince funda— ince the units of the spot rate and the forward premium premium are funda- different, the two series were normalized by subtracting from 1entally entally different, ,ach series its mean and by dividing by the corresponding standard ‘ach ‘rror. ,rror. The fact which emerges from these figures is that generally though not always) there is aa positive ex— positive correlation between between the the ex- ,ected currency (as (as measured measured by by the forward premium ected depreciation of the currency ,nn foreign exchange) and the spot exchange exchange rate. rate. This positive positive cor— cor- ·elation may may be be rationalized by by noting that currencies currencies which are ex— ex-elation Iected •ected to depreciate are traded at aa discount in the forward market nd, on average, these currencies .nd, currencies also also command aa lower foreign exchange exchange alue in the spot market. •alue This correlation is interpreted further in he next section. :he -223-223— I \‘ \ \ \ ~· I I '- I’ / ‘ I \,J I I I l \ ~ \ I 0- I I C = C I I I I I I C U- ~I I \ 'r, \ Q C :z C <C ..... .... U-a \ \ I I I,. I '-I I~ I _,,I C <C ....= I I / (\ I = I I IV _,,......,✓ = U-a \ I I I :z "' C <C \ = ::r: LI '-' ‘C X .... U-a US $/UK£ $/UK £ ‘1973 1973 Figure 8: Figure ‘1974 1974 ‘ - 1975 ‘ 1976 1976 - 1977 1977 ‘ ‘ 1978 ‘ - 1979 1979 Monthly ob~ervations normalized Dollar/b Dollar/L spot Monthly obftervations of the the normalized spot excha~ exch~~ rate S~) and the normalized rate (Zn (2.n S~) normalized forward forward premium [£n(Ft/S ) l] Both series series are normalized normalized by by subtracting series ij subtracting from each seri~s mean error: mean and and by by dividing dividing by by the the corresponding corresponding standard standard error: June 1973 -— July July 1979. -224- I ,,~ ,,,, ~ / I Ln(.t) / I / / I\ I I I I I I I I / I I I 'v I I II ! 11 I I \ 11 ,) I I I I I I I \ I I I ,, I I I I I I \ \ I I ~, '-1 I I I ~, I,—’ / i'..1 ,.....- NM InSt I I I I ' I L-, I I I , , — .......... / """ lnS 1 1973 Figure 9: 9: Figure 1974 1974 1975 1975 - 1976 1976 - 1977 1977 US $/FFr - 1978 1978 - 1979 1979 Monthly of Monthly observationj observatio~ of the the normalized normalized Dollar/F.Fr. Dollar/F.Fr. spot spot 1 exchange (Zn 51) and the normalized Forward exchange rite r~te (tn S ) and the normalized Forward Premium Premium 1 [Zn(F~/S normalized by [£n(F /S ?]. ) ] . Bothtseries Bothtseries are normalized by subtracting subtracting from ~acfi each series series its its mean and by dividing dividing by the corresponding corresponding standard standard error: error: June June 1973 1973 —- July July 1979. —225— -225- In(.1) r U-a = 0- C ( C I I = = C \ \., \ I U.. C, C z C < La-a C = U-a C0 C = LI DC U- US $/DM 1973 1973 Figure Figure 10: 1974 - - 1975 1975 - 1976 1976 ‘ - ‘ 1977 1977 - - 1978 1978 ‘ - 1979 1979 - I ("~n SN Monthly observations of the normalized spot (Zn Monthly observations normalized Dollar~DM DollarNDM soot and the normalized normalized Forward Premium [.Ln(F [Zn(F~/S~Y ]. Both /St)~]. Both seria serit are normalized from each series normalized by subtracting from ieries its mean and dividing by the corresponding standard error: 1973 —by dividing corresponding standard error: June 19i3 July 1979. 19 79. July -226- EXCHANGE RATES, INTEREST RATES AND INNOVATIONS In this this section II analyze the relationship relationship between exchange rates and interest interest rates from the analytical perspective perspective of the the monetary apapproach to the exchange rate. To To set the the stage stage for the analytical analytical dede- velopment useful to recall the the typical typical analysis which which generally generally velopment it is useful predicts aa negative association between the rate of of interest and the exchange rate. According to that analysis analysis, aa higher higher rate rate of of interest , attracts foreign capital which induces aa surplus in the capita capital1 account of the balance of of payments and thereby induces an appreciation appreciation of of the exchange rate). rate). domestic currency (i.e., aa lower spot exchange Another variant variant of the popular approach states that the higher rate of of interest interest lowers spending and thus induces a surplus surplus in the current account account of of the balbal- ance of payments payments which results results in aa lower spot exchange exchange rate. AA third variant of this approach approach claims claims that that the higher higher rate of interest implies (via the interest interest parity parity theory) aa higher forward premium on on foreign exchange and to to the the extent that that at aa given point in time the forward exchange rate is predetermined predetermined by past history, history, (am (an assumption that is is clearly rejected by the evidence on the the comovements of of spot and and that forward rates), the reguf red rise in the forward premium required premium will will be brought about by aa lower spot spot rate rate (i.e., (i.e., by an appreciation appreciation of of the domestic currency). Whatever Whatever the route, this approach predicts aa p~g~~jve exthe rate rate of interest and the spot spot exnegative relationship between the change rate (or alternatively, aa positive positive relationship between the rate rate of interest and the foreign exchange value of the the domestic currency). currency). These predictions, however, do not seem to to be in in accord with with the the broad facts. Over the recent period the rise in the rate of of interest in in the U.S. (relative to to the foreign rate of interest) has been been —227— -227- Figure Figure 11 Foreign Exchange Value of the U.S. Dollar and Interest Rate Differentials Ratio Stale Scale March 1973:100 MarCb 1973 =100 110 105 _, ~ Weiqhted Foreign Currency thee Dollar w·htdA e1g e Ayeraoe veraae F ore1an C urrency Value VI a ue of ofthDII o ar - Ratio Scale Ratio MauI 1913:100 M arcb 1973 =100 110 los 105 ~ 100 '' 95 90 100 A 95 \ v- - 8S 80 Percent 5 90 90 ,/ as 85 80 at Percent 5s U.S.-Foreign Interest Rate Differentials f\. 4 J Short-Ter: 3/ 3 2 - 44 \' 33 22 r' J---, ----r-,-. ... / ' , ...'- 0 ., -~ -2 long Termll ..... ,- I •• ~ .... , _ .... 4 \. .3 ...- .J.-'-'I —1 -i ! -2 ·2 :~ .3 '-'1' .4 0 1976 i 1977 1978 1q79 1979 .4 -4 Sources: Federel Reserve Statistical Release H.13; Federal Reserve Reserve Bulletin; Monetary Sources; Federal 4.13; Federal Bulletin International Monetary Fund, International !nternationol Financial Statistics. -Statistics, ii in the the Ll Secondary Secondary market market rates rates for for 90-day 90-day large certificates certificates of deposit deposit in the united United States States less less the weighted wl!lighted average overage of foreign foreign three-month money market market rates. :1_ U.S. 2. us. long.term gov~rnment bond bond yields less the overage of Ieng.term Iong.tern~govornment yielas less the weighted weighted average of foreign foreign long-term government bond yields. yields, Latest data plotted~May latest data plotted: May Source: Source: D. R. 1. Mudd (1979). (1979)’. -228- depreciassociated with aa rise rise in in the spot spot exchange rate (I.e., (i.e., with aa depreci- ation of the dollar). ll illustrates the point by plotting the Figure 11 against the interest interest rate foreign exchange exchange value of the U.S. dollar against differential. As As is evident, in contrast with the popular prediction, the higher (relative) (relative) rate of interest in the U.S. has has been associated with (i.e., with aa lower foreign exchange exchange value with a higher exchange exchange rate (i.e., of the the dollar). dollar). This exThis contradiction, contradiction, however, however, does does not not arise arise when when the the ex- monetary (or an asset market) perspecPerspecchange rate is analyzed from aa monetary tive to which we mow now turn. monetary approach approach are hypotheses hypotheses The major building blocks of the monetary concerning the properties of the demand demand for money and money market equilibrium and hypotheses concerning the link between between domestic and foreign prices. 77 Consider Consider first the equilibrium equilibrium in the money money markets. markets. N/P and M*/P* The supplies of domestic and foreign real balances are M/P where NMand and PP denote denote the supply and level, where the nominal nominal money money supply and the the price price level, respectively, and where variables pertaining to the foreign country are indicated by an asterisk. Denoting the demands for real balances by LLand and L* (both of which are functions which are specified specified below), equilibrium equilibrium in the the money markets is attained when (7) (7) L L = M/P N/P and (8) L* = M*/P*. 7For theoretical developments and applications of the approach see, for for example, Dornbusch (1g76a, (1976a, 1976b), Kouri (1976), (1976), Mussa (1976a), (1976a), Frenkel (1976), Frenkel and Johnson (lq78), (l~78), Frenkel and Clements (1980), Bilson (1978), Hodrick (1978), and Frankel (1979). —229— -229- From equations (7)-(8), equilibrium in in the money markets implies that the ratio of the two price levels is: (9) (9) PP ~ P* MM 1* M* L* IL · The second building block links domestic and foreign prices. 8 Assuming purchasing power Assuming the simple version of purchasing power parity implies that:8 (10) pP == SP* Using equation (10) in (9) yields ((11) 11} 5S - MM M* LL* M* I which which expresses expresses the exchange exchange rate rate in in terms terms of domestic domestic and and foreign supsup- plies and demands for money. To gain further insight into the deterdeter- minants of of the the exchange exchange rate and to set the stage for the the empirical estimation, assume assume that the demand for money money depends on real real income (y) (y} and the rate of interest interest (i) (i} according to: (12) L = ay~e~ 8For For aa discussion of the choice of of the relevant relevant price price index to be be used in equation (10}, (10), see Frenkel (1978). This simple version of the purchasing power power parity theory is used here to simplify simplify the exposition. To the extent that there are systematic systematic deviations from from purchasing power parity they can be incorporated into the final final exchange rate equaequation. Similarly, Similarly, to to the the extent extent that purchasing power power parities holds in in the long long run but not in the short run, the final exchange exchange rate rate equation equation will reflect these dynamic characteristics. characteristics. To the extent extent that purpurchasing power power parity pertains to traded goods only, the exchange exchange rate equation would would also contain terms which relate relate to to the relative prices orices non—traded goods; for aa formulation along these lines, see of traded to non-traded Dornbusch (1976b) and and for an an empirical application, see Clements and Frenkel (1980). AA more Frenkel (1930). more refined refined specification specification would allow allow for the effects of tariffs on on the relationship between domestic domestic and foreign prices as well as as for short-run effects of of unanticipated unanticipated money on outoutput on prices and the the exchange exchange rate. put rather rather than than only only on prices and rate. -230- (13) L* = Using equations (12)-(13) in (11) (11) and assuming assuming for simplicity simplicity of exex(12)—(13) in momey position that foreign and domestic parameters of the the demand for money ~ we are the the same, same, i.e., that that c~ a= c~, a*, and that ~n = n*, we obtain: ((14) 14) zn S ~n S = v* zn M ++ nn tn i *) CC ++ ~n ~n ~+a(~( i - i*) M* - y where CC s in(b*/a). in(b*/a). Equation (14) relates the exchange exchange rate rate to to the ratios of domestic to foreign money supplies and incomes and to to the interest rate rate differendifferential. tial.99 Most Most pertinent to to the present purpose and in agreement agreement with the facts summarized by Figure 11, relation11, equation (14) (14) yields aa positive relation- rate of interest and the exchange exchange rate. ship between the rate The economic economic The interpretation of this this association in in the context of the U.S. dollar dollar and the inflationary environment environment is is as as follows: aa rise in the the domestic domestic (relative) (relative) rate rate of interest is primarily primarily dominated by by aa rise in the exex- pected (relative) rate of of inflation which induces aa decline decline in the dedemand for real cash balances; for a given path of of the nominal money supply, asset market equilibrium requires a price level which is higher than the price which would would have have prevailed prevailed otherwise. Since the domestic price level level is linked linked to to the foreign price price through some form of purchasing purchasing power power parity, and since the path of the foreign price is 91t should be rt noted that aa similar similar set of of variables would also appear in the reduced form of aa variety of alternative alternative models. The dedependence of the demand for domestic money money on on the domestic rate of ininterest and the dependence dependence of of the the demand for foreign money on foreign A more rate of interest is assumed only for simplicity of of exposition. exposition. A general formulation would recognize that the demands for domestic domestic and foreign monies monies depend on all margins of substitution. substitution. See See Frenkel and Clememts Clemen ts ((1980). 1980) . —231-231- assumed to be given, the higher domestic price can only only be achieved achieved through aa rise in in the spot exchange rate (i.e., through aa depreciation of the currency). of of the positive association association between between interest interest This explanation of This explanation the positive rates and exchange exchange rates has has an intuitive appeal in in that it implies implies an inflationary environment, aa relatively rapid rise in in prices that, in an is associated with high nominal nominal rates of interest as as well as with aa depreciation of the currency currency in in terms of foreign exchange. The traditradiThe tional prediction of of aa negative relationship relationship between interest interest rates and the exchange exchange rate rate may, however, be reconciled with the monetary monetary approach under the assumption that it short—run liquidity it concentrates on the short-run effects of monetary monetary changes. Accordingly, in the short-run, aa higher rate of interest interest may arise from tight money which induces induces an appreciaappreciathan aa depreciation of the the currency.lo should be be ememtion rather than currency.1° It should phasized, however, that during an an inflationary environment environment (like (like the one prevailing prevailing in in the U.S. in recent years) the variations in the rate of of interest are most most likely likely to be be dominated dominated by variations in inflationary inflationary expectations rather rather than by by liquidity effects associated with with changes in the ratio of money to bonds. environment the rate In such an environment rate of interest is expected to be positively correlated with the exchange exchange rate. The discussion provides an illustration of of the difficulties rate of interest as the relevant 111onetary associated with using the rate monetary indicator. Traditionally, the the height of the rate of of interest interest was the !OThe short—run liquidity effects is emphasized in Dornbusch lOThe short-run liquidity effects is emphasized in Dornbusch (1976b). The role of inflationary inflationary expectations in in dominating exchange exchange rate developments developments is emphasized emphasized in Frenkel Frenkel (1976). (1976). Frenkel (1979) and and Edwards (1979) attempt to integrate these two factors. -232—232— aiterion monetary policy iterion for for assessing assessing whether whether monetary policy has has been been easy easy or or tight: tight: 0 high interest rate was interpreted as indicating a tight monetary licy while aa low interest rate was interpreted ,licy interpreted as indicating an an easy easy netary policy. ,netary By By now it is well recognized that during inflationary ?riods it is vital to draw aa distinction between sriods between nominal and real inflationary periods periods the ates of interest and, as aa result, during inflationary ate of of interest may provide aa very misleading interpretation interpretation of the tance of of monetary policy. The same same logic logic applies applies with respect to the nalysis of the the relationship relationship between exchange exchange rates and interest rates. The foregoing analysis also provides provides the the explanation explanation for the obThe ob— ervation (which was noted previously) that generally there is is aa ositive ositive correlation between the forward premium on on foreign exchange exchange nd the level level of of the spot rate. rate. Since Since the spot rate is is expected expected to to be ositively ositively correlated with interest rate differential and since, ccording to to the interest parity theory, that differential must equal on foreign exchange, it follows that the forward he forward premium on remium is also expected to be be positively correlated correlated with the level level of he spot rate.~ rate. 11 That positive positive correlation may may also be be rationalized by currencies which are expected to oting that currencies to depreciate are traded at at a 11 ~For For evidence on the robustness robustness of the interest interest parity relation— relationhip, Frenkel and Levich (1977). The positive association between hip, see Frenkel he spot exchange rate and the the forward premium premium has has been interpreted interpreted in in erms of an explicit posiems explicit monetary model. It is noteworthy that this posiive association would be ex.ive be predicted predicted by any model in which which current current exdepreciahange rate rate reflects immediately the expectations of of future depreciaion. See, for example, Mussa (1976a) and Frenkel .ion. Frenkel and Mussa (1980). (1980). ince the rate ,ince rate of interest and and the exchange exchange rate are dimensionally ncommensurate, their their association raise questions questions that that are familiar familiar rom the discussions of the Gibson ·rom Gibson Paradox. In aa separate paper, paper, II ininend to examine the relationship between exchange rates and the .end the forward remium interest differenti differential) the various ,remi um ((or or the the interest a 1) in in 1light i ght of of the various exolanaexo 1anaiions ons of of the the Gibson Gibson Paradox. Paradox. —233-233- discount in the the forward forward market market and, and, on on average, average, these these currencies currencies also also discount in command aa lower foreign exchange exchange value in the spot market. Prior Prior to proceeding with the empirical evidence on the relationrelationship between exchange rates and interest rates it might be be useful useful to highlight highlight some of the main main features of the monetary approach which are are reflected reflected in equations (11) and (14). First, First, these equations equations demondemon- strate the symmetric roles that are being played by the supplies of of domestic and foreign monies and the the demands for these monies. Since Since variables like like real incomes as as the demands for monies depend on real variables well as as on on other other real variables which underlie expectations rates real variables which underlie expectations and and rates well of of interest, it is is clear that the monetary approach approach does does not not imply that the exchange rate depends only on the relative supplies of money; nor does exthat real variables do not affect the the equilibrium exdoes it imply that rate. change rate. Second, from the policy perspective perspective the monetary approach approach of the homogeneity homogeneity postulate: brings to the forefront the implications of ceteris paribus paribus aa rise in the quantity of of money money results in in an an equiproequiproportionate rise in in the the exchange exchange rate. rate. This illustrates the intimate intimate connection between connection between monetary policy and exchange exchange rate rate policy. Third, the the positive positive relationship between interest interest rates and exchange exchange rates and the the central central role played by by inflationary expectations expectations imply that policies of the currency which attempt to induce an appreciation of currency should aim at reducing inflationary expectations. expectations. The reduction in inflationar9 inflationary exex- pectations would halt the depreciation of pectations of the currency currency in in terms of goods goods and in terms of of foreign exchange, and and would result in lower nominal rates of interest while maintaining (or even raising) raisinq) real rates rates of of interest. -234-234- The discussion in the second section and, in particular, the conconThe ~ibutions l979a) and Dornbusch (1978) emphasized that "ibutions by by Mussa (1977, 1979a) ~e 1e predominant predominant cause of exchange rate movements is news news which which could )t it have been anticipated. It It was also argued in the second section iat the forward rate 1at rate seems to summarize the information that is availavail- ' le le to to the market when the forward rate is being set. set. We may may there— there- Dre express the spot rate at period t as aa function of factors which ire which ave been known in advance and are summarized by by the lagged lagged forward ate, as well as aa function of the “news.” :1.te, news. 11 11 15) in St = a+ a + bb tn in Fti + "news" “news” zn st= Ft-l + The empirical difficulty is in identifying the variable variable which easures easures the “news.’ news. 11 11 Assuming that asset markets clear clear relatively fast nd nd that the “news’ "news" is immediately reflected in in (unexpected) changes in in he rates of interest interest we may write equation equation (15) as 16) in St = a + b in Ft1 +~[(i - i*)t - Et1(i - i*)t] here the bracketed term denotes the innovation in the interest differdiffer— ntial Eti (i - i*)t ntial and where Et-l i*)t denotes the interest interest differential which - in period tt based on the information available available as expected to to prevail in t tt - 1. — The expected expected interest rate differential was computed from aa egression egression of the interest differential on a constant constant and on on two lagged 12 The previous analysis of the relationship alues of the differential.12 alues of the differential. The previous analysis of the relationship 12An alternative way An alternative to compute the the expected differential would se data on on the term structure structure of interest rates. Since data on the ifferential computaifferential of of 2—month 2-month rates are not readily readily available, this computaion would require interpolations. —235— -235- between interest rate differential and the exchange rate implies that between the coefficient cia is expected to to be be positive. Table 44 reports the OLS estimates of equation equation (16) (76) for the three exchange rates over the period June ,June 1973-July 1979. As may be seen, in all coefficients of the unexpected interest all cases cases the coefficients interest differential differential are positive and in most cases the coefficients coefficients are statistically signifipositive statistically significant. In order to to verify the importance of of using the series of innoinno- differential, Table 44 also reports reports estimates estimates of vations in the interest differential, which replace the innovations by the actual series of the regressions which interest differential as well as regressions which which include include both the innovation actual differential. innovation and the actual In all cases the coefficients coefficients of of the actual actual interest differential do not differ significantly from 1~ To allow for a simultaneous determination of interest rates zero.13 zero. To allow for a simultaneous determination of interest rates two—stage— and exchange exchange rates, rates, equation equation (16) was also also estimated estimated using aa two-stageleast—squares least-squares estimation procedure. These results results are reported reported in Table 5, 5, and again in all cases the coefficients of the unexpected interest differential are are positive. These These coefficients coefficients are highly sigsig- nificant in in the Dollar/Pound exchange exchange rate rate but insignificant insignificant in the nificant other two rates, rates. On the whole, the record shows that during during the 1970s l970s exchange rates and interest rate differential have have been associated associated positively and thus indicating that during that that inflationary inflationary period the the rise in same factors which induced a rise in the interest differential differential also inin- exchange rates. rates. duced aa rise in the spot exchange Furthermore, consistent with 13In order to check whether the In order to the dollar rescue policies of November 1978 1978 have had aa systematic systematic effect on the the estimates, these reregressions gressions were also estimated for for the period up to to September September 1978. 