The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
May/June 1991 Vol. 73, No. 3 3 Pu blic C apital and P riv a te S ector P e r fo r m a n c e 16 S u p ervision o f U n d e rc a p ita liz e d Banks: Is T h e r e a Case fo r C hange? 31 T h e FO M C in 1990: O nset o f R ecession 53 U.S. P o lic y in th e B retton W o o d s Era THE FEDERAL > RESERVE X IIA N K of ST.I jOUIS 1 F ederal R eserve B ank of St. Louis R e v ie w May/June 1991 In This Issue . . . A decline in the grow th o f U.S. productivity and international com petitiveness has been caused by deficient public capital formation, accor ding to some public policy analysts. These analysts suggest that a sharp rise in public capital form ation w ill be necessary to restore earlier trends in productivity growth. In the first article in this Review, John A. Tatom examines the production function estimates that provide the underpinning fo r these arguments. The author shows that simply accounting fo r the influence o f energy price movements on productivity and fo r the slowing trend rate o f technological change in the typical production function fram ew ork reduces recent estimates o f public capital's effect on private sector out put by m ore than half. M ore important, Tatom argues, such findings are examples o f “ spurious regression bias,” w h ere variables appear to be statistically significantly related but are not. He explains this statistical problem and shows how it applies to tests o f the public capital hypothesis. W hen the private sector production function is appropriate ly estimated, Tatom explains, the public capital hypothesis is rejected; the public capital stock has no statistically significant effect on business sector output. The author concludes that the pace o f public capital form ation has played no role in accounting fo r movements in U.S. productivity. * * * Concern over the increasing number o f bank failures and the resulting deposit insurance fund losses has led to numerous proposals fo r bank regulatory reform . One recent proposal calls fo r bank super visors to undertake "prom pt corrective action” to achieve stronger en forcem ent o f bank capital requirements at undercapitalized banks, in cluding restrictions on their asset growth, shareholder dividends and in sider loans. In the second article in this Review, "Supervision o f Undercapitalized Banks: Is Th ere a Case fo r Change?” R. Alton Gilbert examines banks whose capital ratios w ere below the specified minimum fo r periods longer than one year during 1985 to 1989 to determine w hether they generally violated the constraints that w ould be imposed under prompt corrective action. He finds that, w hile most long-term undercapitalized banks did not violate these constraints, a substantial m inority did so. He also finds, how ever, that there was no significant difference in the recovery rates betw een the banks that violated the "prom pt corrective action” constraints and those that did not. * * * In the third article in this issue, "The FOMC in 1990: Onset o f Reces sion,” James B. Bullard presents this Review's annual synopsis o f the re- MAY/JUNE 1991 2 cent actions o f the Federal Open Market Committee, the prim e policy making group in the Federal Reserve System. Calendar 1990 was an in teresting period from the perspective o f m onetary policy-watchers because the national econom y slipped into recession in the latter half o f the year, thus providing an opportunity to study policy-making during the onset o f recession. In the context o f a chronology, the author em phasizes some difficulties that the Committee experienced in attempting to implement its stated short-run policy goals. These include the p ro blems o f inaccurate short-term forecasting and the appropriate measure ment o f the policy stance. * * * In the final article o f this issue, Allan H. M eltzer, John M. Olin Pro fessor o f Political Economy and Public Policy at Carnegie Mellon U niver sity, discusses U.S. international economic policy during the Bretton W oods era. His article, which was presented in St. Louis on April 8, 1991, as the fifth annual Hom er Jones Mem orial Lecture, review s the roles o f the U.S. as the w o rld ’s strongest econom y and the dollar as its reserve currency in a system o f fixed exchange rates. Key within this review is the lack o f consistency betw een U.S. m onetary policy, which ultimately (although unofficially) was responsible fo r maintaining the fixed rate system and domestic policy objectives that often ran counter to this goal. By noting this conflict and other tensions within the Bretton W oods arrangements, M eltzer is able to trace the steps that led to the collapse o f the fixed rate regim e in 1973. FEDERAL RESERVE BANK OF ST. LOUIS * * * 3 John A. Tatom John A. Tatom is an assistant vice president at the Federal Reserve Bank of St. Louis. Kevin L. Kiiesen provided research assistance. Public Capital and Private Sector Performance A GROWING BODY o f literature now argues that the public capital stock has significant, positive effects on private sector output, pro ductivity and capital form ation.1 Most o f this literature suggests that a decline in the grow th o f the public capital stock since the early 1970s caused a "productivity slump" in the private sec tor low ering profitability and investment.2 Unless these trends are reversed, say the studies, the nation’s standard o f living w ill be further threatened. This article explains this public capi tal hypothesis and evaluates the evidence sup porting it. THE PUBLIC CAPITAL HYPOTHESIS AND PRIVATE SEC TOR PRODUCTIVITY Public capital comprises federal, state and local governm ent capital goods. It includes high1 This argument will be referred to as the public capital hypothesis; it has been developed most fully by Ratner (1983), Aschauer (1989a), (1989b) and (1990) and Munnell (1990). Also see Deno (1988) and Eberts (1990). The hypothesis was suggested earlier by Schultze (1981), Ar row and Kurz (1970), Eisner (1980) and Ogura and Yohe (1977). 2See The National Council on Public Works Improvement (1988), Malabre (1990) and Reich (1991) for analyses that attribute such consequences to the slowdown in public capital formation. A previous article, Tatom (1991), ex plains how several factors account for the decline in the rate of growth of the public capital stock. These factors suggest that a reversal of this past decline would not be ways, streets and roads, mass transit and air port facilities, educational buildings, electric, gas and w ater supply facilities and distribution sys tems, w astewater treatm ent facilities, and ad ministration, police, fire, justice and hospital facilities and equipment. The public capital hypothesis is that the stock o f public capital raises private sector output both directly and indirectly. The direct effect arises, according to the hypothesis, because pub lic capital provides intermediate services to pri vate sector firms, or the marginal product o f public capital services in the private sector is positive. The indirect effect arises from an assumption that public and private capital are "com plem ents” in production—that is, the partial derivative o f the marginal product o f private capital services w ith respect to the flow o f public capital services is positive.3 Thus, a rise in public capital raises the marginal productivity o f private economically justifiable, even if the public capital stock has the effects emphasized by the public capital hypothesis. 3The notion of complementarity and substitutability used here has been called q-substitutability and q-complementarity. It refers to the effect of the quantity of one resource on the marginal product of another resource. The concept of p-substitutes or p-complements is more common; these terms refer to the effect on the demand for a resource of a rise in the price of another resource, holding other resource prices and output constant. See Sato and Koizumi (1973) for a discussion of this distinction, or its use in Tatom (1979b). MAY/JUNE 1991 4 Figure 1 Business Sector Output per Worker and Public Capital Stock per Worker Thousands of dollars per worker (1982 prices) 22.5 Thousands of dollars per worker (1982 prices) 7.5 1947 53 59 65 71 1989 Note: capital stock is measured from the end of the previous year. capital services so that, given the rental price o f such services, a larger flo w o f private capital services and a larger stock o f private assets p ro ducing them are demanded. The rise in the m ar ginal product o f capital increases private capital formation, fu rth er raising private sector ouput. The indirect effect o f a rise in public capital on private output, how ever, is not necessarily positive. In fact, this effect is negative if public and private capital are substitutes. Economic theory does not dictate w hether private and public capital are complements or substitutes.4 The analysis below focuses on estimating the direct effect, the private sector’s marginal p ro duct o f public capital. I f public capital does not enter the production function fo r private out 4There is, however, a growing literature that suggests that government spending is a substitute for private sector spending. Recent attention to this view owes much to its elaboration by Aschauer (1985). Other research that develops the direct substitution channel for crowding out include Kormendi (1983), Kormendi and Meguire (1986) and (1990) and Tatom (1985). Critics of the Kormendi and FEDERAL RESERVE BANK OF ST. LOUIS put, as is demonstrated below, the sign o f the indirect effect must also be zero. The Productivity Decline and Public Capital Formation: A Look at the R ecord Advocates o f the public capital hypothesis argue that a slowdown in public capital form a tion caused a "productivity slump” beginning in the early 1970s. Some perspective on this issue is provided by figure 1, which shows output per w ork er in the business sector and the real nonmilitary net stock o f public capital (1982 prices) per business sector w o rk er from 1947 to 1989. Public capital per w ork er is measured by Meguire view include Barth, Iden and Russek (1986), Feldstein and Elmendorf (1990), and Modigliani and Sterling (1986) and (1990). 