1978. The results did not change materially. -236- Table 4 interest Late Interest Rate Differentials and Exchange Rates Monthly Data: Data: June 1973 -— July 1979 Monthly June 1973 July 1979 (standard errors errors in parentheses) (standard in parentheses) Dependent Variable ,n S Constant in Ft_i in F t-l (1-i*)t (i_i*)~ t 2 R.2 D,W. D.W. .027 .95 1.73 .388 (,165) (.165) .026 .96 1.77 .546 (,199) (.199) .026 .026 .96 1.80 1.80 .029 •.79 79 2.11 2.11 ..377 377 (.132) .028 ,81 .81 2.22 2.22 .540 (.206) .028 .81 2.32 2.32 .032 .93 .93 2.02 2,02 .601 .601 (,271) (.271) ,031 .031 .94 2.08 2.08 .583 (.349) .031 .94 2.08 2.08 a,e, 8.e. f(i-i*)t-Et_ 1 (1-i*)t)} {(i_i*)~_E~i(i_i*)~)l Dollar /Pound Dollar/Pound .959 .017 (.096) .032 .032 (.019) (.019) (,025) (.025) .030 (. 017) (.017) ( .024) (.024) .019 .019 (.019) .968 (. 024) (.024) —.155 -.155 (, 111) (.111) —.335 -. 335 (.100) •.776 776 (.067) (.067) .184 (,125) (.125) —.301 -. 301 (.079) (.051) —.231 -. 231 (.104) .851 (.070) (.070) —.070 -.070 (. (. 43) .926 .926 .237 . 237 (,045) (.045) (.229) -.037 —.037 (,027) (.027) ,955 .955 (.032) -.040 —.040 (.046) (. 046) .952 (.047) ( .047) .961 .961 Dollar/Franc ' w "' "' N) U) —4 .801 .801 .195 (.188) (, 188) Dollar/OH Dollar/OM NOTE; NOTE: .024 .024 (.290) Interest one—month (annualized) Euromarket rates differential Et_ (1-i*)t Interest rates are the one-month rates. The expected interest interest rate differential 1 computed from from a regression of the interest differential on a constant and on lagged values of the was computed differential. (I —- i*)t i*) denotes denotes actual actual interest differential where where ii denotes the rate of interest interest on on securities different.ial. (i interest rate rate differential denotes the rate of securiti.es denominated in U.S. u.s. dollars and i* denotes the rate of interest securities denominated in foreign currency. d(;'1\0minated interest on securities currency, the unexpected interest rate differential. ((1-i*)t-Et-l(i-i*)tl denotes denotes the unexpected interest rate differential. Table 55 table lntna.t Bat. Oifferential sad Interest Rate Differential and takings Exchange Beta Rates Instrumental Variables Monthly Data: June Jia. 1973 —- July 1979; 1979; Istaseatal (standard errors in parentheses) Dependent Da~n~u~ Variable J!.n st R22 ft D,W. D.W. .027 .027 .95 1.69 .024 .024 .97 .97 1.79 1.79 .023 .97 1.78 .036 .69 2.18 2.18 .216 (.165) .029 .80 .80 2.22 2.22 .165 (,199) (.199) ,034 .034 ..73 73 2.21 .034 .034 .92 2.11 .034 .034 .92 2.07 .034 .92 .92 2.07 2.07 p.i(i_i*)t) a... l((i—i*)~—t (i-i*) t-Et-1 {i-i*) t) I1 s,e, Sn Ft-1 in (i..i*)~ (i-i*) t (.020) .966 (.026) -.153 —.153 (.117) .027 (, 016) (.016) .963 .965 .435 .43.5 (,022) (.022) (.164) (,164) Constant poller/Pound Oollar/!i?omi.ll .020 .020 .017 .017 (.017) (.017) Dollar /Franc Dollar/Franc Dollar/I}l1 Dollar/$t —.143 -.145 -·.142. —.142 ( ,098) (.098) ,')09 .909 -.301 —.301 (.234) (.125) (.085) (.085) —.279 -.279 (.081) (.081) ((.053) .053) —.184 -·.184 (.126) .883 ((.086) .. 086) —.260 "".260 (, 225) (.225) -.006 —.006 (.047) (.047) .987 .987 -.135 —.135 (.049) (.049) (.307) —.040 -.040 (. 031) (.031) .951 ((.036) .036) —.040 -.040 .931 .951 ( .055) (.055) (.053) BOUt NOTE: .971 .971 (.022) .826 .816 ,425 .425 (,160) (.160) .555 (. 380) (.380) —.001 -.001 ((.003) .003) .555 .555 (. 402) (.402) t are the the one—month one-month {annualized) Euromarket rates. The 1be expected expected interest interest rate differential differential IE (i-i*)) Interest rates rate. are (annualized) luroinrtst (i—i interest differential differential on a constant sad and or, on two lagged values of the §Ilferential. was computed from tree aa regression of of the interest the 5iher.ntL. Two-stage least square. squares astianion estimation method was used. instruments for the the interest interest dIfferential differential were a a content constant and Two-stage nsed. The imanants two lagged valves of the the differential imstr,aents for the unexpected differential two values of differential and the the instruments differential were a a constant constant and and Dufl.in’a Durbin's (1 —- iS) i*)f denotes denotes actual interest ate rate differential differential where where ii denotes denotes the rat, rate of interest on securities rank variable (i denominated in U.S. dollars do lars sad and j* i* denotes the the rate securities denominated in foreign foreign currency. currency, demoatmatad rat, of interest on securities (i..i*) rc-1”~~ ~1denotes demotes the iat.rest rate ((i-i*)t-Et-l(i-i*)t] the —e--nted unexpecte--4_ interest rate differential. differential. ' co M N ' the hypothesis that current changes in exchange rates are primarily a response to response to new information, information, the evidence evider;ce shows snows the the importance importance of of the the innovations Innovations in the interest differential. The principle that current exchange rates already reflect expecexpec- tations concerning the future course course of events implies that changes in in exchange rates are are primarily due to innovations. In the present secsecapp"lied to the analysis of the relationship tion this principle was applied between exchange rates and interest rate differential. differential. however, is general. The principle, For example, it implies that the relationship between a deficit in the balance of trade and the exchange rate depends crucially on whether the deficit was expected or not. A deficit that was expected may have no effect on the exchange rate since the latter already reflected these expectations. In contrast, an unexpected deficit in the balance of trade may contain significant new infonnation information 4 that is likely to induce a strong strong effect on the exchange rate) rate. 14 EXCHANGE RATES AND PRICES One of the striking facts concerning the relationship between prices and exthange exchange rates during the 1970s is the extent to which the evolution of prices and exchange rates have not coincided. The origiorigi- nators and proponents proponents of the purchasing power parity doctrine (Wheatley and Ricardo during the first part of the 19th century and Cassel during the 1920s) have viewed the doctrine as an extension of the quantity 14 ‘4For aa further elaboration on the relationship between between exchange rates, rates, and and the current current account, account, see Dornbusch Dornbusch and Fischer (1978) and Rodriguez special emphasis Rodriguez (1978). (1978). For For aa special emphasis on the the role role of of innovations innovations in in the trade balance, see Nussa Mussa (l979c) (197gc) and for empirical evidence, see Hakkio (1979b). -239- theory of money to the open economy. By now now the consensus seems to be that that purchasing power parities can be expected to hold in in the long-run, lonq-run, if most of the shocks to the system are of aa monetary origin which which do if not require changes in relative prices. prices. To the extent that most of of the shocks “real” changes (like differential growth rates among shocks reflect "real" sectors), the required changes in sectoral relative prices prices may result in in a relatively loose connection connection between exchange exchange rates and aggregate price levels. The experience during the the 1970s 1970s illustrates the extent to to which real shocks shocks (oil embargo, supply shocks, commodity booms booms and shortages, differential productivity growth) shortages, growth) result in in systematic devideviations from purchasing power parities. As illustrated illustrated in in Figures Figures 2—4, 2-4, short-run short-run changes changes in exchange rates have not been closely linked to short-run short-run differentials differentials in the corresponding national national inflation inflation rates, rates, particularly particularly as as measured by consumer consumer price price indices. indices. loose link seems seems to be be cumulative. Furthermore, this As illustrated in in Figures Figures 5-7, As 5—7, divergences from purchasing power parities, parities, measured in terms of of the relationship between exchange rates and the ratio of consumer consumer price indices, seem to persist. The The loose link between prices and exchange rates is is illustrated illustrated in Table Table 66 which reports the the results results of regressions of changes chanoes in in the exchange exchange rates on changes in (wholesale) prices. As may be seen, for the Dollar/Pound and the Dollar/Franc Dollar/Franc exchange rate, the slope coefficoefficients are very close to to unity; for the Dollar/DM Dollar/OM exchange rate the slope coefficient coefficient is less close is less close to to unity. unity. slope Furthermore, in all cases Furthermore, in all cases the parameter estimates are extremely imprecise. imprecise. The results are even poorer when the wholesale price price indices are replaced by the cost cost of of living living indices. indices. It be noted, noted, however, extent this It should should be however, that that to to some some extent this -24D-240- Table Table 66 Relative Purchasing Purchasing Power Parity; Relative Parity; Instrumental Instrumental Variables Variables July 1979 Monthly Data: Data: June 1973 —- July 1979 (standard errors errors in parentheses) (standard parentheses) Dependent Variable )ependent Variable iinSt I:::. in st Dollar/Pound )ollar /Pound Constant Constant 6 tn(P /P*) /P*) w w .999 w s.c. s.e. D.W. D.W. .039 1.71 .003 (.005) (. 005) (.653) (. 653) Dollar/Franc )ollar/Franc —.001 -.001 (. 004) (.004) .891 (.682) (. 682) .030 2.38 Dollar/3M lollar/DM —.001 -.001 (.008) (.008) 1.313 1.313 (2.057) .036 .036 1.92 1.92 ~ote: iote: ~I:::. in /p*) denote, percentage change change in SSt and iI:::. Zn(P tn(P /P*) denote, respectively, respectively, the percentage Ln the spot spot exchange rate an’s in wholesale price e~change ratewan~ in the ratios ratios of the wholesale price indices. indices, ;.e. error of the regression. stage least-squares least-squares ;,e. is is the standard standard error ~Rgression. Two stage estimation constant, time, ~stimation method, is is used; the the instruments instruments are are a constant, time, time time quared, and lagged values >quared, values of the dependent dependent and independent independent variables. variables. —241— -241- phenomenon is specific to the 1970s. During the floating rates period of the 1920s, the doctrine of purchasing power parities parities seems to to have 5 been much reliable.’15 much more reliable. The discussion in the second second section section emphasized emphasized that that in periods The which are dominated dominated by “news,” news, which alters expectations, exchange exchange 11 11 rates (and other other asset prices) are are expected to be be highly volatile. volatile. Aggregate Aggregate price indices, indices, on the other other hand, are not not expected expected to reveal reveal such aa degree of volatility since they reflect the prices of goods and services which are less durable and, and, therefore, therefore, are likely to to be less sensitive sensitive to the news news which alters expectations concerning future course of of events. It follows, therefore, that in in periods during which there is amp ample fluctuations 1e “news” news which cause large 1 arge fl uctuat i ans in exchange rates, 16 there will also be large deviations from purchasing power parities.16 II II there will also be large deviations from purchasing power parities. The different degrees of volatility of prices and exchange rates are illustrated in in Table 7, which which reports reports the average absolute monthly perpercentage changes in the various exchange exchange rates and prices. As is evievi- dent, the the mean absolute change in the various spot exchange exchange rates has been about 22 percent per month (and even slightly slightly higher for the changes in the forward rate). The magnitudes of of these changes have been more more than double the magnitudes of the changes changes in most of of the various price indices, as well as in in the ratios of of national national price price levels. For For example, example, 15 For evidence see Frenkel (1976, 1978, 1978, 1980b) 198Db) and Krugman (1978). 16Dn this, see on this, Mussa (1979a). It is is noteworthy that the emphasis been on on the words large fluctuations; fluctuations; this should be in the text has been contrasted with periods during which which there are large secular changes changes in in the exchange rate (like the changes which occurred occurred duang during the German sten hyperinflation). During During such such periods the secular changes do do not stem necessarily from news news and need not be be associated with deviations from purchasing power power parities. purchasing parities. 11 11 -242- Table Table 77 Absolute Percentage Changes in in Prices Prices and Exchange Exchange Rates Mean Absolute Percentage changes Monthly Monthly Data: Data: June 1973 1973 -— July July 1979 1979 Variable Variable ritry n.try COL coL WPI k Stock Ma~t Market Exchange Rates Against the Dollar Dollar ~g~inst spot forward spot coL/c0L~~ COL/COLDS ‘~ . .009 .007 .037 — — . .014 .012 .066 .021 .021 .007 .007 rice nee .011 .009 .054 .020 .021 .003 many .004 .004 .030 .024 .024 .004 e: — All variables variables represent absolute values percentage represent the absolute values of monthly monthly percentage changes in in the data. data. denotes the wholesale wholesale price index and WPI denotes price index coL COL denotes denotes the cost cost of living living index. index. Data on prices prices and exchange rates IMF tape tape (May (May 1979 version).. rates are from the IKF 1979 version) The stock stock market market indices International Perspective, Perspective, monthly issues. indices are from capital Capital International monthly issues. -243- the mean monthly change change in the cost of living price index was .4 .4 perpercent in Germany, .7 percent in in the U.S., .9 .9 percent in in France and 1.2 percent in the U.K. These differences are even more more striking for the detrended series. exchange rates have been volatile is clearly clearly The notion that exchange illustrated by Figures 2—4 2-4 and by Table 7. 7. The comparison of the magmag- nitudes of the changes in in the exchange exchange rates with the magnitudes magnitudes of of the in the the ratios ratios of national price price levels changes in the price indices and in may suggest, according to aa narrow narrow interpretation interpretation of the purchasing power parity doctrine, that that exchange rate rate fluctuations have been “excessive.” excessive. 11 11 The The previous discussion, discussion, however, has emphasized that exex- change rates, being the relative prices of assets, are fundamentally different different from the price price indices of goods and services and, therefore, therefore, are expected to to exhibit aa different degree of volatility in in particular during during periods that that are are dominated by by “news.” "news." An An alternative alternative yardstick yardstick for measuring measuring the degree of exchange rate fluctuations would would be aa comcom- assets. parison with prices of other assets. while exchange exchange rate Indeed, while changes changes have been large relative to to changes changes in in national price levels, they have been been considerably considerably smaller than changes changes in the prices of other assets like like gold, silver, many other other commodities that that are traded in in organized organized markets, and common stocks. For For example, Table Table 77 also also rere- absolute monthly percentage change in stock market ininports the mean absolute dices. As may nay be seen, the mean monthly monthly change in these indices indices ranged from over 33 percent percent in in Germany to over 66 percent percent in in the U.K. By By these it is difficult to argue that exchange exchange rates have been exexstandards it cessively cessively volatile. -244- the characteristics of of exexThe fundamental difference between the also reflected reflected in in their their time change rates and national price levels is also series properties. The monthly changes in exchange rates exhibit little or no little no serial correlation while national price levels do do exhibit exhibit aa degree degree of serial serial correlation. The The serial correlation correlation of national national price levels has been rationalized in recent macroeconomic theorizing existence of nominal conin terms of costs costs of of price price adjustment, adjustment, the existence nominal contracts, confusion absolute prices tracts, confusion between between relative relative and and absolute prices and and confusion confusion changes. between permanent permanent and transitory changes. This This difference difference between the time series series properties of exchange rates and prices prices is reflected in the low correlation between the the practically random month-to-month exchange month—to—month exchange rate changes changes and the serially correlated differences between national rates of inflation. Given the short-run short—run deviations deviations from purchasing power parities, parities, it is relevant to explore explore whether whether these deviations tend to diminish with with time or tend to persist or even grow in size. In order to to examine examine the the patterns of the deviations, II have computed the autocorrelation funcfunc- tions and the partial autocorrelation functions of of these deviations for the wholesale and the cost of of living price indices. The deviation from purchasing power parities during month ttis is denoted by A 6 and is defined defined as: (17) At = An St - £n(PIP*)t. 12—14 illustrate the Figures 12-14 the patterns patterns of the the deviations for the three exchange rates. As may be seen, the the general pattern is very similar similar for the three exchange exchange rates and and for the two price indices. In all cases the autocorrelation autocorrelation function tails off at at what seems to be be an an —245— -245- Figure 12 The from PPP PPP with The Dollar/Pound: Dollar/Pound: Deviations Deviations from with Wholesale Wholesale Price Price Indices Indices .. 01,, ··•~" ' ~;_;•~ 'J:1~.1 •• , 4:9 .hH ..,. "·" ,.04 .. cir,co,~« ~wr,u,a ,,.,, ' ',... ' ~. 02 ~ - - - -- _ ---~ E~1, SIO <asoa ~J’hr~__~ ___ Jc• ~O• “S S. S~O.SS · ·~· ,.,.•c•""•~• 0.00 0.50 .0.0 ------~--- ~: -- --. -—~——‘ - ‘--- • — --~----—:——~~-~.—:—--—---— —~-—--‘ . -=111=-·-~:- ~.-;...,·:,no;;;;,_.,,o~ ,~~<,,o~- •-,,. ,~. ""'"' The Deviations from cost of Indices The Dollar/Pound: Dollar/Pound: Deviations from F?? PPP with with Cost of Living Liv~ng Indices ... """"'~<< ~°~“~Ek~ 5005 0000C0S.CL.,00SS :~.il:N~49:Nf~,1~,~H9i~J~H !l~~;_ '~:i ...,l-o, «o ,a~o• ,ck •o• u,. sto ,u~• O!H(O<M< -.- ···0-.--- 11••> £50. «1. o-• , — '""""•--- - ~fl.StSncoSS(S.flSSS ,_ . ,• ------- -"••-~-- ..:___ "•""''- --.:;-,o,o;.,·~.-;,o~-,;;;;,;;;;;;-o;:--,·~; "~"" · -246-246- -- Figure Figure 13 13 The from PPP Pt’? with Dollar/Franc: Deviations Deviations from with Wholesale Wholesale Price Price Index Index The Dollar/Franc: ... ... O"rf~H•C< uo•i •~:~d-o~ 2:0 2:22 t:F. 0.05 0.05 0.0’ 2:12 tO 2,.I0.0.22.0.40.0.13.0.35—l.32.0.1e.0. 2:22 1:11—2:21-8:11 —1:81 .o.52.0.021,1l 0.22 0. 50’, OIHUE•C[ 000 O•O, II•& ) £I—0I’I~s0 ~02 0.69 - 2 .0.02—0.02 00c051cs,705,,Fosc110s 10 2 — -. o,56.50510, 0—0 - 5.2 ~ -— 500 50 ~0S ‘05 Sb 0.02 - 0,0 — - 0.5.C0-0.l2.o.t0.0.t3 0.0 —~ —— 0.0 0.0 -— ——---: —_ — The Deviations from P1’? with of Living Living Indices The Dollar/Franc: Dollar/Franc: Deviations from PPP with cost Cost of Indices ,.,.,~,"".. ~,'"~"" 01"("'"" 0—0) • 00 0—~1 ~ "·~ •> "O"•• " Q5(•U I ••• -·. ·--·o ----.; o·-""" —1:D—188.1:i1—1:1 .8:44 .1:1: :1:81 :1:14 0.03 Ol'H•••u .,.~, .8:81 4:4? .1:84 5005 ~•,s ) 0.00 0.02 I .155.151.55 0.22 0.12 2,0Ij.s105100000s.TI0s5 --—- 1.21 5.1’ -- 050. 500 — - >•12 ~--- £0’0c0l,,0LbT0c05l1,c5lls100,E,0~it5 •utoi:ol!,iC,, ,o,. ,u.,;1 ,o,. o, ,~, " " ' " - -----§: -------··--- ·-·--,· ... ·····. . , ,0 •O,O •O•• •Ool ~.O ···. O,< O.• ....... O•• •-••·· ........ ------OoO ) ,0 ,. ------lt·-~-------- )g- -·- ______ -~~::r--- "· ". wit_______~-- ------ ........... • ••,r:u;:-.ut.o••·L•110k--,,Y.~i,o, o, ,..,",,.-,,-.- - . : —247-247- I — Figure 14 14 Figure The Dollar/DM: W71olesale Price Price Indices Dollar/OM: Deviations from PPP with Wholesale o,,.n,~« • LOSS 00 .---.-. DlNOOU ,, •• , <1•8 ,000C000,soOlO’-S 1:1; 1:14 1:U .1:1? .5:42 - ‘~.27 —0.~ _5.1S’5.1S~S. ~;.;•~ 'tt~oi-o, lb—Il 5250 ‘‘02 ) 5.~0’0.2a.I.l0 2.02 1:51 1:11 1:11 .1:81 —1:41 .1:18 —1:18 :U~ .l0’0.15’I.2O’5.29 ‘1.10.1,21 2lO0I0S0lT0cl0l_5l_~5l 5005 ~0.22 5.o~ 0.2 20.20 55’. 500 ,.,. "" 00500 ,.~,,. '"" ""· 1:14 tU 1.50 —0.12.0.05 ~5.O0 —- ,~i.o• __ B l , 510 550. ___ f ~a a O ~ - - - 0.02 0.2* o.ti -i:··-· -•-·--··--·•-·---·•-···---:·-·-·-·--':"-------·-·!: - --·-·- - - - ·-·-------·-·.----·-· 5055! 005 5525(051005 ·t : . _!!: --·--- --- ):.' :a~- -- -11- -~=~ C· ·,= ~- - I -~-~ if: ·: - --—- - -: " ------~i _______ ~- - - - - - - - - - - The Dollar/DM: Dollar/OM: Deviations Deviations iron from F?? PPP with cost Cost of of Living Living Indices . 01u;oo<f Loss ., Hoa~ '&:i,.l-o. - . ... 0000Cl0IlL00lL~0 .1i~.9ii~t~ 1 05’ _~9 ~ d21s0o1L15:__~_ :~:~:~il O!r,fO<M<:E - n-•1 ll-8 , —— ------;?--•·:.:•···••··=~·-•---••·---••;·•·•···•-·:·-·H••·"•'.•·•·· ·•·•• ·•· i - it· - - — '•"---••---• — — ----- ,,. ~ ______ ,s:_ _ _ _ ________,__;,;~-;·;~;:-~;;;-;;:;;;~:~•7,o~ ,;;;;1,,o~ •-·~..;,-.,.,., o .o••. •••· •0,0 •O•• O,< 0,0 O,< 0,, —~ 0,0 0,. ·--··········-····---·········· ·•···••··············•· ··················•-*•··-··········-~ - - - - - - -248—248- funcexponential rate and, in all cases, the partial autocorrelation function shows aa spike at the first lag. pattern seems to indicate indicate This pattern night have been expected on the basis of the time series properties (as might of arid price indices) purof exchange exchange rates and indices) that the the deviations from pur- autoregressive process, process. chasing power parities follow aa first order autoregressive It is noteworthy, however, that that in all cases the value of the autoregresautoreqres- sion tern term is about 0.9, indicating the possibility possibility that the series may not satisfy the stationarity requirement requirement.0 To allow for this possibilthis possibil- ity, II have also examined the autocorrelation functions and the partial autocorrelation functions of of6t -6t-l', i.e., i.e., of the first difference - of the deviations from purchasing power parities. pa ri ti es. The results indicate indicate that these differences are serially uncorrelated, uncorrelated, thus thus implying that that 7 In view of this 17 the the deviations 6t follow aa random walk walk process) process. of this possibility, II conclude conclude that the the deviations deviations from purchasing power power parities seem seem to follow aa first order order autoregressive process but but that the data do not provide sufficient evidence evidence to to reject the alternative alternative hypothesis of aa random walk. Finally, it it may be be noted that the maim main difference between accepting the AR(l) AR(l) rather than the random random walk hypothesis relates to the economic interpretation of of the two alternaalternative processes. implies that deviations from The random walk process implies purchasing power parities do not tend to diminish with the passage of of time time while the stable AR(l) process implies that there are mechanisms mechanisms which operate to ensure that in the long—run long-run purchasing purchasing power power parities 17 the 17 If 1f the not entail (ex of equilibrium equilibrium deviations follow a random walk process, then they do deviations follow a random walk process, then they do ante) unexploited profit opportunities opportunities. For an analysis 0 For deviations from purchasing power parities, see Saidi (1977). -249- are satisfied. For For the the purpose of forecasting the near near future, future, how- between using the AR(l) process ever, there is aa very little difference between with an autoregressive coefficient of 0.9 and using the random walk process. - CONCLUDING REMARKS CONCLUDING In this paper I examined some aspects of the operation of flexiflexiWe ble exchange rates. The analysis was based on the experience of the lglOs. 1970s. The principle conclusions which may be be drawn from the empirical work are: (1) In spite of the extraordinary turbulence in the markets for exchange1 it seems that to aa large extent the markets foreign exchange, have operated efficiently. that in It should be noted, however, that the concept of “efficiency” "efficiency" is somewhat narrow in this context the that it only refers to the notion that the markets do not not seem enta i 1 unexploited profit opportunities. AA broader perperto entail spective should deal with 1<ith the social cost of volatility in terms of the interference with the efficiency of the price tenns system in guiding resource allocation, as well as with the cost of alternative outlets for for the disturbances that are currently reflected reflected in the volatility of exchange rates. (2) The high high volatility of exchange rates (spot and forward) rereflect an intrinsic characteristic of the relative price of monies and other assets. The price of gold and the price of stocks as well as exchange rates between national monies depend critically on expectations concerning future course of events, and adjust rapidly in response to new infonnation. information. In this —250— -250- perspective the exchange rate (in contrast with the relative price of national outputs) is being viewed as as aa financial variable. (3) During During inflationary periods periods variations in nominal rates of interest are dominated dominated by changes in inflationary inflationary expectations; as aa result, high expected to be high nominal rates of interest are expected be associated with elith high exchange exchange rates (a (a depreciated currency). This relationship was demonstrated demonstrated within the analytical frameframework work of of the monetary monetary approach to the exchange exchange rate, and was supported supported by by the empirical work. In this this context the key finding was the dependence of exchange rate changes changes on the interest. changes in the rates of interest. in accord This finding is in current exchange exchange rates with the analytical prediction that current already reflect current expectations about the future while primarily reflect reflect changes in in the current exchange rates primarily changes in these expectations which, by by definition, arise from new information. (4) The experience of the the 1970s does does not support the predictions of power parity doctrine doctrine of the simple version of the purchasing power which relates the values of current current prices to current exchange exchange rates. The empirical empirical work showed that deviations from purpurpower parities can can be characterized characterized by aa first order chasing power autoregressive process. One of of the key analytical analytical insights insights that that is is provided provided by by One the key the the monetary (or the asset asset market) market) approach to the exchange exchange rate circumrate is that exchange rates reflect not not only only current circumstances those circumstances circumstances which stances but but also also those which are are expected expected to to -251— -251- exprevail in the future. This anticipatory feature of the ex197gb) does not change rate (which is emphasized emphasized by by Mussa, 1979b) to such such aa degree) the prices of national characterize (at least to outputs. outputs. As a result, during during periods periods which which are dominated dominated by frequent changes in expectations about about the future, one may exexdeviations from purchasing purchasing power power parities pect to find frequent deviations 18 when the the latter latter are computed using current prices.18 when recognized by Gustav Cassel 18This phenomenon phenomenon was was recognized by Gustav Cassel -- the the most most recognized proponent of of the purchasing power power parity doctrine. Since this paper was prepared for presentation presentation on October October 20, 1979 -- the the date of Cassel ‘s birthday (Cassel was born on Cassel 's on October 20, 20, 1866) it it seems appropriate to to conclude conclude with the the quote that reflects this key idea. —- —- The international valuation of the currency will, then qenaenerally show aa tendency to anticipate events, events, so so to speak, speak, and become more an an expression of the the internal value that that the currency is expected expected to possess in aa few months, or perhaps in aa year’s year's time (Cassel, 1930, pp. 149-50). -252—252— DATA DATA APPENDIX APPENDIX 1. Exchange Rates exchamge rates are end-of-month end-of—month rates obtained from the The spot spot exchange IMF version, updated to July July 1979 using the TMF tape (May 1979 version, November November 1979 issue of the International International Financial Statistics) Statistics) obtained from the International Monetary Fund. The The forward exexchange rates are one month are end—of—month end-of-month rates rates for for one month maturity. maturity. The The change rates forward rates for the U.K. Pound and the DM for the period June 1973 - June 1978 1978 are bid prices obtained from the International International (1MM). For Money Market Market (IMM). For the period period July 1978 - July 1979 they are sell prices obtained from the Wall Street Street Journal. The The forward rates rates for for the the French French Franc Franc for for the the period June June 1973— 1973July 1974 are bid July bid prices calculated from the Weekly Weekly Review publication of the the Harris Bank Bank which which reports the spot rate and the the forward premium; in each case the closest closest Friday to to the end end of the month was chosen. For For the period August August 1974 1974 - June June 1978 1978 the the rates are bid rates rates obtained from the 1MM IMM and for the period period July 1978 - July 1979 1979 they are sell sell prices obtained from the Wall Wall St refil_J.p ur11<1lSSI?ItlolAlmaI. — — - — 2. Prices The wholesale wholesale and and cost cost of living price indices are period averages obtained from the the IMF IMF tape, lines 63 63 and 64, respectively. 3. Rates of Interest 1—month Eurocurrency rates obtained All interest rates are 1-month obtained from the Weekly the Harris all cases figures the Weekly Review Review of of the Harris Bank. Bank. In In all cases the the figures used correspond to the last last Friday of each month. month. 4. Stock Markets The stock market market indices correspond correspond to to the last trading day of of the month. The sources are Capital International Perspective. Perspective, Geneva, Geneva, Switzerland, monthly issues. issues. —253— -253- REFERENCES KEFERENCES “Fixed and Flexible Artus, Jacque R. R. and John H. Young, "Fixed Fl exi b1 e Exchange Exchange Rates: AA Renewal of of the Debate,” Debate," National Bureau of Economic Economic Research, Working Paper Paper Series, no. 367, 367, 1979. Rational Expectations Rate.” In Bilson, John John F. 0.,” O., "Rational Expectations and the Exchange Exchange Rate." 0. Johnson (eds.), J. A. Frenkel Frenkel and H. G. (eds.), ]c~9!i~1ofExciiafl9~ The Economics of Exchange Rates: Selected Studies. Reading, 1978.. Reading, Mass: Addison—Wesley, Addison-Wesley, 1978 'Speculative Efficiency' Hypothesis." ‘Speculative Efficiency’ Hypothesis.” -________.,“The - -nanuscript, - -. , "The University manuscript, University of Chicago, 1979. 1979. Unpublished Cassel , Gustav. Money Cassel, Money and Foreign Exchange After 1919. London: London: 1930. Macmillan, 1930. “Exchange Rates, Clements, Kenneth W. and and Jacob A. A. Frenkel, "Exchange Rates, Money and Relative Prices: l920s.” Journal of Prices: The Dollar—Pound Doliar-Pound in in the 1920s." of International Economics, 10, 10, 1980, forthcoming. “Capital Mobility, Flexible Exchange Rates Dornbusch, Rudiger (l976a) (1976a) "Capital and Macroeconomic Equilibrium.” Equilibrium." In E. E. Classen Classen and P. Salin (eds. (eds.), Recent Issues in International Monetary Economics. Amsterdam: Amsterdam: North-Holland, 1976, Pp. 29-48. 29-48 . North—Holland, of Flexible Flexible Exchange Exchange Rate Regimes Regimes and Theory of _________ “The Theory - ~ ~ -. (1976b) "The Macroeconomic Policy.” Scandinavian Journal of Macroeconomic Policy." of Economics 78, No.2 (May 1976): 1976): 255—75. H. 0. 255-75. Reprinted in J. J. A. Frenkel and H. G. Johnson Johnson (eds.), The Economics Economics of of Exchange Rates: Selected Selected Studies. Reading, Mass.: Addison-Wesley, 1978. 1978 . __________ Exchange Rate Flexibility,” Flexibility," in . , “Monetary "Monetary Policy Under Exchange Managed Exchange—Rate --~Mra-n~aged Exchange-Rate Flexibility: The The Recent Experience, Federal Federal 90— Reserve Bank of of Boston Conference Series, No. No. 20, 20, 1978, 1978, Pp. 90122. Dornbusch, Rudiger Rudiger and Stanley Fischer, “Exchange "Exchange Rates and the Current Dornbusch, Account,” Account," unpublished unpublished manuscript, Massachusetts Institute of Technology, 1978. 1978. Edwards, Determi"A Simple Monetary Monetary Model of of Exchange Rate DetermiEdwards, Sebastian, “A nation in the Short Run Run - Some Preliminary Results for the Peruvian 1950—1954.” unpublished manuscript, University Peru vi an Experience 1950-1954." of of Chicago, 1979. — Frankel, Jeffrey. “On "On the Mark: The Theory of Floating Exchange Rates Rates Differential.” American Economic Review, Based on Real Interest Differential." 610—22. 69, No. 44 (September, 1979): 610-22. -254- Frenkel, Jacob A. “A "A Monetary Approach to the Exchange Rate: Doctrinal Aspects and Empirical Evidence.” Evidence." Scandinavian Journal .Journal of EcoEconomics 78, 2O0—2~4. Reprinted 78, No. 2 (May (May 1976): 1976): 200-24. Reprinted in J. A. A. Frenkel and H. 0. G. Johnson (eds.), (eds.), The Economics of Exchange Exchange Rates: Selected Studies. Reading, Mass.: Addison—Wesley, Addison-Wesley, 1978. 1978 . _________ The Forward . , "The Forward Exchange Exchange Rate, Expectations and the Demand ---,f"o_r_'Money: Hyperinflation." American Economic Review for Money: The German Hyperinflation.” 653—70.. 64, No. 44 (September (September 1977): 1977): 653-70 . , "Purchasing Purchasing Power Parity: Doctrinal Perspective and EviEvi--~d-e-nc-e of International dence from the 1920s." l920s.” Journal of International Economics 8, No. 22 (May 1978): 169-91. 169-9 . _________ Demand for Money - - ~ - ~ ·.,• “Further "Further Evidence on Expectations and the 0emand during the German Hyperinflation.” Hyperi nfl ati on." Journal of Monetary Economics, 5, No. l1 (February, 1979): 81-96. 81—96. (l980a) “The (1980a) "The Forward Premium on Foreign Exchange and CurCurrency Depreciation During the German Hyperinflation,” Hyperinflation," American American Economic Review 67, 1980, forthcoming. __________ (l980b) “Exchange (1980b) "Exchange Rates, Prices and Money: Lessons Lessons from the l920s’ American Economic Review, 20, 2 (May, 1980), forthcoming. 1920s" forthcoming. __________ Frenkel Frenkel,, Jacob A. and Kenneth W. Clements, “Exchange "Exchange Rates in the l920s: AA Monetary Approach” 1920s: Approach" in June M. Flanders Flanders and Assaf Assaf Razin Development in an Inflationary World, New York: Academic (eds.), 0evelopment Press, 1980, forthcoming. Frenkel 0. Johnson (eds.), The Economics of Frenkel,, Jacob A. and Harry G. of Exchange Rates: Addison—Wesley, 1978. Rates: Selected Studies. Reading, Mass.: Addison-Wesley, 1978. Frenkel, Jacob A. and Richard M. Levich, Levi ch, “Transaction "Transaction Costs and Interest Arbitrage: Arbitrage: Tranquil Versus Turbulent Periods.” Periods." Journal (December, 1977): 1209—1226. of Political Economy 85, 85, No. 6 (0ecember, 1209-1226. “Efficiency of Foreign Exchange Frenkel, Jacob A. and Michael L. Mussa, "Efficiency Markets and Measures of Turbulence,” Turbulence ,t' American Economic Review, 70, 2 (May, 1980) forthcoming. Hakkio, Craig S. (1979a) “Expectations "Expectations and the Foreign Exchange Market.” Market." Unpublished Ph. D. University of Chicago, 1979. 1979. 0. Dissertation, University __________ (l979b) "Exchange “Exchange Rate and Innovations in the Current - ~ - - (1979b) Account,” Account," Unpublished manuscript, Northwestern University, 1979. 1979. ‘Predictors of Hansen, Lars P. and Robert J. Hodrick, "Predictors of Future Spot Spot Rates: An Economic Analysis,” Analysis," Journal Jou ma 1 of Political Po 1 i ti cal Economy 88, 4 1980) forthcoming. (August, 1980) Hausman, Jerry A. “Specification Speci fi cation Tests in Econometrics.” Econometrics. Econometrica, 11 H 46 46 (Novermber, ( Novermber, 1978): 1251—71. 1251- 71. —255— -255- Hodrick, Robert J. J. “An "An Empirical Analysis Analysis of the Monetary Monetary Approach to Rate.” In the Determination of the Exchange Exchanqe Rate." In J. J. A. Frenkel and H. G. Exchan e Rates: Selected Studies. Johnson (eds.), (eds.), The Economics of Exchange Reading, Mass.: Addison-Wesley, ison- es ey, 1918. ‘The Exchange Rate and the Balance of Payments in Kouri, Pentti, J. S. "The in the Short Run and in the Long Approach.” Long Run: AA Monetary Approach." Scandinavian Journal of Economics 78, No. 22 (May 1976): 280-304. Krugman, Paul. “The Efficiency of the Forward Exchange Market: Evidence Paul. "The from the Twenties and the Seventies." Seventies.” Unpublished manuscript, Yale University, 1977. Purchasing Power ---.=cc·, !!Purchasing Power Parity and Exchange Rates.” Rates. Journal Journal of _________ 11 International Economics 8 (August, 1978): 397-407. 397—407. “Further Results on the Efficiency Levich, Richard M. "Further Efficiency of Markets for Foreign Exchange" Exchange” In Mama ed Exchan Managed Exchangee Rate Flexibilit Flexibility:: The ç~p~iem~,Federal Experience, Federal Reserve Ban Bank of Boston, Boston, Conference Recent Series, No. 20, 1978 1978.. . , "The Efficiency Efficiency of Markets for Foreign Foreign Exchange," ~.,“The Exchange,” In --~R'u'd~iger A. Frenkel (eds.), International — Rudiger Dornbusch and Jacob A. Economic Polic Policy:: Theor Theory and Evidence. Ba1timoriTJNi~i~Tlb~ldns Baltimore: Johns Hopkins University 79. University Press, 11979. McKinnon, Ronald I. “Floating "Floating Foreign Exchange Rates 1973—74: 1973-74: The Emperor’s New Clothes." Clothes.” In Karl Brunner and Allan Meltzer (eds.), Emperor's ~Institutional Arrangements and the Inflation Problem, Vol. Vol 3 of the Carnegie—Rochester Carnegie-Rochester Conference Series on Public Policy, A A Monetary Economics, 1976, Supplementary Series to the Journal ~ 1 offMo et~yEconomics, 1 pp. 79-114. . Mudd, Douglas, R. “Do "Do Rising U.S. Interest Rates Imply aa Stronger Dollar?” Dollar?" Federal Reserve Bank of St. Louis Review, 61, No. 66 (June, 1979): 9—13. 9-13. Mussa, Michael, (l976a) “The Exchange Rate, the Balance of (1976a) "The of Payments and Monetary and Fiscal Fiscal Policy under a Regime of Controlled Floating.” Floating." Scandinavian Scandinavian Journal of Economics 78, No. 22 (May 1976): 1976): 229-48. Reprinted in J. A. Frenkel and H. G. Johnson (eds.) (eds.),, The Economics ~ of Exchange Rates: Selected Studies. Reading, Mass.: Addison— AddisonWesley, 1978. Experience with with Fixed and Flexible (1976b) "Our “Our Recent Experience ---~change Rates: A Comment." In Karl Brunner Brunner and Allan Allan Meltzer A Comment.” ExEx(eds.), Institutional Arrenagements and the Inflation Problem, Vol. 3 ~ Vol. 3 of of the Carnegie—Rochester Carnegie-Rochester Conference Series on Public Policy, aa -ial of Mpnetaryjçonomics, supplementary series to the ~p Journal Monetary Economics, 1976, pp. 123-41. __________ 11 Uncertainty: Causes, Consequences, and Exchange Rate Uncertainty: Policy Implications.” Implications." Unpublished manuscript, University of 1977. Chicago, 1977. —256— -256- - ______ -~~ • , 7 Mussa, Michael, (1979a) “Empirical "Empirical Regularities Regularities in in the the Behavior of of ExExchange Rates and Theories of the Foreign Exchange Market.” Market." Vol. 11 11 of the the Carnegie-Rochester Carnegie—Rochester Conference Series on Public Policy, a supplementary ~ supplementary series to the Journal of Monetary 9-57 . Economics, 1979, Pp. 9—57. . (197gb) (1979b) "Anticipatory __________ “Anticipatory Adjustment of aa Floating Exchange 7 --~R'a t-e." Unpublished manuscript, University of Chicago, 1979. 1979. Rate.” - _________ - ~ ~ - · (l979c) (1979c) “The "The Role Role of the Trade Balance in Exchange Rate Dynamics,” 1979. Dynamics," Unpublished Manuscript, 1979. “Expectations and Efficiency in Obstfeld, Maurice. "Expectations in the Foreign Exchange Market,” unpublished paper, Massachusetts Market," Massachusetts Institute of Technology, Technology, 1978. 1978. "The Role of Trade Trade Flows Flows in Exchange Rate DeterDeterRodriguez, Carlos A. “The A Rational Expectations Approach,” mination: A Approach," Unpublished manumanuscript, Columbia University, 1978. 1978. Saidi, Nasser, Nasser, “Rational "Rational Expectations, Purchasing Power Parity and the Business Cycle” Cycle" Unpublished manuscript, University University of Chicago, Chicaqo, 1977. Shiller, Robert J., J Do Stock Prices Move Too Much to be Justified by "Do Subsequent Dividends?” Unpublished manuscript, Subsequent Changes in Dividends?" University of Pennsylvania, Pennsylvania, 1979. 1979. .,“ -257—257— INTERNATIONAL STABILIZATION POLICY UNDER FLEXIBLE EXCHANGE RATES H. Heller H. Robert He 11 er Being the only speaker speaker at this conference to represent represent the busibusiness sector, II will focus my remarks on the effects of the flexible exexchange change rate system -- as as it it has operated throughout the seventies seventies -- on on -— the business business sector. the topic: topic: —- In particular, II will will discuss three aspects of of First of all, all, I will address myself to to the question of exchange rate system and its actual operation in whether the flexible exchange the economy, and in particular, particular, the the years since 1971 have served the economy, business sector well. Second, II will offer offer my views as to to what policy policy changes operation of of the the present present system. system. would improve improve the the operation changes would Third, Third, II will adopt aa longer perspective perspective and indicate what intermational international monemone- tary reforms might improve improve the the operation operation of the the system. THE CONSEQUENCES CONSEQUENCES OF FLEXIBLE EXCHANGE THE OF FLEXIBLE EXCHANGE RATES RATES The operation of 1971 has of the flexible exchange rate rate system since 1971 has entailed a significant significant increase in costs to to the business business sector. In particular, there are adverse effects on international trade, trade, internainterna- tional capital movements, and and foreign foreign investment. investment. II will also argue that the the increased costs costs to the private sector were not offset offset by aa greater freedom for the policymakers to macroto pursue more more appropriate appropriate macroeconomic stabilization policies or other direct savings realized by by the public sector. Dr. is Vice President for for International Bank of Dr. Heller Heller is Vice President International Economics, Economics, Bank of America, NT NT++ SA, San San Francisco, California. California. -258- But II should like to make it clear clear at the outset outset that there are But at present no no viable alternatives to to the the flexible exchange exchange rate system. As long long as there are large differences in inflation inflation rates among nations, a fixed exchange exchange rate system will not not be viable. What What we perceive as the cost of flexible exchange rates is therefore truly the cost cost assoasso- ciated with high and differential inflation rates. Nevertheless, the nonflexible exchange rate system does little to to make make countries adopt adopt non- inflationary policies. It is in that sense that the flexible exchange rate system also been associated with fluctuating exchange rates systen has also and the costs thereof. International_Trade International Trade The thesis thesis has has been been advanced that flexible exchange exchange rates rates disdis- courage foreign trade. trade. There There are several reasons reasons for expecting aa dampdamp- ening effect on foreign trade under under aa system of flexible exchange rates. First First of of all, there there is simply simply the the increased increased uncertainty of of exex- will have to be be borne by one or the other other change rate fluctuation that will party to to trade transactions. It is important to to note note that we are are not not involved game here. here. involved in in aa zero—sum zero-sum game While in in simple arithmetic arithmetic terms one party’s gain must party’s loss, party's must be the the other party's loss, the increased uncertainty will affect both both parties parties to to the transaction. transaction. As long as people are risk-averse, there will be aa net assomet loss because the welfare losses associated with aa 50/50 chance of losing one million dollars are greater than the welfare gains gains of of aa 50/50 chance of of winning winning one million dollars. That’s the reason why we find corporate presidents wagering That's find few corporate wagering last quarter’s corporate earnings on aa double or nothing bet on quarter's on the the outcome