5 the capital stock at the end o f the previous year divided by the average level o f business sector employment during the year.5 The grow th o f output per w ork er slowed from a 2.5 percent annual rate from 1948 to 1973 to nearly zero (-0 .1 percent rate) from 1973 to 1982, before rebounding to a 1.8 per cent rate from 1982 to 1989. Advocates o f the public capital hypothesis emphasize only the post-1973 slowing. The grow th o f the public capital stock per w ork er also slowed abruptly in the early 1970s. Public capital per w ork er rose at a 3 percent rate from 1948 to 1971, then showed no grow th from 1971 to 1982, much like the 1973-82 slowing in productivity. Since 1982, how ever, the grow th in the stock o f public capital per w ork er slowed further, falling at a 1.6 percent rate. This inconsistent shift in trends after 1982 does not contradict the fact that both variables g rew m ore slowly after the early 1970s than they did before then. The simple correlation coefficient fo r the logarithms o f the tw o measures shown in the figure is 0.95 fo r the period 1947 to 1989. This strong positive relationship is a classic case o f a spurious time-series relationship. In fact, changes in the public capital stock per w ork er are not statistically significantly related to changes in business sector output per w orker. The correla tion coefficient fo r changes in the logarithm o f each measure is negative and equals only -0.03 fo r the period 1948 to 1989. The reversal in the size, significance and sign o f the correlations among levels and first-differences illustrates the importance o f the issues explored below in assessing the public capital hypothesis. THE BUSINESS SECTOR PRODUC TION FUNCTION AND PUBLIC CAPITAL FORMATION The most direct aggregate evidence on the positive effect o f public sector capital formation 5The net public and private capital stock data used in this article were provided by John Musgrave from data he prepares for the U.S. Department of Commerce, which is described in U.S. Department of Commerce (1987) and Musgrave (1988). The series are constructed by deducting depreciation from gross stock measures, which cumulate gross investment less discards (assets that are scrapped). The depreciation methods use straight-line depreciation for service lines equal to 85 percent of the U.S. Treasury Department’s Bulletin F service lines. Constant cost measures (1982 prices) are used throughout this article. can be obtained from production function estimates. The aggregate production function in dicates the maximum output that can be p ro duced w ith labor and capital given technology and other factors influencing production. The marginal product o f each resource is assumed to be positive and inversely related to the quan tity o f the resource (diminishing returns). The Conventional Approach Ratner (1983) provides the first model that ex plicitly adds public capital to the production function to test w hether the marginal product o f public capital is positive. M ore recent analyses by Aschauer (1989a) and Munnell (1990) use a similar approach. Ratner assumes that the business sector production function (Q) can be represented by a Cobb-Douglas function: (1) Q, = A h “kfkgf e (r,+l,), w h ere A = a scale parameter, h t = business sector hours, kt = the flo w o f services from K „ the constant-dollar net nonresidential stock o f private capital at the end o f the previous year, kg, = the flow o f services from KGt, the public capital stock at the end o f the previous year, r = a rate o f disembodied technical change, t = a time trend, and £t = a norm ally and independently distributed random disturbance term. The same utilization rate, c„ capital as fo r private capital, private capital services, kt, is o f services from governm ent ctKG,.G The utilization rate is is used fo r public so the flo w of ctKt and the flo w capital, kg„ equals measured by the 6The questionable assumption of an identical utilization rate for public and private capital is not required, however. The derivation of equation 2 requires only that the use of public capital be proportional to that of private capital. In this case, the constant term A in equation 2 will include this factor of proportionality. MAY/JUNE 1991 6 Federal Reserve Board's index o f manufacturing capacity utilization. Ratner also assumes that the production func tion is characterized by constant returns to scale, w hich means that a proportional rise in each resource raises Q by the same proportion; this assumption is expressed as a + ft + 6 = 1. Thus, the production function can be rew ritten as: (2) ln(Q,/kt) = InA + aln(h,/kt) + dln(KGt/K,) + rt + £,. The public capital hypothesis that KG affects Q is tested by determining w hether d, the output elasticity o f public capital, is positive.7 The out put elasticity is the marginal product o f public capital services divided by the average product o f these services (Q/kg). The coefficient a is the output elasticity o f labor which, in principle, should equal the share o f labor cost in total cost. Ratner estimated equation 2 using data fo r 1949 to 1973. Since then, the data have been revised numerous times, including changing the base period fo r computing constant-dollar out put and capital stock data. W hen this equation is estimated fo r the original sample period 1949-73, using the latest available data, the estimate is (t-statistics in parentheses): 7There are two alternative transformations of the variables that could be used to derive the theoretical specification. Equation 2 arises from the substitution (p = 1 - a - d), but a or d could have been eliminated instead. The alter native expressions are analytically and statistically equivalent, however; in particular, the estimates are nor malized on output in all three specifications. 8The output elasticity of a resource equals the ratio of the resource’s marginal product to its average product. Ratner’s original estimate implied that the ratio of the marginal product of private capital to that of public capital was 3.8 times the ratio of the public to the private capital stock. During the 1949-73 period, the latter averaged 55.4 percent, so that the private gross rate of return (rental price of capital and marginal product) was about 2.1 times the respective measure of the public sector, according to his earlier estimate. The updated estimate of the relative marginal product of private capital in equation 3 is 35 per cent of the marginal product of public capital. 9When equation 3 is estimated with a first-order autocor relation correction term, its coefficient is not statistically significant. 10Munnell (1990) uses a capital input series prepared by the U.S. Bureau of Labor Statistics (BLS) to measure the ser vices of capital rather than the services yielded by the constant dollar net nonresidential private capital stock. The BLS series is described by Oliner (1989). This series is relatively new and is intended to measure the flow of ser vices as measured here, but does not appear to be much different from the capital stock measure used here. The FEDERAL RESERVE BANK OF ST. LOUIS (3) ln(Q,/kt) = 1.410 + 0.548 ln(ht/kt) (11.52) (9.32) + 0.277 R2 = 0.93 ln(KGt/Kt) + 0.0128 t (2.81) (6.24) S.E. = 1.02% D.W. = 1.65. The statistically significant output elasticity of public capital is estimated to be 27.7 percent. This is much larger than Ratner’s earlier estimate o f 5.8 percent.8 The rise in the output elasticity o f the public capital stock arises from data revisions subsequent to Ratner's study.9 Aschauer (1989a) and Munnell estimate similar production functions fo r the business sector over the period 1949-85 and the non-farm business sector over the period 1949-87, respec tively.1 T h ey both find a positive and significant 0 output elasticity fo r public capital; their estimates, how ever, are about 30 to 40 percent, somewhat larger than that in equation 3.1 1 Three Potential Shortcomings o f Existing Estimates Estimates like that in equation 3 are suspect fo r three reasons. First, they ignore the signifi cant influence o f the relative price o f energy on capital stock measure and its utilization rate used in this article are also used by Ratner and Aschauer. Both Mun nell and Aschauer (1989a) include the capacity utilization rate in manufacturing as a separate variable to capture the influence of the business cycle on productivity. They pro vide no theoretical justification, nor do they indicate whether the capacity utilization rate is intended to capture any influences besides the varying use of the stock of business sector capital. 