weekend’s football game or at at the roulette wheels in Las Vegas. of this weekend's —259— -259- Second, while it is possible to engage engage in in forward currency trans— transactions to eliminate the foreign exchange risk, one one should keep in in mind that there is not only aa brokerage cost associated with these transactransactions, but that there exist exist no no organized forward markets markets for the vast majority of of currencies currencies -- especially especially those of the developing developing countries. —— The system system therefore has has an an inherent bias bias against trade with the develdeveloping countries -- some of which have have the widest exchange rate fluctuafluctua-— tions due to their high inflation rates. rates. Third, Third, there is the natural competitive instinct that that makes makes the “maybe II should businessman think -- "maybe should wait one more more week before II cover cover —- in the forward market market to obtain aa better rate. Or worse yet, if I obtains aa better forward rate next cover this week, and my competitor obtains week, then II will lose the contract contract altogether.” altogether." If the rate turns in aa If the businessman may may then then not even bid on on the disadvantageous direction, the contract. Fourth, there are costs costs for the individual firm associated with the necessity to collect information on exchange rates, to ensure that proper accounting accounting and legal legal procedures procedures are followed, to maintain maintain staff to call up the banks, make make appropriate calculations, keep records, and perhaps even hire an economist or consulting firm to to prepare aa foreign exchange forecast. All All these costs are deadWeight deadweight losses losses to the private sector as as aa whole, because we are playing a zero zero sun sum game where where one firm’s firm 1 s foreign exchange gains will will be another firm’s firm's foreign exchange losses. Gone old days when one merely merely had to outwit outwit the central bank are the good old mind as to to how much longer it should attempt attempt that could not make up its mind had long long ago become unrealistic. to maintain an exchange rate that had -260—260— Instead, the the private private sector sector has has to to maintain maintain all all the the required required ancillary ancillary Instead, services just in an attempt to stay even and not not to wind wind up on on the the losing side of currency seesaw, of the c1rrency seesaw. United States businessmen are particularly particularly affected by the introintro- duction of flexible exchange rate because most international trade used used to to be denominated denominated in in U.S. dollars. Now, Now, only half of of world trade is denominated in dollars and half in in other currencies. currencies. In particular, firms entering entering new markets often find that they have have to adapt to local local conditions conditions if they wish to penetrate new new marmarkets. kets. National pride of some of the newly developing countries countries may also play an important important role in their insistence to utilize utilize their own currency currency to to an an increasing extent. While While some empirical studies studies failed to find an effect of of flexible exchange volune of exchange rate on on the volume of international trade, II find this evievi- dence hard to to believe. The simplest simplest of all possible possible calculations show that foreign trade increased at an annual rate of 8.8 percent in real terms during 1963- 73, while while the rate of increase in 1973-75 amounted amounted to only only 4.3 percent per annun. annum. That of increase in the volume of That is, is, the rate of of internationinternation- al trade was cut in in half under the flexible exchange rate system. While it it is true that this this does not imply that that the flexible exchange rate system caused this this decline—— decline.-- and II will have to say more on that that topic later -- it is certainly certainly true that the flexible exchange rate —— system did not prevent prevent the decline in the trade volume either. Finally, II need not point out excharge out that the myth that flexible exchange rate would always always balance balance our international trade is nothing but that -- a myth. -— People who who drew the the opposite opposite conclusion from textbooks textbooks in in —261— -261- international international economics economics forgot to read the fine print: print: namely that it was assumed that there were no no international capital movements. Only in such aa world can perfect purchasing power parity hold imhold and will will imports ports automatically be be balanced by exports of equal value. movements are very very much with with us, us, and In the real world, capital movements they are not so much determined by actual actual international price differendifferentials, but by the expectation of price level level changes at at some some future date. Capital Movements Movements Capital This brings me to the second second major point to be be covered: the imimpact of of flexible exchange exchange rate on international capital movements. First of all, all , it is clear by by now that international capital movemovements have not served served as as the great stabilizer stabilizer of exchange exchange rates that they are supposed to be. be. According to theory, well-informed well-informed private speculators will act to to stabilize stabilize the exchange rate, buying the currency currency when it is low and selling it it when when it is high. informed informed speculators? But who are these these well— well- The actors actors with the greatest amount of of expertise in the area, the large cormiercial commercial banks, are highly reluctant to to take open foreign exchange exchange positions. tion-oriented. tion-oriented. They They are trade—oriented, trade-oriented, not not speculaspecula- Much Much more money is to be be made made by by actively trading in earning a small small spread on each transaction than than by the market, and earning maintaining maintaining an an open open position and hoping for the best. The situation situation is not of aa grocery store owner, owner, who not unlike unlike the the one one of grocery store who makes makes his his money money buying buying and selling tomatoes, and not by hoping hoping to to make a killing killing in in the market market when the tomato crop crop in in Mexico goes sour and the price skyrockets. U.S. Treasury data data show that on on average average U.S. commercial commercial banks were —262— -262- holding less than $100 million in in open foreign exchange exchange positions. This This aggregate aggregate amount for all U.S. banks is not much much larger than some of the individual transactions foreign exchange traders are called called upon to execute. This leaves private corporations and and individuals as as the potential market stabilizers. take foreign exchange posiposiWhile corporations do take they are typically typically designed to to offset some some commercial transactransactions, they tions rather than as as aa deliberate deliberate attempt to take an an open open position. The corporate corporate treasurer who attempts to make aa career out of of realizing foreign exchange profits profits is aa rare, and and probably short-lived, breed. more properly properly be described described Instead, the typical corporate strategy can more as one one of foreign exchange exchange loss-avoidance loss-avoidance rather than of foreign exexchange speculation. The final group -- private individuals individuals -- is certainly certainly increasincreas—— ingly ingly active in in the market. market. -- They are probably probably more active in the oror- ganized non—bank (Internanon-bank foreign exchange exchange markets, such as as the 1MM IMM (International tional Monetary Market, aa division of the Chicago Mercantile Exchange) Exchange) than than in the commercial bank market. Prior Prior to the Herstatt Herstatt calamity, calamity, private speculators had access to the bank bank market market largely largely through small small banks. banks. Since 1974 most major banks have reduced the foreign exchange exchange lines made available to to the smaller smaller banks, banks, thereby sharply sharply limiting their access access to the interbank market. Consequently, most Consequently, individuals are active in the 1MM IMM and the New York exchanges. As aa it may may be be said said that these exchanges exchanges are equal to rough generalization, it inthe transactions carried out by one major U.S. bank bank as as far as its in- fluence on the market is concerned. -263—263— Considering all all this, it it still still remains true that that aa speculator is is able to make make profits profits more consistently by running with the markets rather than by taking taking aa position and hoping for a turn-around in in the the market. To try to pinpoint market turn-arounds is exceedingly exceedingly diffidiffi- cult as everyone who who has tried his luck at it knows. The upshot is that the herd herd instinct in in foreign exchange markets markets still very powerful and the is still the well-informed speculating loner is the exception rather than the rule. Consequently, speculative speculative activity activity may well accentuate rather than than reduce exchange rate fluctuations. Investment The uncertainty surrounding the exchange value of the currencies has also taken its toll on on the willingness of investors to to engage in in foreign direct investments and in long—term long-term construction construction activity abroad. Increasingly, foreign countries countries insist on on denominating long—term long-term construction contracts in their own currency, currency, forcing the American businessman to shoulder shoulder the foreign exchange exchange risk. risk. Foreign direct inin- take five or or even vestment and long-term construction projects that may take ten years to complete are particularly affected by by the the exchange rate uncertainty because because there are no no organized in which organized forward markets in such long—term may be hedged. hedged. long-term exposures exposures may In addition, many of these projects are located countries for not even projects are located in in countries for whose whose currencies currencies not even regular regular forward markets or capital markets markets exist, thereby making hedging an imimpossibility. Under such circumstances the only options open to the businessman are are to to assume the foreign exchange exchange risk or to forget about the contract. -264- Foreign investment activity is is also also greatly complicated by by Foreign changing currency values. What might be be aa profitable profitable foreign operation What at at one one exchange rate may rapidly become unprofitable unprofitable as as the foreign foreign exex- change rate changes. In addition, arbitrary accounting rules -- such th-bitrary accounting —- have significant significant impact impact on aa firm's and loss loss as FASB 88 -- may have firm’s profit and —— position regardless regardless of of the profitability of the underlying underlyinq manufacmanufacturing activities. At best, the effects of of exchange rate changes on on the balance sheet make profitamake it much much more difficult to evaluate evaluate the profita- bility of the investment. investment. At worst, worst, it leads to erroneous erroneous investment decisions and ultimately ultimately aa retreat from international activities. decisions and The Public Sector The question arises whether the additional costs imposed imposed upon the The private sector of the economy are counterbalanced by benefits to to the public sector of the economy. While While this this is is aa difficult difficult question question to answer, II believe that it must be be answered in the negative. an economic economic Benefits may accrue to the economy by the creation of an about aa greater greater freedom to pursue approapproenvironment that would bring about priate economic economic policies, foster higher growth, or lessen inflationary inflationary pressures. On all these counts the actual actual experience with flexible flexible exex- has been discouraging. change rates has Of course, course, the ultimate proof of Of any of these propositions is is impossible to to attain. It It would require aa under a fixed exchange exchange rate regime -- and that is is replay of history under -— clearly impossible. Economic inference inference makes it it also difficult difficult to see why aa floating exchange rate regime should be be characterized characterized by high growth and little inflation. inflation. The fundamental fundamental point is that the flexible exchange rate —265— -265- system does not lessen the balance of payments payments constraint -- it merely —— changes its nature. :hanges It is difficult difficult to decide whether aa loss loss of foreign exchange rereserves serves or aa fall in in the foreign exchange rate provides provides aa more rigid policy constraint. constraint. policy But while the loss of foreign exchange reserves under aa fixed exchange rate system provides not only aa self—limiting self-limiting constraint in that no country country has either unlimited unlimited reserves or unlimitunlimit- the loss of reserves reserves eventually eventually ed access to international credit, the more restrictive monetary policy which which will forces the adoption of of aa more tend to bring the country in in line with the global inflation rate. rate. Flexible exchange self—limiting properexchange rates do not have such self-limiting proper- it has has instead been suggested that the depreciation of aa curcurties, and it rency circles where curcurrency may well lead to the development of vicious circles rency of its immediate rency depreciation brings about more inflation because of impact on on the price of imported imported commodities. rekindled inflationinflationThe rekindled may force aa further depreciation and so on. ary forces in turn may While the statistical evidence evidence on the validity of this theory is is far from complete and doubtful, it stands to reason that aa fixed exexequalizer of international inflation inflation change rate system operates as an equalizer accentudifferentials, while while a flexible flexible exchange rate system tends to to accentu- ate inflation differentials. As far as the international businessman is concerned, it is is clear which one constitutes the more attractive attractive environment: environment: given aa choice between similar -- even if high -- inflation inflation rates in all countries and between —— —— an environment or widely divergent inflation rates, rates, the the businessman is an likely to to choose the former one. -266—266— which one of the two alternatives is is However, it is questionable which best for all people of the world. For the central central banker, banker, floating rates do not seem to have brought aa more relaxing lifestyle either. Gross foreign exchange marmar- ket intervention intervention on behalf of central central banks amounted to aa record of of $72 billion 1979. billion dollars in the half year ending ending July July 31, 31, 1979. To To put this this number into into proper perspective, let us remind ourselves that the total foreign exchange reserves of all countries in the world totalled the sane amount in 1971, the last year of same of the the fixed exchange exchange rate system. Increasing, rather rather than than less, official intervention intervention has has been the hallhallmark mark of the flexible exchange rate system in in the seventies. The International International Monetary System System The exchange exchange value of the dollar against the DM OM (deutsche mark) or or such aa SFR (Swiss franc) has been been cut in in half over the last decade. That That such world’s leading precipitous decline decline in in the value of of the world's leading reserve currency cannot be without impact impact on the role of this this currency in the world world and and on the international monetary system itself goes without saying. official foreign exA A superficial glance at the percentage of official exof dollars shows that the market share share change reserves reserves held in in the form of of the the dollar has remained virtually virtually constant constant at 80 percent. percent. these figures are -- in in my opinion -- highly misleading. misleading. -— —- However, While While high U.S. Treasury officials officials have argued that the dollar purchases purchases on behalf behalf of foreign central banks were were proof of their their confidence in in the U.S. dollar, it it is probably more appropriate to argue that these official dollar purchases were largely the result of of intervention intervention designed designed to stop an an even further slide of the dollar. -267- The The foreign central banks were the the reluctant reluctant victims of aa declining dollar and not the exuberant investors investors they are made out to be. In fact, foreign central banks of floating currency countries have reduced the share of dollars in their foreign exchange reserve portfolio from over 90 90 percent in 1970 1970 to less than 75 percent in 1976. So have the central central banks of countries other than the main main industrialindustrialized countries, who who acquired the dollars dollars as aa result of their interintervention policy. In other words, those central central bankers that were free to consider the dollar dollar as a portfolio portfolio investment instead of am an interintervention currency did in in fact switch away from the dollar. The decline of the dollar in official foreign exchange exchange portfolios was also masked to some extent by the even faster decline of of the BritBritish ish pound in international significance. Central banks have switched pounds and purchased purchased DM OM over the last decade, so that the posiout of pounds position tion of of the pound is now held by the mark. mark. It It stands to reason that central banks would DM anyhow, and had would have have wanted wanted to to acquire OM had it not been for the fact that the pound was even even weaker than the dollar, the certainly have been more pronounced. switch out of dollars would certainly In addition to the decline in the value of of the U.S. dollar, dollar, there are other other reasons that make it attractive for central banks to diverdiversify their their foreign exchange exchange portfolios portfolios to an increasing extent. First it is clear that aa diversified currency portfolio increases increases its its of all, it overall stability. Second, as exchange exchange rates fluctuate fluctuate it may be prupru- one’s trading currency of one's trading partners. dent to hold reserves in the currency Third, the same argument applies applies to the denomination of the currency in which the country’s country's external debt is denominated. connecIn that connec- tion it it is is important important to to note the very rapid swing away from -268-268- dollar-denominated international bond bond issues in recent years. dollar—denominated In 1976 In the value of still of dollar-denominated international bond issues issues was still three times as large as the value of OM bonds, but by 1978 the OM volume was equal to the dollar volume. Consequently, Consequently, the need to make amortization amortization and and interest payments in in marks will will continue to increase desirability of holding increase in the future and and with it the desirability holding marks as as liquid reserve assets. We may therefore therefore conclude that: one, the flexible exchange rate system system has been been associated with with aa significant significant increase in costs to the private sector; two, that that it it has not brought brought about aa climate climate for the conduct conduct of more effective effective stabilization stabilization policies; three, three, that that it has not decreased the cost cost of intervention intervention for central banks; and four, it has fostered the decline of that it of the dollar as as the world’s world's leading currency. II will now consider several efseveral measures that might improve the ef- fectiveness of stabilization policies under the flexible exchange rate rate system. IMPROVING THE OPERATION OF OF THE FLEXIBLE FLEXIBLE EXCHANGE EXCHANGE RATE SYSTEM At the present present time, there exists no viable alternative to to the flexible exchange exchange rate system. simple: The The main reason reason for this conclusion is is as long as differential inflation rates among countries countries prepre- vail, it is not possible to to impose or to to achieve exchange rate stabilstability. The framers framers of the new new Article IV of of the IMF IMF (International (International MoneMone- iclesof cement were fully aware of this tary Fund) ~ Articles of Agreement this point: point: exex- change rate stability cannot be achieved without internal internal stability stability in economies. the relevant economies. To blame the flexible exchange rate system system for —269— -269- especially by the private private the additional costs that have to be borne -- especially —- sector -- would be to blame the messenger messenger for the bad news. —— Nevertheless, there are certain improvements in the operation operation of exchange rate system that can be be made in order order to to enhance enhance the flexible exchange its effectiveness and to reduce the costs associated with it. it. These are the lessons we can learn from the experience gained during the the seventies to to enhance the operation operation of the international monetary system during the eighties. It will be convenient to group the suggestions into two broad categories: those pertaining to improving improving U.S. monetary monetary and exchange exchange rate policy and those relevant for the international monetary system. Possible U.S. Policy Policy Improvements It should be be feasible to improve improve U.S. monetary and exchange rate policy with aa view towards enhancing the stability stability of of exchange rates. viewtowards The first set of suggested steps pertains to to the conduct of U.S. U.S. monetary policy, and it is gratifying that that the Federal Reserve has alalready announced announced the adoption of monetary targets and their supremacy supremacy over interest rate targets. The experience experience of having to chase chase the marmar- higher and summer of 1979 ket interest rates higher and higher during the summer 1979 while real interest interest rates remained remained negative negative and the money money supply grew out of of control was an important deciimportant factor in influencing the the October October 1979 1979 deci- sion to use bank reserves instead of the Federal Funds rate rate as an imimmediate operating target. Of Of course, both the Federal Reserve and and the other market participartici- pants will have to gain experience and confidence in in the operation of of the new system to ensure its proper proper functioning. -270- In In that connection it it is somewhat syssomewhat disconcerting to note that the introduction of of the the new sys- tem was was not handled in implementation as tem not handled in aa fashion fashion designed designed to to make make its its implementation as smooth as possible, but was conducted in an abrupt and disruptive fashfash- ion that resulted resulted in the introduction of uncertainty, confusion over Federal Reserve, and thereby greater market market ininthe intentions of the Federal stability symptoms that stability -- the very very symptoms