11Schultze (1990) has criticized such estimates for implying implausibly large estimates of the rate of return to in frastructure. Aaron (1991) also questions the magnitude of the effect, the conceptual basis for such an effect and whether the estimate is spurious. A counterpart to the relatively high rate of return, at least in equation 3, is that the output elasticity of hours is only 54.8 percent. This is well below the theoretically expected value, which equals the share of labor cost in total cost of about 66.7 percent. The output elasticity of private capital is about 17.5 per cent, about half of the expected value. 7 productivity found in similar studies (see the shaded insert on page 10). Second, they omit a significant time trend or reductions in the trend found in other studies. Third, they contain variables that are not stationary, raising the possibility o f spurious estimates.1 2 Consider the Cobb-Douglas production func tion including the flo w o f energy, Et: (4) Q, = A hrkfkgf E\ e (r,+t,), w h ere y is the output elasticity o f energy. The quantity o f energy is assumed to satisfy the first-order condition fo r its employment, Et = yQt/pt, w h ere p ' is the price o f energy measured e relative to the price o f business sector output.1 3 In addition, the production function is assumed to be characterized b y constant returns to scale, (a + / + y + d = l ) . 1 Substituting these tw o assump ? 4 tions into equation 4 and taking logarithms o f both sides yields: 12Eberts (1990) raises the issue of whether there is “ reverse causation” in estimates of the effect of public capital on private output; in other words, does a significant positive correlation indicate that public capital raises private output or does a rise in private output raise the demand for and quantity of public capital? He provides regional evidence suggesting that causality runs both ways. 13The quantity of energy is assumed to be proportional to stock of energy-using capital and to its services. It is in cluded because conventional measures of the flow of capital services, k (or kg), are not expected to reflect the differential effect of energy price changes on the economic value of the capital stock and its flow of services. Reduced energy usage is only one of several reasons why higher energy prices affect private sector output. The domestic and foreign capital stocks (for example, the pools of oil and gas, beds of coal and hydroelectric power sources) that produce the energy used in U.S. production are not included in the measured domestic nonresidential capital stock. Therefore, if the relative price of energy rises and producers respond by using less of this capital, the reduc ed flow of services from this capital will not be reflected in kt. Moreover, the decline in the real value of the rest of the capital stock due to higher operating costs also is not reflected in kt, since replacement cost rather than market prices are used to measure the value of existing assets in computing the constant dollar net stock. See Rasche and Tatom (1977a), (1977b) and (1981). Also see Helliwell, Sturm, Jarrett and Salou (1986) for an alternative approach. (5) ln(Qt/kt) = InA* + a*ln(h,/kt) + d*ln(KGt/Kt) + y*Inp' + r*t + a' d* = a / (l- y ) = 6/(1- Y), y* = —y/(l —y). * r = r/(l -y ), A* = A (1,|-y)y (T,1-r)) and £ ,* = £t/(l —y). The omission o f en ergy price effects on produc tivity after 1973 could result in attributing energy-related productivity losses to the decline in the grow th o f public capital. The second potential shortcoming o f existing tests using production functions is that they omit significant time trends or significant breaks in the time trend found in similar studies.1 5 Trends are intended to control fo r the influence o f the pace o f technical change; their omission could bias the coefficients and the standard er rors fo r the included variables, especially those correlated w ith the omitted time trends.1 6 tion in return for estimates of the coefficients that are more efficiently estimated and more readily interpreted as estimates of the theoretical parameters. The importance of this issue and the inability to reject this constraint is discussed in Tatom (1980). 15For example, Munnell (1990) includes no time trends and Aschauer (1989a) includes no shift in the trend. 16A decline in the trend rate of technical change in 1967 is discussed in Rasche and Tatom (1977b) and several studies that discuss this trend-break are cited there. A break in the trend rate (r) in 1967 for the business sector was not significant in the data available at the time of Ratner’s study, but it is significant in later data. See Tatom (1988), for example, for a discussion of this change in significance. Darby (1984) argues that a declining quadratic trend arises from a post-depression and postWorid War II catch-up in the level of technology. 14The use of the constant returns to scale constraint in estimating production functions is quite common because of the intuitive appeal of this property and, more important ly, because the high correlations between hours, the flow of private capital services, the time trend and, in this case, the flow of public capital services, are expected to make it difficult to interpret the coefficient estimates and to raise their standard error estimates without this constraint. When the constraint can be rejected, its imposition trades off some explanatory power in fitting the production func MAY/JUNE 1991 8 The effect o f the first tw o shortcomings on an estimate o f the production function like equa tion 3 can be seen by including the relative price o f energy, as in equation 5, and by allow ing fo r a quadratic trend component t2. Nearly identical results arise from allowing fo r a one time decline in the linear time trend from 1.5 percent per yea r b efore 1967 to a 1.0 percent rate afterw ard.1 For the period 1948 to 1989, 7 estimate is: (6) ln(Q,/kt) = 1.595 + 0.614 ln(ht/kt) (15.29) (12.88) + 0.132 ln(KGt/Kt) - 0.048 lnp't (2.77) (-6 .4 1 ) + 0.019 t - 0.0001 t2 (8.42) (-4.23 ) R2 = 0.97 S.E. = 0.95% D.W. = 1.49 This estimate indicates a statistically significant, positive effect o f the public capital stock on out put, but is less than half that given in equa tion 3 or the estimates obtained by Aschauer and Munnell. Both the relative price o f energy and the slowing in the time trend are statistical ly significant. Updating the equation 3 estimate, but including the energy price and time trend slowing, does not alter the statistical significance o f the public capital stock effect, h ow ever.1 8 The third potential shortcoming in regression estimates like equations 3 or 6 is that they con 17The standard error of estimate using the time trend shift in 1967 instead of the quadratic trend is 0.99 percent; the estimated output elasticity of the public capital stock is 0.135 in this case. Equation 3 and the Aschauer and Mun nell estimates are representative of estimates without a declining time trend and energy price effects. For exam ple, when the quadratic trend term and energy price term are omitted from equation 6, the output elasticity of the public capital stock is 0.306 (t = 5.27), about the same as in equation 3; when a significant first-order autocorrelation term is added, this output elasticity rises to 0.343 (t = 4.91). 18Equation 6 contains the 1948 data point, as well, to in clude all available data. This does not affect the results, however. When the relative price of energy is added to the 1949-73 estimate in equation 3, its coefficient (-0 .1 1 7 ) is not statistically significant (t = 1.35). Its inclusion lowers the coefficient on ln(KG/K) to 0.206, and it too becomes statistically insignificant (t = 1.88) at a 95 percent con fidence level using a one-tail test. When a first-order autocorrelation correction term is added to equation 6, its coefficient is not statistically significant. 