that the Federal Reserve action should —— have helped to to alleviate alleviate rather than than to foster. Nevertheless, the overall thrust of the new policy is good, and once the dust has settled the the targeting on the monetary aggregates should prove to be a significantly significantly better system system than the interest— interesttarget approach used in the past. The operation operation of the system could be further enhanced enhanced by the anannouncement of of intermediate range monetary targets as guideposts for the Federal Reserve. Such three three to to five—year five-year targets could be very helpful helpful in signalling to the private sector the the clear clear intention intention of the Federal levels and Reserve to reduce monetary growth rates to non-inflationary levels to provide provide aa framework for orderly and sustained sustained economic economic growth. Of Of course, course, such targets must be strictly adhered to, so that confidence confidence in the policy of the authorities will will be be enhanced. policy statements statements of To use use the the announcement announcement of official targets to influence expectations without appropriate follow—through follow-through and implementation merely creates aa climate climate in which all policy pronouncements pronouncements will be doubted and will therefore therefore become less and less less effective. it is also important to have a realistic realistic monemoneIn that connection it tary growth target supported by by a coordinated fiscal strategy. To anan- nounce aa reduced monetary growth target while the public sector borrowreduced monetary borrow- expected to increase increase drastically might not ing requirements are expected -271— -271- policy package in that context. constitute aa credible policy policy Monetary policy cannot work in isolation isolation and must be seen as one ingredient ingredient in in aa cocoordinated ordinated poliãy policy package aimed at achieving achieving economic stability. erThe central bank can also play an important role in reducing er- ratic exchange rate fluctuations as as the November 1978 policy actions showed. There is aa significant difference between intervention to maintain an an exchange exchange rate rate that has has become unrealistic, and interveninterven- tion to turn around aa market trend that has become disequilibrating. Central banks have now learned the lesson that there there is is little little to be Central gained by trying to maintain maintain an unrealistic exchange exchange rate. Not only only Not are the foreign exchange losses incurred staggering, but the domestic such ill—advised ill-advised intervention are are also disadvantageous. disadvantageous. consequences of such A exchange markets to A central bank that sells its currency in foreign exchange keep it from appreciating increases the monetary base by providing providing more more keep of its own currency. This in turn turn increases inflationary inflationary pressures pressures at aa later date, thereby leading to to domestic instability. Similarly, aa central bank that depends on on unrealistically unrealistically high losses exchange rates will soon find that the foreign exchange reserve losses drastic exchange are staggering and will be forced to permit a more drastic exchange controls with rate realignment at a later date date or or to impose exchange all undesirable consequences consequences attached to to such such measures. In contrast, central central bank intervention to turn the foreign exexIn change market market around and to end a trend that has has clearly clearly become destadesta1978 U.S. policy bilizing can be be highly successful as the November 1978 measures showed. The The essential ingredient to to the success of of such an intervention policy is the simultaneous adoption of domestic monetary monetary root cause of of the the exchange rate rate policy measures measures that attack the root —272— -272- movement. It will be recalled that from November 1978 until April 1979 there was was virtually no no monetary monetary growth in the U.S. This was taken by the markets as aa signal that the Federal Reserve was prepared to pursue aa tight, anti-inflationary anti-inflationary monetary monetary policy and the dollar remained stable during that period. In April April 1979 the money money supply again began to excessive rate, driving up interest rates, increasing to grow grow at an excessive increasing ininflationary pressures, pressures, and and bringing bringing the dollar under renewed pressure, pressure, thereby necessitating the November 1979 policy actions. November 1979 Possible Improvements in the International_Monetary_System International Monetary System It It is my belief that the world world economy could could function quite quite well global unit under aa dollar standard, where the dollar is the the dominant dominant global of account, transaction currency, and store of of value. An indispensable indispensable precondition for the functioning of such aa system is the unquestioned stability of the dollar in terms of real real purchasing power. power. Oomestic Domestic inflation and the accompanying erosion of forof the the currency’s currency's value in foruneign exchange and international commodity commodity markets will have the un- avoidable consequences of reducing the dollar’s dollar's international role. The British inflation and decline in the value of the pound resulted in the elimination of of that currency from the reserve currency status that it once enjoyed. Continued double-digit inflation in in the United States will undoubtedly bring about the demise demise of the dollar as as aa reserve curcurrency as well. It is is up to the United States to get its own house in in order order if if It she wishes to preserve preserve the international international stature stature of the dollar. The The benefits flowing to the international community community as aa result of such such acaction would undoubtedly undoubtedly be great. great. -273—273— The most likely alternative to aa dollar standard is at present multiple-currency reserve standard, where several the development of aa multiple-currency currencies, in addition addition to the dollar, dollar, will serve an international role. However, it it should be realized that such aa multiple-currency standard is inherently unstable and is likely to lead lead ultimately ultimately to to severe financial and economic economic disturbances. For the same reason reason that bimetallism proved to be be unstable, unstable, it it will be be found that relatively small differences in national national inflation rates among the different different key currencies will lead lead to relatively large shifts in in capital flows among these countries. countries. Such capital flows will exacerbate balance of of paypay- difficulties, as capital is likely to flow into aa country country that ments difficulties, already enjoys aa current current account surplus. exchange rate rate Consequently, exchange movements will be accentuated, official intervention will will have have to bebecome even larger, or capital controls will have to be be introduced. UlUl- timately, it is likely that capital controls controls cannot cannot be avoided, avoided, and the very benefits benefits of aa liberal international international financial order order will be be destroyed. multiple—currency The only feasible realistic realistic alternative to a multiple-currency SDR (Special Drawing system is at present aa system based on the SOR Rights). The recent decline of the dollar dollar has has consequently consequently led to to aa renewed interest in the SDR SOR as an an international reserve asset. This This turn of events 1968 events is not without irony, because the the SDR SOR was born born in 1968 out of fear that there might be be aa shortage shortage of dollars when when the U.S. balance of payments would return to surplus. Instead, the SOR SDR is now likely to assume aa larger role on on the international economic economic scene bebeperceived surplus of dollars. cause there is aa perceived -274- The renewed interest in a dollar/SDR dollar/SOR substitution account is is the natural outgrowth of these developments. To move SDR firmly to the center of the international monemove the SOR monetary system would would require at least three steps: to base exbase the IMF ex- SDR, to make the SOR SDR usable -- that clusively on the SOR, that is transferable -— SOR inflation—proof. inflation-proof. among private entities, and to make the SDR Let me elaborate on briefly elaborate on each one of these these points. Recently, the Economic Counsellor Counsellor of the IMF, Mr. J. J. J. J. Polak, set forth a plan plan that would make the SOR SOR the centerpiece centerpiece of all IMF operations. would signifiThis innovative and farsighted suggestion would signifi- cantly cantly enhance the importance of the SOR SOR and make it a more central asset in the international monetary system. In addition, addition, such aa move would would also have the advantage of of unifying unifying many of the Fund Fund operations that are now proliferated among an ever larger larger and more more complex variety 11 of 11“accounts” accounts'* and “facilities.” facilities. 11 Second, the SOR should be made made transferable among private as well as public holders. When the SOR SDR is freely traded in in international fifi- nancial nancial markets its its usefulness usefulness and and liquidity liquidity will will be be greatly greatly enhanced. The SOR is not not likely to assume The SOR is likely to assume aa significant significant role role in in world world financial financial markets until transactions that until it is also used widely for comercial commercial transactions create aa need need to to effect payments in in SORs. But if SORs are are not freely transferable between private and official holders, it is unlikely that they will assume assume an an important role role as an international means means of paypay- ment. Transferability Transferability of the SOR among private parties is therefore therefore essential international monetary system is essential if the international is to be be based firmly on the SOR. SOR. —275— -275- SOR should be turned into a true global standard of Third, the SDR inflation—proof. value by rendering it inflation-proof. Traditionally, gold has has fulful- acfilled the role of an an international standard of value but official actions and the recent speculative fever have have deprived deprived gold of its its status as aa stable stable measuring rod. Instead, it has become a highly speculative speculative commodity. As presently constituted, the SDR SOR offers some protection against the risk inherent in in differential inflation inflation rates by providing providing the basket. holder with a diversified currency basket. But it it should be be noted noted that the value value of this currency currency basket in in terms of real purchasing power dedeteriorates along with the weighted average of the inflation inflation rates exexSDR basket. perienced by by the sixteen sixteen countries represented represented in in the SOR AA superior inflation inflation hedge is always available available to to the investor investor -- be be it aa —— monetary authority or or aa private entity countries. of high inflation countries. —— holding the currencies by not holding The SDR, SOR, as as presently presently constituted, constituted, forces inflainvestor to to accept the depreciating currencies of of the high high inflathe investor tion countries that do not enter the SOR interest rate calculations based on the five most important currencies only. Hence, the SDR SOR as presently constituted is not aa particularly particularly attractive asset. inflation—proofing of the SOR would The inflation-proofing would make it it a truly superior international international asset that could could play an increasingly important role role on on the world financial scene by by providing providing a universal unit of account, a monetary of value. monetary transaction medium, and aa stable store of Such an inin- flation—proofing of flation-proofing of the SDR SOR could be accomplished by by linking it to a price index of the sixteen countries countries making up the SOP SOR basket. Of course, there are many operational problems to be considered, but but these -276—276— more complex than those that had to to be be resolved when are not inherently more the SOR as currently constituted constituted was created. Of course, there remains a very disturbing thought: all the the If all individual necessary individual countries are unwilling or unable to take the necessary steps to bring inflation under control, why why should we we assume that all these nations acting in concert through an institution would would be any more willing willing or or able to act in in aa manner that would expose expose their their own own shortcomings7 shortcomings? Nevertheless, it may be be possible possible to achieve an an interinter- national consensus on the creation of such an an asset asset because the alteralternative of continued continued international monetary disruption is is associated with high high costs for all. The only other other feasible alternative for the eighties is a rapid inflation rate, such such that that the international role reduction in the U.S. inflation of the dollar will be preserved preserved in in the decades to come. Without aa Without stable dollar that can serve as the anchor of the international monemone- tary system system there is is not likely to to be be exchange exchange rate stability. The The elimination of inflation in the U.S. and in other countries will theretherefore fore be aa precondition for the the improvement of the operation operation of the international monetary system. Stability cannot be be imposed by the international monetary system or found by manipulating manipulating the system. International monetary International monetary and exchange rate stability stability can be be achieved achieved only by first attaining domestic stability. by -277—277— FLEXIBLE EXCHANGE RATES AND MONETARY POLICY: AA DISCUSSION OF THE THE FRENKEL AND HELLER HELLER PAPERS PAPERS David Laidler Laidler If aa conference such as this one, one, dealing with United States’ States' macro—stabilization macro-stabilization policy, had been organized ten years ago it is unun- likely that anyone would have have suggested devoting an entire session to to the operation operation of the international international monetary monetary system. If If the the suggestion suggestion had been been made, it would certainly certainly have been been greeted greeted with aa loud “why?” "why?" The very fact that this session is included included in in this conference epitoepitomizes the most important lesson of doof all that we have learned about domestic stabilization stabilization policy in the last last decade-—namely decade--namely that that it it cannot sensibly be discussed without explicit reference reference to the international environment within which it it is being implemented. By this II do not simply mean that United States States domestic domestic policies theworld have implications for the rest of the world that policymakers should be interested in, or that there are interesting debates about the organiorgani- zation of of the international international monetary monetary system, system, the outcome of of which which will influence the ease with which American business business can operate operate in in interinternational markets and which ought therefore therefore to to concern American policypolicymakers, makers, though both both of these observations are surely true. Rather II mean that the way way in which monetary policy impinges impinges upon upon traditional Or. Laidler is Ontario. This grant from the Canada. Canada. Professor of Economics at at the University of Western Professor on work completed under aa research paper draws heavily on Social Social Sciences and Humanities Research Council of -278- domestic targets, targets, employment, employment, prices prices and and the the like, like, is is intimately intimately linked linked domestic to to the operation of the international monetary monetary system. of the papers papers that II am am discussing has much to say Since neither of explicitly explicitly about these domestic domestic matters, matters, and that is not to criticize criticize either of them, because one can on~ysay on'.y say so so much in one paper, paper, II bebe- lieve that it will be useful for me to use 0s2 these discussant’s discussant's comments comments to explore this this area in the the light of the arguments presented by Frenkel to and Heller, rather than to provide a detailed critique critique of those arguarguments. Both of of the papers papers before us deal with the operati operation Both on of aa system exchange rates. of flexible exchange That is only right and proper, given that is the system (more or less) under which the world is currently currently this is operating. However, I believe that it would be be wrong for anyone to to conclude international factors for United conclude that that the new importance importance of international States States domestic policy stems from the the adoption adoption of aa system system of flexible exchange rates per se. se. exchange In the 1950s l95Os and 1960s, l96Os, United States policymakers were able to operate "as “as if” beif" the economy they were were dealing dea 1i ng with with was closed, not because cause the Bretton Woods system was aa fixed rate system, but because because it was a dollar standard system. As we now now know, with the the benefit of of hindsight, and as some--notably, some——notably, for example, Robert Triffin (1961)-(1961)—— argued at at the time, this did not mean that that the United States could indefinitely operate operate its domestic policies while completely ignoring countries used to be called “the "the balance of payments payments conconwhat in in other countries straint.” straint." “constraint” operated sufficsufficHowever it did mean that the "constraint" which domestic domestic iently slowly that, relative to the time horizon for which -279—279— stabilization policies policies are conceived, conceived, it seemed seemed unimportant. It would It only be be if the world were to to return to aa dollar standard that this happy, for United States policymakers, state of of affairs would be be rere- stored. However though II understand Robert Heller’s Heller's nostalgia nostalgia for such a system, II am much much less less sanguine than is is he he about the possibility possibility of of the restoration of aa dollar standard. standard. the The breakdown of the Bretton Woods system system has forcefully reminded us that the amount of can extract from his of seignorage which aa banker can clients depends upon their willingness to to pay up. If he tries to exexIf tract too much, they will, not without difficulty to to be sure, take their business elsewhere. At At the risk risk of oversimplifying, oversimplifying, under Woods, the banker, namely the United States, tried to to extract extract Bretton Woods, too much seignorage. The The current chaotic chaotic international monetary syssys- tem is the result of the customers trying, as best they can, to find somewhere somewhere else to do their banking business. AA dollar standard standard is not be restored unless it is clear clear to to the rest of the world that going to be the United United States has mended its ways, and is not going going to repeat its previous policies-—either policies--either willfully willfully or inadvertently. The pjfi~evi_()Jl_l,)' evi- dence dence that that the past decade has produced to support this view is is the recent recent announcement of monetary monetary policy changes by by Mr. Volcker. Volcker. If If that that announcement announcement is followed up up with action, and and past evidence evidence suggests that this cannot be taken for granted, and if the new policies policies are adadthat hered to long enough to erase erase the memories of inof fifteen years of in- stability, then the possibility possibility of restoring restoring aa dollar standard standard night might arise. However, II believe that we would be ill advised to to hold bold our breath in anticipation anticipation of of the event. -280- will Inevitably inevitably remain on Now this is not to say that the world will the present flexible exchange rate system into the indefinite indefinite future. The problems of operating operating under under such a regime as Heller describes describes are real ones, although how much they are exare the result of the flexible ex- much of of the underlying monetary ininchange rate regime per se, and how much stability stability that that forced the adoption adoption of that regime in in the first place, place, is aa point that one might want to argue about. There is is undoubtedly undoubtedly aa demand for aa stable stable monetary unit to to serve as aa means of of exchange, exchange, unit of account, and store of marof value in international transactions, and mar- kets have aa way way of of evolving in order to to meet meet such demands in a manner that verges on on the inherently inherently unpredictable. unpredictable. After After all, the the Bretton Woods system was was not designed to put the world world on on the dollar standard, nor did or indeed could the outthe United States in in any way way force this this out- come; it it arose as aa result result of the voluntary choices choices uf aa host of of instiinstitutions and individuals and and the evolution evolution in in question only appears appears ininevitable with the benefit of hindsight. In the current state state of knowledge, economic economic theory enables enables us us to say that, so long as domestic monetary monetary policies remain uncoordinated uncoordinated international monetary system and unstable, unstable, then then the the international system will will also be be ununstable, stable, whatever its formal formal institutional framework, and and that that as such policies become stable stable and harmonized, then then so will the international monetary system become stable and perhaps perhaps adopt aa new reserve currency, or indeed currencies. It does not enable us to say anything positive positive about the form that such an evolution is likely to to take. Nevertheless, given the array of of inflation rates, monetary expansion rates and such at present present ruling in in various parts of the world, one is tempted to conconclude the first step step towards towards reestablishing some sort of clude that even the -281- unified world world monetary monetary system has yet yet to be taken. unified The European European MoneMoneThe estabtary System is regarded by some as being the first stage in estab- lishing an an important important regional base from which such aa system might evolve; whether it is or not not depends upon whether its its members succeed in developing the means means to coordinate their domestic policies so as as to make them compatible with the maintenance of of the System, System, and they show show no signs signs of doing this. Be that as it may, as as aa practical practical matter matter any discussion of United States’ stabilization States' stabilization policies that is is to be be of of current relevance relevance should take aa flexible exchange exchange rate rate system as its its background. background. Thus Frenkel's paper, though it it the theoretical and empirical material in Frenkel’s will will look rather unfamiliar unfamiliar to to many specialists specialists in the analysis of domestic monetary policy, policy, is of considerable considerable relevance to to their conconcerns. cerns. II will now turn to some of the issues issues involved. It should should go without saying saying that if one is going to discuss the macro—stabilization policies are likely to work against way in which which macro-stabilization the background of aa flexible exchange rate regime, regime, one ought ought to know something something about the way way in which which the foreign exchange exchange market itself operates. Frenkel deals with this this matter matter from the point of of view of what what may may be be referred referred to to as the “Asset "Asset Market Approach Approach" to exchange rate beyond doubt doubt provides a simple and powerful theory, an approach which beyond method of analyzing the problem area. Nevertheless, anyone reading Heller’s paper immediately after Frenkels Heller's Frenkel's must wonder wonder where where many of of the concerns he raises, particularly about the large large amount of dollar— dollardenominated assets held abroad, fit into into Frenkel ‘5 's analysis. II believe believe that the answer here is that, although the theoretical framework framework which -282- underlies Frenkel’s Frenkel's work can deal with these issues, the particular underlies “monetary” version of the asset market approach which he "monetary" he explicitly explicitly sets out does does so only implicitly, implicitly, and in a way way that his his empirical empirical evievidence suggests suggests is inadequate. itThe basic monetary model of of the exchange rate rate is simplicity itself. self. With national price levels tied to each each other other by purchasing power parity and aa stable deiiand denand for real balances function in each each country, domestic price levels, inflation, nominal interest rates and the the exchange rate are simultaneously determined by the the behavior behavior of the “real” "real" arguments in in the demand functions in in question, and by that of the the supplies of of nominal money money in in the two countries. countries. What does this this analysis tell us about the role role in in influencing the exchange rate of U.S. dollar—denominated dollar-denominated assets left over over from from the period when when the dollar was the reserve currency, and and currently held abroad? This is a be of key problem which many commentators, including Heller, believe to be importance. The monetary model model tells us, II believe, that these assets have no special significance. They They are interest-bearing interest-bearing assets, and, according to the monetary version of the more more general asset market market apapaccording to of return return on them adjusts to to compensate compensate their holders proach, the rate of for any anticipated change in their purchasing power over goods and over assets denominated in in other other currencies. Variations in in that rate of return are taken account of in in the model because the nominal interinterest rate rate they bear is is an an argument in the U.S. demand for money funcfunction. The The above reasoning hinges upon purchasing power power parity always holding, but Frenkel’s Frenkel's empirical evidence shows that at the very best it does does so only on average over rather long long time periods, and in in aa -283- rough and ready fashion at that. This means that variations in the dollar—denominated assets cannot stimultaneously rate of interest on dollar-denominated compensate compensate for variations in in their purchasing power power over goods goods priced priced in U.S. dollars and goods priced in foreign currencies. currencies. This in turn means that, although some U.S. dollar-denominated dollar—denominated assets may be be perfect perfect substitutes for those denominated in foreign currencies, others are are not. That being the case, the currency currency in which which they are denominated must be be aa relevant relevant property of of at least some classes classes of securitieS, securities, and fluctuations in the supply and demand for such securities are likely to impinge upon upon the behavior of of the exchange rate. rate. The behavior behavior of the dollar—deutsche mark mark exchange exchange rate gives Frenkel more trouble trouble than any any dollar-deutsche other, and surely that is is not an accident, given that mark-denominated mark—denominated assets assets have so often been been the destination of of funds realized by selling dollar-denominated securities. securities. dollar—denominated There is another characteristic characteristic of of the U.S. U.S. dollar’s dollar's place in the the There international monetary system system worth worth noting: it it is is the unit of account for many international transactions, not the least of which are those involving oil. That means that many international prices are going to to be particularly sensitive sensitive to the conduct conduct of U.S. U.S. domestic monetary policy, and that that policy still has aa considerable power, for good or ill depending depending upon how it it is used, over over the international economy, aa power which it pf~ices in it would would not have were prices in that economy to be be set set in in other currencies. The frequent references in in U.S. debates to to oil price increases as domestic policy increases as being being exogenous exogenous to to domestic policy shows shows that that it it is is not not yet appreciated that oil oil prices in the world economy economy respond to U.S. domestic policy and that attempts to cushion their their effect by by domestic domestic monetary monetary expansion are not just useless but but actively actively harmful harmful. -284- To put put all all this this in in another another way, way, if if goods goods markets markets cleared cleared as as fast fast To as long—run equilibrium where the as asset asset markets, if we we were always always in long-run prices of national moneys and of of national outoutconcepts of the relative prices puts were were interchangeable, interchangeable, the above problems would not arise. HowHow- asset markets markets do clear faster, and in in the short short run do dominate dominate ever, asset of the exchange rate, so that the distinctions upon which which the behavior of the asset asset market approach approach focuses are important. implication of of the evidence that Frenkel presents. That That surely is one This This very fact howhow- ever ever seems to to me to imply that that the asset market approach to to exchange exchange rate determination must be carried beyond aa simple monetary formulaformula- tion, as it is, for example, by Boyer (1978), to incorporate explicitly explicitly other aspects aspects of of portfolio portfolio behavior, and to to incorporate other other aspects it can of using aa particular particular currency as a unit of account, before it claim to provide us us with aa complete toolkit for dealing with foreign exchange rate problems, not least those which Heller raises. NevertheNeverthe- less, if our it is still our toolkit is incomplete, incomplete, it still the best best one that we have. As Frenkel’s Frenkel's paper shows, the the asset market approach to to anaana- lyzing lyzing exchange rates rates is extremely useful, useful, and its use does enable us to to aa clearer understanding understanding of how how to conduct domestic domestic policy come to against a background of exchange exchange rate flexibility. One of the best established pieces of conventional wisdom in One international monetary economics is that high interest interest rates are assoassoweak one, one, ciated with aa strong currency and low interest rates with aa weak but one of the best established established facts of the last few years is that the high interest rates in in fact are are associated associated with weak currencies, and vice versa. As Frenkel Frenkel shows, the latter prediction prediction is what what follows -285- from the asset market approach, and, as he also shows, that theory’s theory's predictions in this respect are confirmed confirmed by by evidence, generated generated moremoreover by by an experiment experiment whose validity validity does not, as as far as I can can see, in any way hinge upon assuming that purchasing power parity holds. holds. Though Though II can find nothing to disagree with in anything that that Frenkel explicitly explicitly says about this this matter, there are aa few things that he didn’t didn't say that do seem to me to be of particular relevance to to the the theme of this conconference. The conventional wisdom about the relationship between interest otherwise of aa currency has has its historical rates and the strength or otherwise roots in the operation operation of the gold standard, and in particular in the role played by by the central central bank rediscount rate in in the conduct conduct of monemonetary policy under such well—known, such aa system, a role summarized in that that well-known, but now sadly outdated, aphorism “Seven per cent will draw gold from aphorism "Seven the moon" moon” (which I have been been unable to to track track down to its original source). source). Under such such aa system the long-run time paths of money money and prices in the international economy were given by the rate of of change change of the stock of gold. Though this rate of change was not always smooth and and steady, steady, because important new gold discoveries discoveries were were from time time to time made, made, on on average it was. Given that, and given an unquestioned unquestioned commitment of central banks to maintain the convertibility of of domestic commitment money into gold, the anticipated inflation rate was, by by comparison comparison with recent experience, not far short of being an an exogenous exogenous constant. MoreMore- over the aim of.monetary not to and over the principal principal aim of monetary policy policy was was not to control control income income and employment but but simply to maintain maintain convertibility. convertibility. employment In such such a world, world, any any In increase in aa central bank’s bank s discount discount rate represented an increase in 1 the real cost of borrowing from the banking system, and hence hence led to to aa -286- contraction (or (or at at least least aa slowdown slowdown in in the the rate rate of of expansion) expansion) of of domes— domescontraction tic credit. credit. tic The monetary consequences of of that that in in turn turn led to to aa balance balance The of payments payments surplus and hence aa “strong” "strong" currency. currency. of different than that The world of of the last last ten years years has been very different that whieh which I have just just described. standWith nothing to replace the gold stand- ard’s guarantee guarantee of expectaard's of long—run long-run price price predictability, predictability, inflationary inflationary expectadomitions have become become endogenous and volatile, and their movements movements domi- nate fluctuations in nominal nominal interest rates. rates. It is these these factors which which have led to to the association of of high interest interest rates and weak currencies. Both are the the consequence of an an adverse response response of inflation inflation expectaexpectations to undisciplined undisciplined and expansionary expansionary monetary monetary policies, as Frenkel has argued. II believe that the forgoing considerations considerations have two important important imim- conduct of domestic monetary policy in in the United plications for the conduct States, both both now now and in in the future. States, emFirst, though at long last an em- phasis on controlling monetary aggregates is replacing replacing an an emphasis emphasis on on interest rate targets targets in in the conduct conduct of be foolish to of policy, it would be believe that the battle battle here has has been finally won. being fought. Rather it it is still The advocates of of controlling monetary aggregates have inferalways based much of their case upon the difficulty of of drawing infer- ences from aa particular particular value of the the interest rate about whether whether policy policy is “tight” do so. tight or “easy,” easy,1' and will continue to do 11 11 11 The The forqoing forrioing analyanaly- sis sis surely helps to to bolster bolster their case, case, for it it shows that there is is an to the problems to which which they have important international dimension to been pointing, a dimension dimension that adds adds weight to to the argument against using interest rates as as aa policy indicator. -287- The second implication worth pointing pointing out out is not of of such such imedi— immediThe ate concern, as important. concern, but is surely just as forgoing argument argument The forgoing proposiamounts to to presenting aa special case case of the following general proposi- tion: the way in which monetary policy impinges upon the domestic economy, and the way in in which which domestic monetary variables variables should be interpreted by the authorities depend critically critically upon the the state of of the international monetary system and the nature of the country's country’s exchange rate regime. regime. II believe that many of of the United States’ States' current policy apprecidifficulties have arisen from aa failure of the authorities authorities to to appreciinternational factors are of prime rather than ate the fact that these international in the design of of policy. secondary importance in To put the matter matter in its its it is not just the way way in in which United States policy simplest terms, it affects the rest of the the world that varies with the exchange rate regime and the conduct of policy in other countries; the way in which which it it afaffects the United States is is also profoundly influenced by these matters. imII will now turn to aa more specific specific discussion of this point point as as it it imupon the conduct of pinges upcn of policy policy under the present regime. There is no doubt about the nature of of the current macro macro policy problem facing the United States: it is how how to to reduce the inflation inflation rate without at at the same time causing more of aa real real contraction than is absolutely necessary necessary (however (however much that that might be). It is is also also true that there is is aa wide consensus that getting the monetary monetary expansion rate "under key role in tackling this problem. problem. “under control" control” must play aa key Debates it comes to the question question of how to implement such a policy, arise when it monetary expansion expansion “under "under control” control" means in of specifying what getting monetary practice. At At one extreme extreme are those who who follow the lead lead that (I an am glad -288- to learn from Neil Wallace) Sargent Sargent and Wallace Wallace (1975) never meant to give. They argue for aa rapid, pre-announced, monetary monetary slowdown slowdown which will, by way of a by by now well—known well-known "rational “rational expectations" expectations” mechanism, only impinge mainly upon prices and will affect output and employment only extent that the pre-announcement to the extent pre—announcement is not believed. extreme are those like Modigliani (1977) (1977) who believe At the other extreme that aa monetary monetary contraction contraction can be fine tuned, while in the middle middle stand those who would support aa gradualist gradualist contractionary contractionary policy policy of the type advocated Allan Meltzer. advocated at this conference by Allan To obTo aa foreign ob- server, the striking characteristic of of this this United States States policy policy debate is the way in which the openness of of the United States economy and the nature of the exchange exchange rate regime are virtually virtually ignored by by all all particparticipants. Nevertheless, the theoretical theoretical and empirical results presented presented by Frenkel at this conference, not to to mention a good good deal of work on stabilization problems problems in open economies that that has has been been carried on on mainmainly outside the United States, is extremely extremely relevant relevant to these issues. ly Two key questions questions underlie underlie current debates about stabilization policy. The first concerns the speed speed with with which the private sector sector of the economy can absorb information about policy and translate translate that ininformation into into price changes, changes, and and the second, analyzed by by Lucas Lucas (1976), concerns the stability stability over time of the mechanisms whereby information is is absorbed absorbed and acted upon and the independence independence or otherwise otherwise between those mechanisms and policy actions actions themselves. If If one one believes believes that information is is absorbed absorbed and acted upon quickly, then rapid monetary monetary contraction anti—inflation policy. contraction is an appropriate anti-inflation policy. If If one believes that reactions here here are slow, but but that their their time path in in the future can be inferred reliably from past behavior behavior then one will advocate fine -289- tuning. tuning. mechanism AA slow but unstable, and hence hard to predict, mechanism underpins the case for gradualism. (May I note here in passing that II (May believe Meltzer’s Meltzer's analysis of the case case for gradualism, which which II largely accept, would be be enhanced if he would lay lay more stress upon the ~pp~~— unpredictability of of the lag lag structure structure of of his his model in any particular particular ininstance, and less upon its drawn out and backward looking se.) looking nature per se.) ‘s empirical work shows that the foreign exchange exchange market Frenkel 's that all available information, including including is efficient, in in the sense that information about policy, is translated quickly quickly into movements movements of the about policy, exchange rate. exchange rate is, is, therefore, aa price price that, other other The exchange things equal (the qualification is and I will will return return to it i~jj!9~~q~j (the qualification is important important and to it in in a moment) adjusts rapidly to policy changes. AA number number of recent papers augof the the aggregate aggregate demand—expectations demand-expectations augpapers have analyzed versions of mented Phillips curve model, which underlies underlies so so much United United States policy debate, extended explicitly to incorporate a foreign sector. Though such exchange rate such work is most most highly developed for fixed exchange regimes——see, e.g., Laidler (1975), Jonson (1976), Jonson, Moses and regimes--see, Wymer (l979)——some results are now availWymer (1976), (1976), Bilson (1978), (1978), Burton (1979)--some avail- able for aa flexible rate regime. Thus Laidler Laidler (1977) (1977) shows, albeit in an extremely primitive model model with zero capital mobility, that even where errors are made made about the domestic price price level, perperwhere systematic systematic errors fect foresight about the exchange exchange rate rate is is sufficient to to guarantee guarantee that that domestic monetary policy impinges solely upon domestic prices and not at all on output. orate much more elabelabBurton (1979 and forthcoming), in aa much model that does does incorporate capital mobility, mobility, aa variety of stochastic shocks, and rational expectations, finds that that the behavior of the exchange rate is aa key source of information for agents and that -290- the more rapidly information information about it it is is available to them, them, the the more direct is the linkage between domestic monetary monetary policy and domestic prices. One must be careful not to to read too much in the way way of policy imin— plications from analytic analytic exercises such as these. Nevertheless, the work that II have referred to does point to the the conclusion that that aa flexflexible efficient market, narket, imparts to an ible exchange rate, determined in an efficient an economy an an extra degree of price price flexibility that it it does not have under aa fixed rate. This in turn suggests that estimates of the output output that might be lost in the United States States while while bringing inflation under under been generated from data on the fixed exchange exchange rate control that have been period are likely to be exaggerated, even if there is nothing else wrong with the techniques used to to derive them. there is a very important qualification to be be added to However, there all this. The theoretical theoretical results results to which which I have alluded are premised premised on on the price level, and implicitly the money market, in the rest of the theoretical experiment from world remaining undisturbed during the theoretical which they are derived. To put the matter in terms of Frenkel’s Frenkel's frameframe- work, they apply to situations in which which nothing happening abroad disdis- turbs equilibrium in the market for foreign money, money, or or foreign assets in in exchange rate originate originate in the the general, so that all disturbances to the exchange money supply. behavior of the domestic money Why this is an an important important qualiquali- fication is easily seen by considering Frenkel ‘s 's analysis and his his emempirical results. If aa foreign monetary contraction begins at at the same time question tells us that, that, given given tine as aa domestic one, the analysis in question for the of simplicity that the relative relative sizes of these contraccontracthe sake of nothing will happen to to the exchange rate. tions are appropriate, nothing —291— -291- In In that case domestic money markets must be cleared by domestic output and price price level level fluctuations without help from aa quickly adjusting foreign exchange exchange market. Frenkel ‘s 's results on on purchasing purchasing power parity lend weight view that that domestic prices adjust adjust slowly disweight to to the the view domestic prices slowly to to monetary monetary disturbances. Thus there there is every reason reason to suppose that in in this this case, and in the short run, which may nevertheless persist tine, persist for aa long time, much of of the effect will be be on output. The implications of of looking at Frenkel Frenkel 's empirical results results on the ‘s empirical efficiency of of the foreign exchange exchange market market in the the light of the macromodels II have cited in the preceding preceding section may may be be summarized as follows: a single single economy seeking to tackle an inflation inflation problem against the the background background of an otherwise tranquil world economy will will find that the existence existence of an efficient market market for foreign exchange exchange under aa flexible rate enhances the flexibility of of domestic domestic prices. Such Such an an economy will will enjoy an easier transition to aa lower inflation inflation rate rate than one would expect from studying closed economy models. However, However, if if that same economy is one among aa number number faced with aa similar problem, then, even with with a flexible exchange rate, the pressures pressures of domestic deflation deflation will, if other simultaneously deflating, be concentrated other countries are simultaneously on domestic output. In general general,, the extent to which which this this happens in any one country country will vary with the conduct of of policy abroad. In the current state of of knowledge, II do not believe we can say any more than this, but II would would claim that even even this much is important important to know. know. Our consideration of open economy aspect of of the open of stabilization policy has, after all, led led us to argue argue that the lags with which informinformation will become available, available, and hence aa basis for action, will vary —292-292- with the way way in in which policy policy is is conducted not only at home but also also abroad. The length and variability of such lags are, are, therefore, in in any such lags particular instance, going to be next to to impossible impossible for policymakers to predict. predict. However However such unpredictability is the very essence of the the case case for gradualism. The The analysis we have been considering considering does, therefore, make an important contribution to to the current U.S. U.S. policy debate. —293— -293- REFERENCES REFERENCES 0. (1978), “A Devaluation,” Canadian Bilson, J. F. O. "A Dynamic Model of Devaluation," Journal of of Economics 11, ll , 194—209. 