19This bias is explained by Granger and Newbold (1974) and (1986). Some analysts refer to this potential bias as arising only when two random walk variables are used in a regression, because this was the example used by Granger and Newbold (1974). Granger and Newbold (1986) FEDERAL RESERVE BANK OF ST. LOUIS tain variables that are not stationary, and so are subject to a spurious regression bias.1 First9 differencing typically renders the data stationary and rem oves the problem o f justifying or ex plaining the existence o f a deterministic trend or trends. The evidence concerning this poten tial difficulty and its implications is explained below. Are the Production Function Variables Stationary? Table 1 reports Dickey-Fuller tests fo r a unit root fo r the levels o f the variables in equation 6— ln(Q/k), ln(h/k), ln(KG/K) and lnp'—and fo r their first-differences. The relevant statistic fo r the unit root test is the t-statistic fo r the co effi cient on the lagged level o f the variable (Zt_,) whose first-difference is used as the dependent variable; this coefficient is labeled b in the table. If this coefficient is significantly different from zero w hen the tim e trend is statistically insigni ficant and, therefore, omitted, then the variable Z is stationary. W hen the time trend is signifi cantly different from zero, its coefficient, d, is included in the reported test equation. In this case, if b is significant, the variable Z is said to be trend-stationary.2 Only one lagged depen 0 dent variable is statistically significant in any o f the tests fo r the levels o f the data in the table; these significant instances are reported in the use other nonstationary variable combinations. Engel and Granger (1987) explain that a linear combination of sta tionary and nonstationary variables is nonstationary, unless the nonstationary variables are cointegrated. Thus, the er ror term in such an equation is potentially nonstationary, giving rise to a potentially spurious regression. 20The t-statistics for the b coefficients are estimates of Dickey-Fuller statistics called (t „), when the time trend is omitted (d = 0), and r T when the time trend is included. , The critical values of t„ and tt for this size sample are given in Fuller (1976) and equal about - 2 .9 5 and -3 .5 3 , respectively. 9 Table 1 Tests for Nonstationarity Test Equation fo r Levels o f Variable (Z): AZ, = a + bZ M + d t + eAZ, Levels o f Variable (Z): a (Sample period: 1950-89)1 S.E. D.W. 0.29 0.023 2.00 0.12 0.034 2.31 0.10 0.034 2.44 0.22 0.031 1.75 0.420 (2.81)* 0.15 0.088 2.03 b d e - 0.647 (-4 .2 9 )* -0 .0 0 2 (-3 .5 9 )* - ... ln(Q,/k,) 0.101 (3.86) ln(h,/kt) -0 .3 2 4 (-2 .0 9 ) -0 .1 4 0 (-1 .8 1 ) - 0.003 (-1 .3 3 ) ln(h,/kt) -0 .1 2 6 (-2 .9 1 ) - 0.040 (-2 .3 3 ) — ln(h,/k,) -0 .1 8 5 (-4 .1 5 ) -0 .0 5 9 (-3 .5 0 )* -0.25 5 R* ( -1 .7 9 ) ... ln(p;) 0.252 (1.42) -0 .0 6 3 (-1 .4 0 ) ln(p;) 0.401 (1.92) -0.11 1 (-1 .9 4 ) 0.002 (1.34) 0.436 (2.94)* 0.17 0.087 2.06 - 0.005 (-0 .4 6 ) -0 .0 1 9 (-1 .0 3 ) -0 .0 0 0 5 (-3 .5 2 )* 0.602 (6.77)* 0.68 0.007 1.91 d R; S.E. D.W. — 0.68 0.025 2.03 0.68 0.025 2.07 0.57 0.036 1.57 0.64 0.032 1.72 0.29 0.089 1.98 0.27 0.090 1.98 0.23 0.008 1.53 0.41 0.007 1.84 ln(KG,/K,) Test Equation fo r D ifferences (AZ): A2 = a + bAZ, , + d t + £, Z, First-difference (AZ,) a (Sample period: 1950-89)2 b Aln(Q,/kt) -0 .0 0 6 (-1 .5 1 ) -1 .3 5 2 (-9 .0 9 )* Aln(Q,/kt) -0 .0 1 3 (-1 .6 4 ) -1 .3 6 4 (-9 .1 5 )* Aln(h,/k.) -0.03 1 (-4 .3 8 ) -1.14 1 (-7 .2 1 )* Aln(h,/k,) -0 .0 6 2 (-4 .9 9 ) -1 .2 6 6 (-8 .4 1 )* Aln(p;) 0.004 (0.28) -0 .6 1 2 (-4 .0 9 )* Aln(p;) 0.0004 (0.014) -0 .6 1 3 (-4 .0 4 )* Aln(KG,/K,) -0.001 (-0 .9 9 ) -0 .3 2 3 (-3 .5 3 )* Aln(KG,/K,) 0.006 (2.35) -0 .4 2 5 (-5 .0 0 )* 0.0004 (1.03) ... 0.001 (2.94)* — 0.0002 (0.14) ... -0 .0 0 0 4 (-3 .5 5 )* * Significant at a 5 percent level. 1 When e is not significantly different from zero at a 5 percent level of significance, it is constrained to zero and the period used begins one year earlier. 2 The lagged dependent variable is not significant at a 5 percent significance level for any of these variables, so it is omitted. MAY/JUNE 1991 10 Alternative Hypotheses About the Productivity Decline Several hypotheses have been put forth to explain the slow dow n in productivity grow th besides the one focusing on the slow dow n in public capital form ation originally form ulated by Eisner (1980), Schultze (1981) and Ratner (1983). These other explanations include a slowing in the trend o f resources m oving from agriculture to the industrial sector; the return to “ norm ality” from the tem porarily rapid postwar grow th o f output and produc tivity associated w ith reversing the adverse effects o f the Depression and W orld W a r II on the private sector; a slow dow n in research and developm ent spending; the costs o f increased governm ent regulation; a slow dow n in the grow th o f the private capital stock per w orker; and the rise in energy costs in 1973-74 and 1979-81.1 Rasche and Tatom (1977a) and (1981) ex plain h ow a rise in energy prices reduces the economic capacity o f the typical firm and renders capital obsolete. In the short run, firms alter their optimal production tech 1Kendrick (1979) contains papers dealing with most of these hypotheses. Baily (1986) presents a more recent summary of these hypotheses. Also, see Dennison (1974), (1979) and (1985). Darby (1984) develops the hypothesis about the postwar and post-Depression tem porary surge in growth. Griliches (1988) examines the R&D hypothesis. Crandall (1980) and Gray (1980) ex amine the regulatory hypothesis. 2Jorgenson (1988) provides evidence of these ad justments within individual industries, while Baily (1981) and Griliches (1988) have focused on energy price- table. No lagged value o f the dependent variable is significant (or reported) in the bottom half o f the table w h ere the presence o f a unit root fo r the first-differences is tested. The evidence fo r the levels o f the variables shown at the top o f the table indicates that ln(Q/k) is trend-stationary. The level o f ln(h/k) appears to be stationary when the insignificant lagged dependent variable is included to reduce the extent o f autocorrelation indicated b y the relatively high Durbin-Watson statistic (D.W.). W ithout this lagged dependent variable, the hypothesis that ln(h/k) has a unit root, or FEDERAL RESERVE BANK OF ST. LOUIS niques, reducing their use o f energy and, in some cases, obsolete capital, substituting labor and other capital to econom ize on higher energy costs. In the long run, reduc tions in the productivity o f labor and capital resources lead, in the case o f capital, to a smaller desired capital stock and flo w o f its services.2 Except fo r an unexplained shift or a slow ing in the time trend, only the energy price rise and associated slowing in the grow th o f the capital-labor ratio provides explanations that are consistent w ith the timing and magnitude o f productivity movements since 1973. Tatom (1982), fo r example, provides evidence that the entire decline in productivi ty grow th from late 1973 to 1981 resulted from the rise in energy prices and the associated reductions in the capital-labor ratio.3 Tests o f the effects o f public capital on private output have not controlled fo r these effects. induced obsolescence of capital and its effects on pro ductivity. Also, see Tatom (1979a) and (1982). 3The energy-price hypothesis is not universally ac cepted. For example, Berndt (1980), Dennison (1974), (1979) and (1985), Darby (1984) and Olson (1988) have been critical of its significance, arguing that the share of energy in costs is too small or, in Darby’s case, that a quadratic trend fits the data without any energy price effect, so that the slowing in the 1970s was not a puz zle. With Darby’s view, the productivity pick-up in the 1980s becomes a puzzle, however. is not stationary, cannot be rejected. For this reason, ln(h/k) is considered to be nonstationary here. According to the tests, lnp' and ln(KG/K) are also nonstationary. The latter has a signifi cant trend term (d), but ln(KG/K) is not sta tionary w hen it is included. Based on the level results in the table, nonstationarity cannot be rejected fo r the fou r variables entering equation 6. The bottom panel o f the table conducts the same test fo r unit roots fo r first-differences o f the variables. Th e test fo r each o f the variables 1 1 rejects a unit root, but tw o variables, Aln(h/k) and Aln(KG/K), are trend-stationary. The levels o f ln(Q/k) and lnp' are integrated o f order 1, 1(1), which means that these variables must be differenced once to achieve stationarity. The levels o f ln(h/k) and ln(KG/K) are 1(2), because they must be differenced tw ice to achieve sta tionarity.2 The presence o f a significant trend 1 in the first-differences suggests that there is a significant quadratic trend in the levels o f the data. A first-difference version o f equation 5 in volves only stationary and trend-stationary vari ables. The first-difference o f the time trend term, rt, in equation 5 is the constant, r, which is the constant term in the first-difference equa tion. If the time trend consists o f broken linear segments, then the average o f the coefficients on these linear trends also is captured in the constant term. I f there is a deterministic quad ratic trend, first-differencing results in a linear trend remaining in the first-difference expres sion. Since tw o o f the variables in equation 5 are only trend-stationary, how ever, a time trend must be included in the first-difference regres sion to maintain the desired stationarity. This is consistent with the presence o f a deterministic quadratic trend in the production function, so that differencing does not avoid the considera tion o f deterministic trends in this case. Estimating a first-difference equation avoids both the problems arising from nonstationarity and the difficulties o f selecting ad hoc breaks in the time trend in equation 5.2 2 21The evidence that, in one test, ln(h/k) is stationary in table 1, while its first-difference is trend-stationary may appear to be inconsistent. In the level estimate for ln(h/k), the trend term is statistically significant when a statistically significant break in 1967 is included or when the quadratic term is included. In each instance, however, the hypothesis that ln(h/k) has a unit root still is not rejected. When the ln(h/k) equation containing the time trend and its break in 1967 is first-differenced, Aln(h/k) is stationary; that is, the addition of a time trend is not statistically significant and a unit root is rejected. In this case, Aln(h/k) is sta tionary, not trend-stationary. When the ln(h/k) equation containing the significant quadratic trend term is firstdifferenced, the result is that given in the table indicating trend-stationarity. Whether the appropriate inference is that Aln(h/k) is trend-stationary or stationary is not essential for the analysis below, however, because another variable, ln(KG/K), also has a trend-stationary first-difference. NEW ESTIMATES OF THE EFFECT OF PUBLIC CAPITAL ON PRIVATE SECTOR OUTPUT The variables used in equation 6 do not ap pear to be stationary, so the statistical signifi cance o f the public capital effect found there is potentially spurious. This problem is avoided by estimating the production function parameters in a first-difference specification with a time trend.2 3 The first-differenced (A) estimate o f the pro duction function fo r the period 1949 to 1989 is: (7) Aln(Q,/kt) = 0.025 + 0.042 Aln(KGt/K,) (6.30) (0.33) + 0.737 Aln(ht/kt) (14.34) -0.058 Alnp' - 0.0005 t (-3 .2 3 ) (-3 .0 5 ) R2 = 0.85 S.E. = 1.05% D.W. = 2.25 The coefficient on public capital, w hile positive, is much smaller than in estimates based on the levels o f the variables, like that in equation 3 or even in equation 6.2 M ore importantly, however, 4 tion by detrending these measures. This adjustment only affects the t-statistic for the trend, which was -3 .0 5 without the correction. A slight rise in the trend coefficient does not show up in the rounded value of the coefficient. 24When equation 3 for the earlier period (1949-73) is firstdifferenced and t is added, the government capital stock coefficient reverses sign ( -.0 0 4 ) and is not statistically significant (t= -0 .0 3 ). When the first-difference of the logarithm of the relative price of energy is included, its coefficient (-0 .0 2 3 ) is not significant (t= -0 .2 8 ) and the public capital stock result is unaffected. In both cases, the time trend is not statistically significant ( - 1 .4 4 and -1 .2 2 , respectively). 22The first-difference of the variable ln(KG/K) is not trendstationary for the sample period used in equation 3, 1949-73, or in 1949-89. In particular, for data from 1949 to 1989, the coefficient on its lagged growth rate in a regres sion of its second difference, including a significant trend, is -3 .4 2 , which is smaller in absolute value than the critical value of -3 .5 0 (5 percent significance). 23The standard errors and t-statistics have been corrected for the time trend in Aln(KG/K) and Aln(h/k) in the estima- MAY/JUNE 1991 12 this coefficient is not statistically significant.2 5 The coefficient on energy prices is significantly negative, and that on hours per unit o f capital remains significantly positive and rises to a value that is closer to its theoretically expected level.2 Thus, equation 7 indicates that the 6 public capital stock has no significant influence on business sector output, given the capitallabor ratio and the relative price o f energy.2 7 The statistical insignificance o f the public capital stock arises from first-differencing the data; it does not arise from the inclusion o f energy prices in the estimation o f equation 7 or from allowing fo r a trend.2 The omission o f the 8 significant energy price term does not produce a significant public capital stock effect either. W hen it is omitted in equation 7, the coefficient on the public capital stock variable, Aln(KG/K), rises to 0.108; how ever, it remains statistically insignificant (t = 0.77).2 9 The inferences from equation 7 are not sub ject to the spurious regression problem. Since tests o f the variables in equation 6 generally fail to reject nonstationarity, the results in equation 7 o ffe r the strongest evidence on the factors in fluencing, or not influencing, business sector output. This estimate rejects the public capital hypothesis. 25lf the insignificance of ln(KG/K) arose from over differencing an appropriate test equation, then the problem could be corrected by estimating equation 7 with a signifi cant MA1 error process; the MA1 coefficient should be equal to minus one in this case. When equation 7 is estimated with an MA1 error process, the MA1 parameter is only -0 .3 6 9 ; it is not statistically significant (t = -1 .9 5 ). More importantly, it is significantly less than one (t = 3.34). The coefficient on ln(KG/K) is reduced (0.021) and it remains statistically insignificant (t = 0.22) when this term is included. 26This theoretical value is derived in equation 5; it is condi tional on the share of labor in total cost and the coefficient on the relative price of energy. For values of these parameters of 0.667 and -0 .0 5 8 , respectively, this theoretical value is 0.706. 27Rubin (1990) regresses the growth rate of multifactor pro ductivity in manufacturing and 11 two-digit SIC code in dustries on a constant, the growth rate of the Federal Reserve Board’s measure of the industry’s capacity utiliza tion rate, and the growth of core infrastructure. Core in frastructure includes highways, streets, sewers and water systems in her analysis. The period she uses is generally from 1956 to 1986. She finds that there is no statistically significant effect for core infrastructure in any industry ex cept petroleum refining, where a significant positive rela tionship is observed. Her result is consistent with the view suggested above, that the decline in public capital growth is, in part, a proxy for the pattern of increased energy prices. In papers prepared after this research was completed, Hulten and Schwab (1991) and Jorgenson (1991) note the FEDERAL RESERVE BANK OF ST. LOUIS An Alternative Approach: Are Private Sector Output and Public Capital Cointegrated? According to the evidence in table 1, the variables in equations 6 are not stationary; tw o o f them are 1(1) and tw o are 1(2). Engel and Granger (1987), Johansen (1988) and Johansen and Juselius (1989) develop procedures fo r ex amining w hether 1(1) variables have long-run relationships or are cointegrated. These methods cannot be used here because tw o o f the variables in equations 6 are 1(2). Neither pro cedure addresses the problem o f h ow to incor porate a linear time trend, trend shift or quadratic trend in a cointegration test. Stock and W atson (1989) have developed a m ethod fo r testing cointegration among higherorder integrated variables, including variables that are integrated o f different orders. They ex plain that one approach to the problem o f non stationarity is to include significant lags and leads o f first-differences o f the dependent and independent variables as right-hand-side vari ables in tests o f functional relationships. They argue that this practice avoids the spurious regression problem fo r nonstationary variables pointed out by Granger and N ew bold (1974) and that it indicates the presence o f long-run (or fragility of estimates of the marginal product of public capital. Hulten and Schwab provide evidence of this fragili ty, but in their first-difference estimates, the private sector input coefficients are fragile as well. 28Without the trend term in equation 7, the coefficient on the public capital stock variable is 0.147, about the same as in equation 6, but it is not statistically significant (t = 1.07). The trend term is necessary to ensure that the error term in equation 7 is stationary. A regression of the firstdifference of the residuals from equation 7 on the lagged level of the residual, with no constant, yields a coefficient on the lagged residual equal to -1 .1 5 2 (t= -7 .3 2 ). Even without the trend term in equation 7, the t-statistic on the resulting lagged residual is -5 .8 6 . Engel and Granger (1987) indicate that the critical value for these t-statistics is -3 .3 7 . Thus, the residuals are stationary in either case. 29Production function estimates are subject to simultaneous equation bias, but this has no effect here. Virtually the same results are obtained using a two-stage least-squares estimation procedure. The instruments for the right-handside variables include the first-difference of the logarithms of real wages, the AAA bond yield, and the relative price of private capital goods, as well as lagged dependent and independent variables. 