194-209. Boyer, R. S. S. (1978), “Financial "Financial Policies Policies in in an an Open Open Economy,” Economy," Economica Economica 45, 39-57. “Expectations and Economy,” unBurton, D. D. A. A. T. T. (1979), "Expectations and aa Small Open Economy," unpublished PhD thesis, University of Western Ontario. (forthcoming), “Expectations Flex"Expectations and aa Small Open Economy Under FlexRates,” Cana Canadian - - iible bl e Exchange Rates," di an Journal of Economics. Economics. Jonson, P. D. D. (1976), “Money "Money and Economic Activity in the Open Open Economy, The The United Kingdom 1880—1970,” 1880-1970," Journal of of Political Economy 84, 84, 979-1012. 979-1012. Jonson, P. D., “A Minimal Model of D., E. R. Moses and C. R. Wymer (1976), "A of the Australian Economy,” Economy," Reserve Bank of of Australia Discussion Paper 7601. 7601. “Price and Output Fluctuations in an Open Laidler, D. D. E. W. w. (1975), "Price Economy," Essays on Money Money and Inflation, I nfl ati on, Manchester, Manchester, Economy,” Ch. 9 of Essays Manchester Manchester University Press and Chicago, University of Chicago Press. _____ “Epectations and the Behaviour (1977), "Epectations Behaviour of of Prices and Output Under - - FFlexible l ex i bl e Exchange Rates,” Rates," Economica Economi ca NS 44, 44, November, 327—336. 327-336. Lucas, R. E. E.,, Jr. (1976), in Karl Brunner and Allan H. Meltzer (eds.), Labor Markets, Carnegie—Rochester Carnegie-Rochester Public The Phillips Curve and Labor Policy Journal_of Policy Series supplement to the Journal of Monetary Economics. Economics. “The Monetarist Controversy Modigliani, F. (1977), "The Controversy or Should we we Forsake Stabilization Policies?” kierican Economic 1—19. Stabilization Policies?" American Economic Review 67, March, 1-19. “Rational Expectations and the Sargent, T. J. and N. Wallace (1975), "Rational Theory of Economic Policy,” Policy," Research Research Department, Federal Federal Reserve Bank of Minneapolis Studies in Monetary Economics 2. Triffin, R. (1961), Gold and the Dollar Crisis, The Future ConvertiConverti~jjj~, UniversitjTress. bility, New Haven, Yale University Press. -294— -294- FLOATING EXCHANGE RATES IN THE 1970s: 1970s: AA DISCUSSION OF OF THE HELLER PAPER PAPER Geoffrey E. Wood Heller’s paper asserThe first section section of of Dr. Heller's paper consists of four assertions about the consequences for the world economy of the move to to Floating Floating exchange exchange rates. On On the basis of these four assertions Dr. leller ieller proceeds to to make recommendations recommendations first for the future conduct of of J.S. J.S. economic policy, and second for the future shape of the interintermtional 1ational monetary system. In these comments comments it will be be argued first that his four asserassertions of exchange exchange rate rate flexibility tions on on the the consequences consequences of flexibility are are at at the the least least nisleading and, in some cases, not supported supported by any evidence evidence at present ~vailable.It ,vailable. It will then then be shown shown that his policy policy recommendations recommendations for the future of the international monetary system are based on on misunder— misunder;tanding of exchange rate volatility and the reasons ;tanding both both the causes causes of capital movements. movements. for international capital The comments conclude with aa sumsum— nary of what appear to be the true lessons lessons of the the floating exchange exchange rates 1970s. rates experience experience of the 7970s. DR. HELLER’S HELLER'S ASSERTIONS ASSERTIONS Dr. Heller asserts “the operation Or. asserts that "the operation of the the flexible exchange rate system since 1971 has significant increase in costs rate system since 1977 has entailed entailed aa significant increase in costs to to )r. Wood is aa member of the Centre for Banking and International Dr. Finance, Finance, The City University, London, England. This is a revised and ?xpanded version of of comments comments made the conference conference on ,xpanded version made at at the on Dr. Dr. H. H. Robert Robert leller’s ,eller's paper. The author author is indebted to several several conference conference partici— partici,ants for remarks which have have improved these coments. comments. Dants —295— -295- the business sector.” sector." The trouble with that that statement statement is is that that Dr. Dr. Heller does not make clear what comparison he he is is making when he he says says costs have increased. There has been a substantial increase in the There dispersion of inflation rates in the O.E.C.D. O.E.C.D. (Organization for Economic l970s as compared to the Cooperation and Development) area area in in the 1970s as compared 1 1960s.1 Had exchange rates rates remained pegged despite despite this change, they been kept so so by by an an increasing proliferation of exchange exchange could only have been controls to restrict capital movements and of of tariffs and quotas to rerestrict trade, trade, and by increasing volatility of of national monetary policies. policies. It is is impossible to to believe that these developments developments would would not have imposed costs costs on on the business sector, and Dr. Heller Heller certainly does not demonstrate that these costs would would be less less than the costs imimposed by floating exchange rates. rates. Indeed, it should be be pointed pointed out that that there is absolutely absolutely no no evievidence dence in in support support of of Dr. Dr. Heller’s Heller's view that that floating floating exchange exchange rates rates have have inhibited international trade. trade. This issue has has been studied fairly exex- tensively, and there is not one one study which has found that that floating trade. rates have had any dampening dampening effect whatsoever on world trade. But dede- spite that, there may be some truth in in this this particular belief. All studies so so far undertaken have looked at the effect of of the exchange rate regime on on international trade as aa whole. Recent 1‘See Wood and Nancy Aamon Ji anakopl os, "Worldwide see Geoffrey Geoffrey E. Wood Jianakoplos, “Worldwide Economic Expansion: Are Convoys or Locomotives the Answer?” Answer?" Federal Reserve Bank Bank of of St. Louis Review, July 1978. —296— -296- theoretical empiritheoretical work by Ronald Ronald McKinnon,22 supported by by forthcoming empirical work by Stephen Carse, John Williamson and the present author,33 cal work by Stephen Carse, John Williamson and the present author, this is not appropriate. suggest that this AA substantial part part of international trade is in primary primary or or semi— semimanufactured goods. sanufactured goods. The prices prices of of such goods are continually held held close together together across countries by arbitrage. Thus traders in in such goods are not affected by by exchange exchange rate fluctuations provided provided that that they hold inventories equal to to their indebtedness arising from trade trade -- and -— the evidence is is that to aa first approximation approximation they do. There is therethere- fore no reason to expect trade in in these these goods to be in any way affected by exchange rate changes, whether or not not these changes are anticipated. In contrast, manufactured manufactured goods do not have their their prices prices quickly arMarbi4 traged goods are traged into into equality equality internationally. internationally. 4 Traders are Traders in in such such goods therefore exposed to exchange risk. Tests for the effects of exchange Tests rate rate fluctuations on on trade should focus on these these categories of goods, goods, rather than than on trade as aa whole; looking at trade as aa whole may have led to the concealing of the effect of of exchange rate fluctuations on aa sub-section of trade. (This hypothesis is currently being explored by by the present author, but no no results sufficiently sufficiently reliable to report are ~t present available.) ,t 2Ronald McKinnon, Money e, Oxford UniverMcKinnon, in International Exchan Exchange, University Press, New New York, 19 1979. 3Stephen Geoffrey E. E. Wood, Financing stephen Carse, John Williamson, and Geoffrey ~c~!jjcfpj~iLftad~, 'ractices in U.K. Foreign Trade, Cambridge University Press, Cambridge, 1980. 1980. 4See Williamson “The British 1i ams on and Geoffrey E. Wood, "The see e.g. John Wil [nflation: Inflation: Indigenous or Imported?”, Imported?", American Economic Review, September ,eptember 1976. -297— -297- absence of confirming evidence, evidence, Dr. Or. Heller may be So, despite the absence correct in saying that trade has been inhibited by by exchange excrange rate fluctfluctuations" uations. But three noints points shou·ld emphasized. Should be emphasized. evidence to support his assertion. evidence provides no First, he provides Second, he should have compared compared Second, what would have have happened to trade under aa fixed rate system defended defended against the consequences of divergent divergent inflation rates by by proliferating controls, with the effect of of exchange rate fluctuations on trade. Third, even if he he is correct that exchange rate fluctuations inhibit trade, it it is far from clear that official exchange market intervention intervention is thereby justified. His second major assertion is that flexible exchange rates have “not brought about a climate for the conduct of "not of more more effective effective stabi— stabi- lization policies.” policies." The The only possible response response to that is to ask why on earth they should. Under Under a fixed exchange rate system, system, the burden by any country’s country's government was was in of mistakes in stabilization policy by part borne by the foreign foreign sector. Excess Excess demand demand was in part met by foreign supply, supply, while deficient deficient home demand was in in part part offset by demand from overseas, so long long as the demand and supply imbalances were at least partly partly due to monetary policy. is the the (An example of this is United Kingdom experience in l96Os; see Williamson in the 1960s; Williamson and and Wood, op.cit.) Floating exchange exchange rates, by eliminating eliminating flows across the close this this safety valve; one should therefore therefore expect foreign exchanges, close of stabilization equal) that the performance of (other things being equal) stabilization policies should deteriorate rather than improve under under floating rates. But wHte very But Dr. Heller did not write very precisely precisely at this point; point; he does not say exactly what 1;hat he he means by the “climate "climate for the conduct of more effective effective stabilization stabilization policies.” policies." -298- He may may mean not the actual achievement of such policies, but rather rather how policymakers policymakers have reresponded to divergences of the economy from its its desired path. is what he means, then then he is pretty clearly wronq. example. If If that The U.K. is aa good It It was was only after after the collapse of sterling’s sterling's foreign foreign exchanoe exchanoe value in 1975 that the U.K. government took any serious measures measures to end the gradually gradually accelerating accelerating inflation of of the the previous twenty years. Why Why they so responded can only be conjectured; but but the explanation may be be that floating exchange exchange rates bring home to the electorate electorate the costs costs of inflationary policies rather inflationary policies rather more more quickly quickly than than did did fixed fixed rates, rates, and and next election. thus may may influence their voting behavior at at the next Dr. Heller next claims that floating exchange exchange rates have not “decreased the cost cost of [foreign exchange exchange market] of [foreign market] intervention intervention to to cencen"decreased the tral banks.” banks." Dr. Heller is really very careless in his use of Or. of the word “cost.” cost." 11 He never tells us what costs he he has has in mind mind in in the present present inin- stance. It is certainly certainly clear, however, that the amounts of interveninterven- tion have been large, large, and and it is on this issue rather than than the undefined one raised by Dr. Or. Heller that that we next comment. Why have exchange exchange rates been so volatile? Where Where have have the private stabilizing speculators speculators been? stabilizing been? these questions. Dr. Heller does not Dr. Heller does not attempt attempt to to answer answer Fortunately, an answer has has been provided by aa large body of previous work. rate volatility is, in large part, Exchange rate the consequence of volatile volatile national monetary policies. policies. This has been l97Os; it was al also 192Os. true not just in the the 1970s; so true in the 1920s. The The concon- clusions of aa recently published paper by my colleague Roy Batchelor summarize the evidence evidence very very well. —299— -299- Stable inflation rates are all all that that is required required to to keep the trend in exchange steady..., efficient exchanqe marexchange rates steady_... markets should keep fluctuations around the the trend within within the same exsame margins as in the the 1920s. 1920s. What What is necessary necessary for exchange rate stability is that monetary expansion expansion be be predictpredictable.. able ... 5 The expressed in the quotation The reason reason for this is admirably expressed quotation from Gustav Cassel with which which Jacob Jacob Frenkel Frenkel concludes concludes the paper paper he he presented at this conference. The international valuation of aa currency currency will, then, then, generally show aa tendency to anticipate movements, so so to speak, and become more an expression of of the internal internal value that the currency is expected to possess in aa few months, 6 to or perhaps year’ss time, or perhaps in in aa year time.6 1 The The more volatile is aa nation’s nation 1 s monetary monetary policy, policy, the more frefre- expected future internal value of of its currency change, change, quently will the expected and so the more frequently will its its exchange exchange rate change. The primary primary source of exchange rate volatility is therefore volatility volatility in in national monetary monetary policies. Understanding that is is central to drawing drawing the the corcor- rect lessons for future policy of the exchange rate experience experience of of the 1970s. of private private Understanding that also helps explain the absence of stabilizing speculation; because of of the volatility of national national monetary expolicies, speculators have had had very very little little basis on which which to form ex- pectations of future exchange exchange rates. In this this context, it is worth pointing out out that that (as (as Jacob Jacob Frenkel Frenkel shows) exchange shows} exchange rates have been no more more volatile volatile than than prices in other 5Roy Batchelor, "Must “Must Floating Exchange Rates be Unstable?" Unstable?” Annual Monetary Monetary Review, Centre for Banking and and International Finance, Finance, The City University, University, London, England. England. 6Gustav Cassel, ~ 199, pp. Money and foreinExchanges Foreign Exchanges after 1919, pp. 149149150, Macmillan, Macmi 11 an, London, 1930. -300-300- asset markets, thus emphasizing the common cause ,sset cause of such volatility. volatility. rurther, stressed that 'urther, it it must must be be stressed that D. D. Heller’s Heller's belief belief that that “speculative "speculative ictivity may ,ctivity may well accentuate rather rather than than reduce exchange rate fluctuafluctuations” tions" is totally contradicted by evidence that that there are no traces traces of ;peculative runs in the foreign exchange markets.77 ;peculative “runs” 11 11 His last assertion is “fostered His last assertion is that that floating floating exchange exchange rates rates have have "fostered :he decline decline of the dollar as the world’s world's leading currency.” currency." By this this he ieams that floating exchange 1eans exchange rates have have led to aa fall in the proportion )f assets in of individuals individuals and if dollar-denominated dollar-denominated assets in the the portfolios portfolios of and :entral banks. He is clearly right. Portfolio Portfolio diversification was to iee expected as aa consequence of the move to floating rates, and it has indeed happened. But so what? Why Why is that that undesirable? Nowhere does Jr. Jr. Heller answer these questions. U.S. POLICY POLICY RECOMMENDATIONS Turning first to his recommendations for the future conduct of 1.5. J.S. policy, these are manifestly sensible -— they comprise recomendrecommend- ng the announcement of of intermediate monetary ranges targeted by base :ontrol so :antral so as to ensure hitting hitting them. them. The empirical and theoretical iork on the causes of exchange rate volatility, which was referred to ,ark ?arlier, clearly clearly indicates such aa policy exchange rates •arlier, indicates that that such policy would would make make exchange rates iuch less erratic in their movements, and such 1uch such aa policy would also also help tabilize the U.S. economy ,tabilize economy as aa whole. 77See for example Donald S. Kemp, “The U.S. Dollar in InternationDonald S. "The Dollar in Internationsee tl,1 Markets, mid-1970 mid—1976,” Federal Bank of Markets, mid1970 to to mid-1976," Federa 1 Reserve Reserve Bank of St. St. Louis Louis teview, August 1976. -301-301- INTERNATIONAL POLICY POLICY RECOMMENDATIONS RECOMMENDATIONS INTERNATIONAL Dr. Heller does not advise aa return to pegged exchange exchange rates; rates; he recognizes that so long as national inflation inflation rates are as as diverse as they they currently currently are such aa move would would not be be sustainable. He does, howhow- ever, encourage encourage official intervention in in the the foreign exchange markets. markets. There are, are, as Dr. Dr. Heller recognizes, costs to such intervention-intervention—in particular, particular, there may very well be an impact on domestic monetary policy. steady and predictable money growth growth is is the the foundation of Since steady reasonably stable exchange rates, there are considerable risks that central bank foreign exchange intervention would buy only short term stability. And what are the benefits of of exchange exchange rate stability stability achieved by official intervention in in the foreign exchange markets? markets? What What can justify justify official intervention? Central banks do not in general have any better better knowledge than does the future course course of economic economic variables. does the private sector of the There can be be occasions when when they do do have such knowledge -- because they -- know their own intentions but have have not published them, or because they are privy to the otherwise otherwise undisclosed intentions of of aa foreign central bank. fluctuIn that case, intervention to prevent aa temporary temporary market market fluctu- nay be justified ation may justified but but such intervention is is inferior to to making the confidential knowledge on which which it is based. public the Making the Making central bank’s bank's intentions intentions public would help stabilize not just the differing degrees, every every other market. foreign exchange exchange market but, to differing Publicity, Publicity, therefore, clearly dominates intervention. A second defense of occasional intervention nay A may exist if if it it is is catefound that fluctuating exchange exchange rates do, indeed, inhibit certain certain cate- gories of trade. If stable stable national monetary monetary policies are being —302— -302- pursued, there may still appear to be aa case for intervention. The case would be that that some of of the benefits from exchange exchange rate rate stabilizastabilization accrue not as profits to speculators speculators on the foreign exchanges, but to traders in goods. There would, would, in other words, be aa divergence between the private arid social social benefits of stabilizing speculation, with the social social benefits outweighing the private ones, ones, thus appearing appearing to to justify intervention. intervention. second best. best. But here, too, exchange exchange intervention is As has emerged from the literature on protection, aa direct subsidy paid to the affected efficient means affected sector is the most most efficient 8 8 of In of assisting aa sector of an an economy. In the present present case, intervening in the exchange markets would mean that that all traders traders in international money noney markets, not only those in in goods affected by fluctuating exchange rates, were being assisted. Here, too, then, then, while exchange exchange market market intervention may may conceivably be justified justified -- although although the evidence -— Nhich may justify is not justify it it is not yet yet in in -- again again the the policy policy is is aa second second best best •hich may —— one. ::me. Two further possibilities remain. remain. An An exchange exchange rate rate may be be changing very rapidly -- sterling in the three months to :hanging to July July 1979 is —- an example. This This was imposing very rapid adjustment adjustment costs on on indusindus- tries already required to respond to aa substantial substantial change in in the oattern of comparative advantage. ,attern If If the authorities in in such aa case can slow monetary control, then there :an slow the adjustment without loss of monetaryc2p~pj,then are ire benefits benefits from from their their doing doing so. But But the the situations when when they can can do do 8See “Domestic Distortions, see J. Bhagwati and V. K. Ranaswami, Ramaswami, "Domestic rariffs, and the the Theory Theory of the Optimum Subsidy,” Subsidy," Journal of Political conomy, February 1963, and Geoffrey E. Wood, ,conomy, l✓ ood, “Senile "Senile Industry ProtecProtection,11 Southern Economic Economic Journal, January 1975. 1975. tion,” -303- so are manifestly rare. The U.K. was able to do so in that episode bebe- cause aa large part of of the inflow seemed seemed to have resulted from aa desire to buy just the kind of of securities the the U. U. K. government would have have had to sell to sterilize the inflow, but the experience experience of Germany in the l960s and l970s 1960s 1970s shows that such such episodes episodes are unusual. This This case, case, then, then, does constitute aa modest modest defense of occasional intervention -- but the —— circumstances are very special. (And there will still still be be aa welfare cost to the nation if the rate of of return earned on international reserves falls short of the rate paid paid on on foreign-owned national national debt.) The fourth, and last, defense defense is is when there is is an increased increased demand on the part of of non-residents to hold the money of some country country -— — not, it should be be stressed, stressed, assets denominated in in that currency, but the currency itself, including of course course bank deposits. deposits. This does does not invariably constitute constitute aa reason for supplying the currency; it may, rather, often often be be an opportunity for reducing reducing the inflation rate. rate. If, however, inflation is at its desired rate, them then the increased demand for currency must be be met by by an increased increased supply, and the simplest way way to be be sure sure of supplying the correct amount is to to operate operate on the the foreign exchange market. But this is aa very special case indeed. Summarizing then, the case for official intervention in the foreign exchanges is is very weak. Recognizing that there can can be substansubstan- tial fluctuations of exchange rates about their equilibrium equilibrium values values does not imply that these fluctuations should be corrected corrected by by official official intervention. Dr. Heller is also concerned about the appropriate appropriate reserve asset for the international monetary system. —304- He believes that the currently currently evolving reserve asset system is inherently unstable, and that it should be be replaced by a single asset system, the the asset being being either either the U.S. dollar or aa somewhat modified SDR (Special Drawing Drawing Rights). It is deal first of aa dollar It is convenient convenient to to deal first with with his his endorsement endorsement of dollar standard. The weakness of such aa system was first diagnosed diagnosed by Robert Triffin.99 His diagnosis can be be summarized very briefly as follows. The reserve asset, the dollar, can be supplied only by the reserve centre, the United States, running running continual continual deficits in its balance confidence in in the of payments -- but that progressively undermines confidence -- reserve asset which is being thus supplied. supplied. nally inconsistent. Such interSuch aa system is inter- Dr. Heller provides provides us with with no reasons for thinking Triffin to be wrong -- indeed, indeed, nowhere does does he refer to Triffin so so his advocacy of of a return to aa dollar standard cannot be taken seriously. The defect with his endorsement of of an SDR-based SOR-based system is that under one set of circumstances circumstances the scheme is unnecessary, while under under the alternative circumstances it it will not work. An international monemone- tary system with all major major currencies serving as as reserve assets is is not, despite his belief belief to the contrary, contrary, inherently unstable. Such aa system will not not be be continually continually destabilized by capital flows responding to reasoninflation differentials -- so so long as these differentials are reason-- ably stable and predictable. And And when these differentials are not stable and and predictable, there will be be sudden and large movements of 9Robert Triffin, Gold and the Dollar Crisis, Yale University Robert Triffin, Gold and the Dollar Crisis, Yale University Press, New Haven, 1960. —305— funds from currency to currency whatever the official reserve asset of the system may be. be. of the international monetary Tinkering with the reserve asset of system cannot substitute substitute for stable stable domestic monetary monetary policies. CONCLUSIONS CONCLUSIONS The lessons for the conduct of international monetary policy policy which have have been been provided provided by by the experience experience of the 1970s can be stated very very briefly. briefly. Exchange rates will be volatile so long Exchange rates will be volatile so long as as national national monetary policies policies are volatile. volatile. It is not clear what harm this exex- change rate rate volatility volatility does, although although the underlying monetary instainstability does cause considerable Meltzer’s paper shows. considerable harm as Alan Meltzer's shows. In any any event, the case for exchange market intervention to to reduce this this volatility is very circumscribed circumscribed indeed. Nor can can any case be be made for trying to prevent prevent portfolio portfolio diversidiversification into into aa range of reserve reserve assets. A multiple A multiple asset system system will will be stable if if national monetary policies policies are stable, stable, and if national monetary policies are unstable then then any international monetary system system will inevitably be unstable also. The l97Os experience of The lesson of the the 1970s of floating rates, as of every earlier floating exchange rate episode, is that the international monemonetary system will only be as stable as the set of of national monetary systems systems which it it links. links. -306- l980s MONETARY POLICY ISSUES FOR THE 198Os Lawrence K. Roos Roos As one of of the sponsors of this conference, it is aa special pleasure to to welcome welcome all all of of you you to to the the Federal Federal Reserve Bank Bank of St. St. Louis. Louis. It is aa privilege, privilege, as well, to have the the opportunity opportunity of joining joining you you in might learn learn from past experience experience in in planning monetary pondering how we might policy for the the future. In the the tine time allotted me, II would like like to share with you some imimpressions of past policymaking that I, in my four years years as president of of the Federal Federal Reserve Bank of St. Louis have gained, and to explore with with you what II believe we we might look look forward to in the years ahead. ahead. Looking l970s, I would be less if II did Looking back in the 197Os, less than than candid ·if not admit admit to deep feelings with the not to some some deep feelings of of frustration frustration with the way way in in which which monetary policy has been underbeen conducted, as as well as to to aa failure to to under- stand how policies which which produced such such adverse consequences managed to persist for so extended aa period. Perhaps the best best way way to express my my feelings is is to to focus focus on few fundamental concepts which feelings on aa few fundamental concepts which have have come come to to doninate my own understanding of the impact of monetary policy on the dominate economy. First, and foremost, is the concept that inflation inflation is fundamenfundamentally aa monetary phenomenon. phenomenon. This is an extraordinarily appealing appealing notion to to me, if for no no other other reason than its generality and sheer sheer Mr. Roos is President of the Federal Federal Reserve Bank of St. Louis. Louis. —307— -307- simplicity. As Beryl Sprinkel recently noted: “It doesn’t "It doesn't take a genius to to know know that if you pump more and more money into into the system, you get inflation." inflation.” Now, I suppose that if Beryl is correct correct that it truly does not take aa genius to understand this, then there is still hope that this this concept will come to be be widely accepted. Unfortunately, Unfortunately, the time time lag necessary for acceptance of this this appears to to exceed, by aa considerable margin, the the time lags with which changes in money affect prices. Because the full impact of changes changes in the rate rate of growth of of money on the inflation rate occurs over over aa considerable period of of time (esti(estimated mated variously variously from three to six years), years), it it is important that monetary policymakers, as well as the general general public, clearly understand that 11 11 the 11“core” inflation rate, core 11 inflation rate, or “underlying” underlying 11 inflation inflation rate, or “basic” basic 11 inflation rate (to mention just aa few of the terms that have been inflation attached to it) it) is determined by the long-term long-term trend rate of of growth in in money after adjustments for changes changes in money money demand. In fact, because long-term long-term changes changes in velocity velocity have, roughly, roughly, had equal and offsetting impacts to that for changes in in output, the the core inflation rate rate is essentially essentially equal equal to to the the trend trend growth in in money. money. BeBe- cause the trend rate of growth growth of of money money has has approximated approximated seven seven percent, 11 the current “monetary—induced” monetary-induced 11 rate rate of inflation is is about seven perper- cent. To put it somewhat differently, had there been been no oil oil shocks shocks or non—monetary indOced other exogenous non-monetary indUced impacts impacts on prices, prices, we we would would nevernever- theless be be currently faced with an inflation inflation rate of about seven perpercent due solely to the growth in money money that has emerged from past monemonetary policy tary policy decisions. decisions. -308- AA careful understanding of the difference between between the actual ininflation rate rate and the monetary-induced monetary-induced or core core rate of inflation inflation is crucial for the proper conduct of monetary policy. Only the monetary— monetary- induced rate rate of inflation should concern monetary policymakers; it it is the only component of of inflation that they can influence. influence. Exogenous shocks such as those caused by by higher energy prices or crop crop failures will will always contribute contribute to to the the current measured inflation inflation rate, rate, but but their impact is transitory. Attention paid to these these exogenous influinflu- ences on prices prices must never divert monetary monetary policymakers policymakers from focusing their actions actions toward controlling, controlling, and reducing, the monetary-induced monetary-induced rate of inflation. A second key concept that has guided my understanding A understanding of of the imim- pact of monetary policy is that abrupt and substantial changes in the pact growth of money, if sufficiently sufficiently prolonged, prolonged, have dramatic dramatic and usually unfortunate consequences for the economy. Unusually rapid growth in in money, if sustained for several quarters, while having some some positive positive effects on employment and output for aa short time, will will ultimately ultimately and inevitably increase the monetary-induced rate of of inflation. inflation. Similarly, Similarly, unusually slow growth in money, if sustained for several several quarters, quarters, will unusually result in reduced growth in in output output and employment -- perhaps, even even aa -- recession, recession, and ultimately ultimately reduce the monetary—induced monetary-induced rate of of inflation. inflation. Careful understanding of the short-run consequences consequences of of sharp proper conduct conduct fluctuations in the growth of money is crucial for the proper of monetary policy. of To avoid undesirable undesirable results such such as as recession or or To an an over—heating over-heating of of the the economy, monetary monetary policymakers policymakers must avoid policy policy capricious changes in the growth of actions that result in sudden or capricious -309- money. They They should, should, instead, instead, conduct conduct policy policy In in such a way way that that changes changes in systematic and and gradual gradual. in the growth of money are systematic A third concept that has guided my understanding is that the A growth of money can best be on the behavior be controlled, not by focusing on of interest rates, but by by controlling the growth of the monetary monetary base. Since Since the the Federal Federal Reserve controls the largest component component of of the monemane- tary base tary —— credit -- growth of of the monetary monetary base is Federal Reserve credit -— directly and completely in the hands of the Federal Reserve. Similarly, Similarly, is considerable considerable evidence that that the multiplier multiplier linking the monetary there is base to to the money money stock is sufficiently stable and and predictable predictable to assure aa reasonably close relationship between growth of of the base base and growth of money money over over all but but the shortest-term shortest-term period. Consequently, the the lesson lesson for policymakers is that, if control of the growth of money is to be be a crucial part part of monetary policy, desired money growth rates should be be linked directly in in the policy process process to the qrowth of of the monetary base. base. Finally, aa fourth concept which has enabled me to understand the impact of monetary monetary policy on on the the economy is that economic markets, markets, especially the financial and foreign exchange markets, markets, are reasonably rational and efficient. efficient. Thus, increased rates of money growth tend to of the dollar on produce higher interest rates and to lower the value of international exchange markets as soon as the financial market market partici participants, pants, who seem to to be be well aware of the association between money come to expect increased future inflation rates. growth and inflation, come It follows that, while so-called “tighter” "tighter" monetary policy may immediimmediately produce higher interest rates, the same result occurs with “looser” looser monetary policy in the longer time time span. span. 11 11 -310-310- Interest rate movements per se are unreliable guides to policy. This is especially true true when we consider that interest rates, rates, which which represent represent the price of of credit, are also affected by aa host of of non—nonetary non-monetary influences. the above concepts is especially complex and certainly certainly Now, none of the none is likely to to be either either new or controversial to to most most of you. HowHow- assessing the likely ever, they do provide an analytical framework for assessing results results of monetary policy actions. actions. It is this basis of analysis that has ny frustration in viewing what has happened has led led to to my happened over the past four years. years. No one, who who believes as as II do that the most most significant component of inflation is is monetary, could could have failed to to have have been been conconcomponent cerned cerned with growth in money that accelerated from five percent over over the period 1/73 to 111/76 to eight percent percent from 111/76 to 111/78, 111/78, period from 1/73 core rate of of inflation. guaranteeing a significant increase in the core No No one, who believes as II do that drastic changes in the qrowth of money money one, produce undesirable economic economic consequences, could have failed to to be con concerned when the the money stock, having grown at the rate of eight percent for two years, suddenly dropped to aa less than two percent growth for the period from September September 1978 1978 to to May 1979, virtually virtually assuring aa major economic economic slowdown. ~nd, And, certainly, no no one, one, who believes as II do that financial markets are rational rational and efficient, could fail to be be disdistightness’ turbed by by the the current current expressions of concern with alleged "tightness" of of monetary monetary policy, as judged by the ‘high’ "high" levels of of nominal interest interest rates. Money growth at rates approachinq approaching 10 percent and an inflation inflation rate of close to to 10 10 percent percent are certainly not reflections of tightness tightness. Certainly Certainly the financial financial and foreign exchange market participants participants have not been fooled; witness the behavior of of interest interest rates and the value of the dollar over the last few months. months. —311— -311- But my frustration is not confined only to the unfortunate conconsequences of past past monetary policy actions. It It also lies with the monemoneobtary policymaking process itself that produced the results we have ob- served throughout the l970s. 1970s. Time and time again, again, II have have observed observed the Reserve’s interest rate target while achievement of the Federal Federal Reserve's while money “desired” target growth was permitted permitted to to wander at will outside its !ldesired target 0 ranges. As II noted in in an earlier earlier discussion in London last last June, the monthly “betting odds” during "betting odds" during the past four and aa half years years have been been only only about one in two that that Ml Ml would remain inside its its target range. Moreover, there is little little doubt that the conduct of of monetary monetary policy, by of interest rates, rates, has produced aa procyclical focusing on stabilization of pattern in the growth in money. That That pattern has tended to exacerbate cyclical movements and exogenous the impact of of cyclical exogenous shocks shocks on the economy. But, But, again, none of this is especially new to you. Many Many of of you have have contributed contributed over over the the past decade decade to to studies critical critical of both both the monetary policymaking process and and policy policy consequences. I, too, too, have been convinced, both exarguments to which II have been been exboth by the economic arguments posed, and by by aa first-hand view of of the disappointing disappointing results of of the policies pursued, that only aa major major change in in the formulation of monemone- tary policy -- away away from concentrating on stabilization stabilization of interest —- rates and towards focusing on the monetary monetary base —- would enhance the prospects successfully achieving achieving the results we desire desire from from monetary monetary prospects of successfully policy. The announcement announcement by by Chairman Volcker on on Saturday, October 7, that the Federal Reserve is changing its procedures of monetary monetary policymaking policymaking controlling growth of of the reserve reserve aggregates to place more emphasis on controlling while while permitting interest rates to fluctuate freely, represents a gaint —312— -312- step In in correcting past mistakes. There is no doubt In in my mind that if this new approach is effectively implemented implemented in the upcoming months and years, we can can achieve control control over the the growth of money money and, and, conseconse‘basic” rate of inflation. quently, control over the the "basic" Similarly, we can avOid the avoid the adverse real sector impacts that have resulted from uninunintended drastic short-run short—rum fluctuations in the growth of money around its its longer—run trend rate of longer-run of growth. Finally, once the financial market participants are convinced convinced that we have indeed seized control over the the growth of of money and intend to bring about the gradual reduction in in money money growth necessary necessary to reduce the the core inflation inflation rate, II believe that we will see an end to the the surges surges in interest rates amd and declines in the value of the dollar dollar which which have proved so so troubling troubling in in the past. Thus, as you enthusiyou may have inferred inferred from my my comments, II am am enthusiastic and encouraged about the change in the policymaking process that has occurred. However, my euphoria is restrained by by aa realization that several problems still remain if this change in policy is to produce results. the hoped—for hoped-for results. To assure maximum effect from the Fed’s Fed's new policy the following steps must be be taken: taken: 1) 1) Instead of placing placing sole emphasis on controlling controlling the growth of policymakers should focus also of non-borrowed reserves, policymakers on on growth in the monetary monetary base and and total total reserves. reserves. There There just too many many slips slips twixt growth in non-borrowed reare just non—borrowed reserves and growth in in money. 2) Policy emphasis must be firmly and fundamentally rediredirected from concern about movements movements in the Federal funds rate to concentration concentration on on growth in in the monetary base base and, and, hence, the money money stock. stock. The substance substance of policy must must go -313—313— beyond merely widening the permtsslble permissible range of movements in in the Federal Federal funds rate. rate. For, if if widened Fed Fed funds funds rate constraints remain even even remotely binding, binding, monetary control cannot succeed. 3) The new method must be given given adequate adequate time time to prove itit- self. The success of the the new monetary control proceproce- dure cannot be reasonably evaluated by observing money stock behavior over aa short time span. Not Not even the most most ardent academic academic advocate of of base targeting targeting asserts that that precise money money control is possible over over aa period of of six six months or less. less. At the very least, least, aa one year testing testing period is necessary for any comparison between between previous methods and the current current one. Moreover, no one should should inflation to dissipate in a matter of months. expect inflation InIn- flation has has been generated over over aa period of of 15 15 years and cannot be eliminated overnight. It It would be be tragic if this new approach to policymaking were were to to be be tried and abandoned after aa short time because of false expectaexpectati ons. tions. 4) Finally, and perhaps most importantly in the short run, the procedures for implementation of of the new policy, the rules of the game, must be be clearly enunciated to the public. As we have observed observed during the first week after announcement of the new new approach, the lack lack of clearly clearly announcement articulated rules rules produced produced aa near panic in in financial financial marmar- kets. There There is no reason reason to shroud policy in secrecy and to to keep keep markets guessing. -314- While While surprises might might have had some value in policies directed mardirected toward money money market counterproductive when ket stabilization, surprises are counterproductive when monetary aggregates become the target. Above all, all, the attention of of policymakers must be be focused on the longer—run impacts of policy. longer-run Unfortunately, as as Arthur Arthur Burns noted in his Per Jacobsson Jacobsson Lecture, Lecture, the “anguish of of central central banking’ his Per the "anguish banking" has has often often cone come from the the short—term short-term political political pressures pressures on on monetary authorities -- pressures for—bad-and—for—worse, the monetary authorities pressures to which, for-bad-and-for-worse, have all too often succumbed. What is needed more than ever ever before is a steady hand on the tiller of monetary policy. subNot only will the Fed’s Fed's new new policy be be sub- jected jected to critical analysis by those who traditionally traditionally have have doubted doubted the feasibility of of monetary control; the the very credibility credibility of this country’s country's central bank is at stake. II trust that we will have the wisdom wisdom to to implement our policy effectively, the openmindedness openmindedness to judge our our propro- the courage to resist whatever pressures might arise arise gress fairly and the to retreat retreat from the historic historic step step we have have taken. taken. —315— -315-