13 cointegrating) relationships betw een variables as the coefficients on the levels o f the variables. This approach was taken in estimating the level o f the production function in equation 5.3 0 Up to tw o leads and lags o f first-differences of each variable in equation 5 w e re examined. The equation estimate containing only significant leads or lags, estimated over the period 1950-88 is: (8) ln(Q,/kt) = 0.489 + 0.105 ln(h,/k,) (10.76) (9.86) - 0.046 lnp' - 0.075 ln(KGt/Kt) (-2 .9 5 ) (-1 .4 7 ) + 0.762 Aln(KGI+I/Kt+1) (2.85) + 0.064 Alnp'_, - 0.065 Alnp'+ 1 (2.11) (-2 .4 0 ) R2 = 0.93 S.E. = 1.36% D.W. = 1.69 In this estimate, the coefficient on the nonmilitary net stock o f public capital per unit o f private capital (-0 .0 7 5 ) has the w ro n g sign and is not statistically significant.3 Like the result 1 reported above, the nonmilitary public capital stock has no statistically significant relationship w ith business sector output. The levels o f busi ness sector output or productivity are uncor related w ith the level o f the nonmilitary public capital stock. The statistically significant tstatistics on the coefficients fo r the levels of ln(h/k) and lnp' in equation 8 suggest that only the variables (InQ/k, lnh/k, lnp') are cointegrated. CONCLUSION An increasing num ber o f people are advo cating increased governm ent capital spending to raise private sector output, productivity and private capital formation. The evidence p re sented here, based on the post-World W a r II ex perience, suggests that a rise in public capital spending w ould have no statistically significant effect on these measures. Earlier claims o f a positive and significant e f fect o f public capital on private sector output have arisen from spurious estimates. In fact, most o f these earlier estimates have ignored a 30The quadratic trend is omitted because it is not an in tegrated stochastic process; thus, production cannot be cointegrated with it. See Stock and Watson (1988), p. 168, for example. trend or broken trends in productivity, as w ell as the statistically significant influence o f energy price changes. Simply accounting fo r these tw o factors reduces the conventional estimates o f the elasticity o f private output with respect to public capital o f about 30 to 40 percent, to about 13 percent. M ore importantly, how ever, both the earlier estimates and those reported here that find a statistically significant public capital effect use equation estimates that con tain nonstationary variables. Thus, these estimates are likely to be spurious. W hen all o f these problems are addressed us ing a first-difference estimate o f the production function, the public capital stock effect on pri vate sector output is not statistically different from zero. Appropriately estimated, the hypo thesis that public capital has a positive marginal private sector product cannot be supported. The same result is found using a m ethod that allows testing a long-run relationship among nonstationary variables. REFERENCES Aaron, Henry J. “ Comment on Historical Perspectives on In frastructure Investment: How Did We Get Where We Are?” presented at the American Enterprise Institute for Public Policy Research Conference on “ Infrastructure Needs and Policy Options for the 1990s,” Washington, D.C., February 4, 1991. Arrow, Kenneth J., and Mordecai Kurz. Public Investment, the Rate of Return, and Optimal Fiscal Policy (Johns Hopkins Press, 1970). Aschauer, David Alan. “ Infrastructure: Spending Trends and Economic Consequences,” presented at the Western Economic Association meetings, San Diego, June 30, 1990. _______ . “ Does Public Capital Crowd Out Private Capital?” Journal of Monetary Economics (September 1989b), pp. 171-88. ________ “ Is Public Expenditure Productive?” Journal of Monetary Economics (March 1989a), pp. 177-200. ________ “ Fiscal Policy and Aggregate Demand,” American Economic Review (March 1985), pp. 117-27. Baily, Martin N. “ What Has Happened to Productivity Growth?” Science, October 24, 1986, pp. 443-51. _______ . “ Productivity and the Services of Capital and Labor,” Brookings Papers on Economic Activity (1:1981), pp. 1-50. Barth, James R., George Iden, and Frank S. Russek. “ Government Debt, Government Spending, and Private Sector Behavior: Comment,” American Economic Review (December 1986), pp. 1158-167. Berndt, Ernst. “ Energy Price Increases and the Productivity Slowdown in United States Manufacturing,” in The Decline in Productivity Growth, Federal Reserve Bank of Boston Conference Series, no. 22 (1980), pp. 60-89. 31A first-order autocorrelation term is not statistically signifi cant and its inclusion does not alter the statistical in significance of ln(KG/K). MAY/JUNE 1991 14 Crandall, Robert. “ Regulation and Productivity Growth,” in The Decline in Productivity Growth, Federal Reserve Bank of Boston Conference Series, no. 22 (1980), pp. 93-111. Darby, Michael R. “ The U.S. Productivity Slowdown: A Case of Statistical Myopia,” American Economic Review (June 1984), pp. 301-22. Denison, Edward F. Trends in American Economic Growth, 1929-1982 (The Brookings Institution, 1985). ________ “ Explanations of Declining Productivity Growth,” Survey of Current Business (August 1979, Part II), pp. 1-24. ________ Accounting for Slower Economic Growth: The United States in the 1970s (The Brookings Institution, 1979). Deno, Kevin T. “ The Effect of Public Capital on U.S. Manufacturing Activity: 1970 to 1978,” Southern Economic Journal (October 1988), pp. 400-11. Eberts, Randall W. “ Public Infrastructure and Regional Economic Development,” Federal Reserve Bank of Cleveland Economic Review (Quarter 1, 1990), pp. 15-27. Kendrick, John W. “ Productivity Trends and the Recent Slowdown: Historical Perspective, Causal Factors, and Policy Options,” in William Fellner, ed., Contemporary Economic Problems 1979 (American Enterprise Institute for Public Policy Research, 1979), pp. 17-69. Kormendi, Roger C., “ Government Debt, Government Spend ing, and Private Sector Behavior,” American Economic Review (December 1983), pp. 994-1010. Kormendi, Roger C., and Philip Meguire. “ Government Debt, Government Spending, and Private Sector Behavior: Reply and Update,” American Economic Review (June 1990), pp. 604-17. _______ . “ Government Debt, Government Spending, and Private Sector Behavior: Reply,” American Economic Review (December 1986), pp. 1180-187. Malabre, Alfred L. “ Economic Roadblock: Infrastructure Neglect,” Wall Street Journal, July 30, 1990. Eisner, Robert. "Total Income, Total Investment, and Growth,” American Economic Review (May 1980), pp. 225-31. Modigliani, Franco, and Arlie G. Sterling. “ Government Debt, Government Spending, and Private Sector Behavior: A Fur ther Comment,” American Economic Review (June 1990), pp. 600-03. Engle, Robert F., and Clive W. J. Granger. “ Co-integration and Error Correction: Representation, Estimation and Testing,” Econometrica (March 1987), pp. 251-76. ________ “ Government Debt, Government Spending, and Private Sector Behavior: Comment,” American Economic Review (December 1986), pp. 1168-179. Feldstein, Martin, and Douglas W. Elmendorf. “ Government Debt, Government Spending, and Private Sector Behavior Revisited: Comment,” American Economic Review (June 1990), pp. 589-99. Munnell, Alicia H. “ Why Has Productivity Growth Declined? Productivity and Public Investment,” New England Economic Review (January/February 1990), pp. 3-22. Friedman, Benjamin. Day of Reckoning (Random House, 1988). Musgrave, John C. “ Fixed Reproducible Tangible Wealth in the United States, 1984-87,” Survey of Current Business (August 1988), pp. 84-87. Fuller, Wayne A. Introduction to Statistical Time Series (John Wiley and Sons, 1976). Granger, Clive, and Paul Newbold. Forecasting Economic Time Series, 2nd ed. (Academic Press, 1986). _______ . “ Spurious Regressions in Econometrics,” Journal of Econometrics (July 1974), pp. 111-20. Gray, Wayne B. “ The Impact of OSHA and EPA Regulation on Productivity,” Working Paper no. 1405 (National Bureau of Economic Research, 1980). National Council on Public Works Improvement. Fragile Foundations: A Report on America’s Public Works, final report of the President and the Congress, February 1988. Ogura, Seiritsu, and Gary W. Yohe. “ The Complementarity of Public and Private Capital and the Optimal Rate of Return to Government Investment,” Quarterly Journal of Economics (November 1977), pp. 651-62. Griliches, Zvi. “ Productivity Puzzles and R&D: Another Nonexplanation,” Journal of Economic Perspectives (Fall 1988), pp. 9-21. Oliner, Stephen D. “ The Formation of Private Business Capital: Trends, Recent Developments and Measurement Issues,” Federal Reserve Bulletin (December 1989), pp. 771-83. Helliwell, John, Peter Sturm, Peter Jarrett, and Gerard Salou. “ The Supply Side in the OECD’s Macroeconomic Model,” OECD Economic Studies (Spring 1986), pp. 75-131. Olson, Mancur. “ The Productivity Slowdown, The Oil Shocks and The Real Cycle,” Journal of Economic Perspectives (Fall 1988), pp. 43-69. Hulten, Charles R., and Robert M. Schwab. “ Is There Too Little Public Capital? Infrastructure and Economic Growth,” paper presented at the American Enterprise Institute Con ference on “ Infrastructure Needs and Policy Options for the 1990s,” Washington, D.C., February 4, 1991. Rasche, Robert H., and John A. Tatom. “ Energy Price Shocks, Aggregate Supply and Monetary Policy: The Theory and International Evidence,” in Karl Brunner and Allan H. Meltzer, eds., Supply Shocks, Incentives and Na tional Wealth, Carnegie-Rochester Conference Series on Public Policy (North Holland, 1981), pp. 9-93. Johansen, Soren. “ Statistical Analysis of Cointegration Vec tors,” Journal of Economic Dynamics and Control (Vol. 12, 1988), pp. 231-54. Johansen, Soren, and Katarina Juselius. “ The Full Informa tion Maximum Likelihood Procedure for Inference on Cointegration— with Applications,” Institute of Mathematical Statistics, University of Copenhagen, January 1989. Jorgenson, Dale W. “ Fragile Statistical Foundations: The Macroeconomics of Public Infrastructure Investment,” com ment on Hulten and Schwab (1991), presented at the American Enterprise Institute Conference on “ Infrastructure Needs and Policy Options for the 1990s,” Washington, D.C., February 4, 1991. ________ “ Productivity and Postwar U.S. Economic Growth,” Journal of Economic Perspectives (Fall 1988), pp. 23-41. FEDERAL RESERVE BANK OF ST. LOUIS _______ . “ The Effects of the New Energy Price Regime on Economic Capacity, Production and Prices,” this Review (May 1977a), pp. 2-12. _______ . “ Energy Resources and Potential GNP,” this Review (June 1977b), pp.10-24. Ratner, Jonathan B. “ Government Capital and The Produc tion Function for U.S. Private Output,” Economic Letters (1983), pp. 213-17. Reich, Robert B. “ The REAL Economy,” The Atlantic Month ly (February 1991), pp. 35-52. Rubin, Laura. “ Productivity and the Public Capital Stock: Another Look,” paper presented at the Regional Analysis Committee meeting of the Federal Reserve System, Oc tober 4, 1990. 15 Sato, Ryuzo, and Tetsunori Koizumi. “ On the Elasticities of Substitution and Complementarity,” Oxford Economic Papers (March 1973), pp. 44-56. Schultze, Charles L. “ The Federal Budget and the Nation’s Economic Health,” in Henry J. Aaron, ed., Setting National Priorities: Policy for the Nineties (The Brookings Institution, 1990). ________ “ General Discussion,” Brookings Papers on Economic Activity (1:1981), p. 65. Stock, James H., and Mark W. Watson. “A Simple MLE of Cointegrating Vectors in Higher Order Integrated Systems,” Harvard and Northwestern University, respectively, November 1989. _______ . “Are The Macroeconomic Effects of Oil Price Changes Symmetric?” in Karl Brunner and Allan H. Meltzer, eds., Stabilization Policies and Labor Markets, Carnegie-Rochester Conference Series in Public Policy, (North Holland, 1988), pp. 325-68. _______ . “ Two Views of The Effects of Government Budget Deficits in the 1980s,” this Review (October 1985), pp. 5-16. _______ . “ Potential Output and the Recent Productivity Decline,” this Review (January 1982), pp. 3-16. ________ “ The Problem of Procyclical Real Wages and Pro ductivity,” Journal of Political Economy (April 1980), pp. 385-94. _______ . “ The Productivity Problem,” this Review (September 1979a), pp. 3-16. ________ “ Variable Trends in Economic Time Series,” Jour nal of Economic Perspectives (Summer 1988), pp. 147-74. _______ . “ Energy Prices and Capital Formation: 1972-77,” this Review (May 1979b), pp. 2-11. Tatom, John A. “ Should Government Spending on Capital Goods be Raised?” this Review (March/April 1991), pp. 3-15. U.S. Department of Commerce, Bureau of Economic Analysis. Fixed Reproducible Tangible Wealth in the United States, 1925-85, (GPO, June 1987). MAY/JUNE 1991 16 Ft. Alton Gilbert R. Alton Gilbert is an assistant vice president at the Federal Reserve Bank of St. Louis. Richard I. Jako provided research assistance. Supervision of Under capitalized Banks: Is There a Case for Change? T h e RECENT REPORT on deposit insorance reform b y the Department o f the Treasury (1991) calls fo r stronger enforcem ent o f bank capital requirements. Am ong the recom m ended changes is “ prompt corrective action” by supervisors in dealing w ith undercapitalized banks.1 Under the Treasury’s plan, banks would be divided into five groups, based on their capital ratios. Those with the highest capital ratios w ould be subject to the few est restrictions. As an incentive to maintain relatively high capital ratios, holding companies with banks in this first group would be permitted to engage in nonbanking activities. As banks move dow nw ard into groups with low er capital ratios, they would be subject to in creasingly stringent sanctions, including restric tions on their dividends and the grow th o f their assets. Banks in the lowest group, with relative ly low but still positive capital ratios, w ould be closed unless their shareholders prom ptly in jected new capital.2 d e p a rtm e n t of the Treasury (1991), pp. 38-42, and Chapter X. The U.S. General Accounting Office (GAO, 1991) recently recommended a different system of prompt cor rective action by bank supervisors. The GAO’s proposal requires that the actions of supervisors be tied to specific unsafe banking practices, defined more broadly than capital ratios below some required level. The GAO study criticizes the Treasury proposal for focusing too narrowly on these ratios. FEDERAL RESERVE BANK OF ST. LOUIS Prom pt corrective action b y bank regulators is intended to reduce both the num ber o f bank failures and the losses by the deposit insurance fund. The Treasury proposal w ould reduce the discretion that bank supervisors have in han dling troubled banks, making the sanctions against banks w ith relatively low capital ratios mandatory. The policy is designed to give healthy banks the incentive to keep their capital ratios above the critical levels at which they w ould be subject to mandatory sanctions. Prom pt correc tive action also w ould constrain the actions of undercapitalized banks which might increase ex posure o f the deposit insurance fund to losses and give the owners o f banks w ith relatively low capital ratios the incentive to inject capital into their banks promptly, if they wish to retain control o f their banks. The effectiveness o f the proposed policy in achieving its goals o f low er bank failure rates and reduced losses to the deposit insurance 2The Treasury proposal is not as specific as some earlier proposals. For instance, it does not specify the criteria for classifying banks into the five groups nor does it provide details about the sanctions to be imposed on banks in each group. For a similar, but more detailed proposal, see Benston and Kaufman (1988). 17 fund depend on w hether the actions to be taken by supervisors under the new policy d iffer substantially from the actions taken by super visors in recent years in dealing w ith under capitalized banks. The case fo r the policy o f prompt corrective action rests on the assump tion that, fo r a given capital ratio o f a bank, sanctions under the proposed policy w ould be m ore severe than those imposed by supervisors in recent years. In essence, the Treasury p ro posal is based on the view that supervisors in the past have perm itted banks to remain under capitalized fo r overly long periods and that undercapitalized banks have been permitted to engage in activities that made them m ore likely to fail and m ore likely to increase the deposit insurance fund’s losses. The purpose o f this paper is to investigate w hether the behavior o f troubled commercial banks in recent years is consistent w ith these assumptions. The paper looks at banks whose capital ratios fell below the minimum required level fo r periods longer than one year. It ex amines w hether these undercapitalized banks violated the types o f constraints that w ould be imposed under the Treasury’s scheme fo r prompt corrective action. The paper also considers w hether such violations reduced the chances that the banks would, once again, achieve ac ceptable capital ratios. policy o f prom pt corrective action might specify higher capital levels as the critical levels fo r mandatory actions. This paper examines the behavior o f banks whose prim ary capital ratios rem ained below 5.5 percent fo r m ore than fou r consecutive quarters betw een 1985 and 1989. This choice o f period reflects the fact that most capital injec tions occur in the fourth quarter o f each year, perhaps because o f the practice o f "w in d ow dressing,” w h ere banks devote special attention to the capital ratios that appear on their yearend balance sheets. By focusing on more than four consecutive quarters, w e include only those banks whose prim ary capital ratios remained below 5.5 percent through the fourth quarter o f the year in which they first became under capitalized. Figure 1 illustrates some o f the characteristics o f the banks included in this study. Undercapi talized banks fall into three groups. Those in one group quickly raised their capital ratios by increasing their capital and/or reducing their assets. Another group consists o f those that w ere closed quickly b y their supervisors. Clear ly, no banks in these tw o groups remained un dercapitalized fo r long. This study focuses on a third group o f banks — those that rem ained undercapitalized fo r m ore than fou r consecutive quarters. ENFORCEMENT OF CAPITAL REQUIREMENTS Slow R esponse to Enforcem ent Actions In 1985, federal supervisory agencies estab lished minimum capital requirements fo r all commercial banks. Th e minimum ratio o f prima r y capital to total assets was set at 5.5 percent. This minimum remained in effect until the end o f 1990, when supervisors began phasing in new risk-based capital requirements. The shaded in sert lays out the components o f prim ary capital and total assets used to calculate the prim ary capital ratio and indicates the effects o f loan losses on this ratio. There are several reasons w h y banks can re main undercapitalized fo r m ore than a year. Some banks respond m ore slowly than others to directives from supervisors to raise their capital ratios, in part because they know that super visors lack the authority to close them because o f their low capital ratios alone. Instead, banks must be judged insolvent (that is, w ith zero or negative net w orth) or nonviable by their char tering agencies to justify closure. Because the objective o f this paper is to ex amine how rigorously and consistently super visors have enforced capital requirements in re cent years, it is necessary to identify undercapi talized banks in terms o f the capital requ ire ments in effect at the time. This paper defines undercapitalized banks as those w ith prim ary capital ratios below 5.5 percent. The proposed Supervisors do have a variety o f enforcem ent actions that they can take against undercapital ized banks short o f closing them down. Am ong the m ore severe are the rem oval o f their officers, the imposition o f fines and the termination of insurance coverage on the banks’ deposits. Supervisors generally try first to induce a bank to comply w ith banking regulations w ith less form al or severe enforcem ent actions, like writ- MAY/JUNE 1991 18 An Introduction To Bank Capital Accounting Th e text assumes a basic understanding o f accounting principles applied to the balance sheets o f commercial banks and the items in the capital accounts o f banks. This insert p ro vides an introduction to these topics. This in troduction abstracts from some o f the detail o f bank capital accounting.1 Th e accounting principles can be illustrated by referrin g to the balance sheets o f a hypo thetical bank in tables A1 and A2, constructed as o f Decem ber 31, 1986. Table A1 indicates the values o f balance sheet items under the assumption that there are no loan losses in 1986. Table A2 is a revised balance sheet fo r the same period, indicating the effects o f $1 in loan losses. Table A1 provides definitions o f items in the balance sheet. One o f the key items fo r our purposes is the allowance fo r loan and lease losses. The position o f the allowance for loan and lease losses on the balance sheet, as a negative item on the assets side, reflects its position in the Report o f Condition, which banks file w ith their supervisors. Increases in the allowance, called provision fo r loan and lease losses, are expense items in the calcula tion o f profit and loss. Through periodic pro visions, the bank increased its allowance fo r loan and lease losses to $3. Loan losses are charged against that allowance, rather than against current income or equity capital directly. At some time in the past, the bank purchas ed another firm fo r m ore than its book value, w ith the difference o f $1 recorded as "good w ill.” The bank’s $4 in equity capital reflects the dollar value o f shares sold to stockholders and accumulated retained earnings. The items under “ Capital Accounts” other than equity capital, that is, limited life p referred stock and subordinated notes and debentures, have fixed m aturity dates. Supervisors consider these items less desirable form s o f bank capi tal than equity because o f their fixed maturi ty dates. Funds raised b y selling stock, in contrast, are not scheduled to be returned to 'For additional detail on the minimum capital re quirements adopted by the federal bank supervisors in 1985, see Gilbert, Stone and Trebing (1985). FEDERAL RESERVE BANK OF ST. LOUIS shareholders at any particular point in time and, to conserve capital, banks can forego dividend payments to shareholders. W ithin limits, the limited life p referred stock and subordinated notes and debentures are in cluded in total capital, but not prim ary capital. In 1985, the federal bank supervisors set the minimum requirem ent o f total capital to total assets at 6 percent. Based on the values o f the items in table A l, the hypothetical bank has a prim ary capi tal ratio o f 6 percent. Table A2 indicates the impact on the prim ary capital ratio if the bank had a loan loss o f $1 in 1986. T o simpli fy the example, suppose the bank pays no dividends or income taxes. A fter recognizing the $1 loan loss, the bank wants to keep its allowance fo r loan and lease losses equal to $3, to cover potential losses on the remaining $50 in gross loans. Th e effects on the balance sheet can be il lustrated in tw o steps. First, the provision fo r loan and lease losses in 1986 is increased by $1, tem porarily raising the allowance fo r loan and lease losses to $4. The increase in the provision fo r loan and lease losses reduces net income fo r 1986 by $1, relative to the case illustrated in table A l . The reduction in net income o f $1 reduces equity capital from $4 to $3, reflecting low er retained earnings. In the next step, gross loans are reduced by $1, and the allowance fo r loan and lease losses is reduced by $1, back to $3. Prim ary capital is reduced to $5 and the prim ary capital ratio to 5.05 percent. These entries il lustrate how the loan losses o f a bank can reduce its prim ary capital below the minimum required level. The nature o f the change in balance sheet items from table A l to table A2 can also il lustrate how losses can make the equity o f a bank negative. Suppose the loan loss w ere $5, instead o f $1. A provision fo r loan and lease losses o f $5 w ould make equity capital equal to negative $1. 19 Table A1 Balance Sheet of a Hypothetical Bank, December 31, 1986 Status of loan losses: none in 1986 ASSETS Cash $ 5 Securities 45 Loans Gross loans Reserye for loan and lease losses Net loans LIABILITIES AND CAPITAL Liabilities Deposits $92 Capital Accounts $50 3 47 4 1 1 Equity capital Limited life preferred stock Subordinated notes and debentures 1 Goodwill $98 $98 Definitions: Allowance for loan and lease losses: the amount set aside to absorb anticipated losses. In creases in the allowance are an expense item in calculating profit and loss. All charge-offs of loans and leases are charged against this capital account, and recoveries on loans and leases previously charged off are credited to this capital account. Goodwill: Purchase price of firms that have been acquired in excess of their book value. Equity capital: Includes the following components: a. b. c. d. Perpetual preferred stock Common stock: the par or stated value of outstanding common stock Surplus: amount received from the sale of common stock in excess of par or stated value Undivided profits: accumulated value of retained earnings Limited life preferred stock: Preferred stock with maturity dates. Subordinated notes and debentures: Debt obligations with fixed maturity dates. They are subor dinated to deposits. If a bank fails, the holders of its subordinated notes and debentures receive payment only if depositors are paid in full. Primary capital: Equity capital, plus allowance for loan and lease losses, minus goodwill = 4 + 3 - 1 = 6. Total capital: Primary capital plus limited life preferred stock plus subordinated notes and deben tures = 6 + 1 + 1 = 8. Primary capital ratio: Primary capital divided by total assets plus allowance for loan and lease losses minus goodwill = 6 + (98 - 1 + 3) = 0.06 Table A2 Balance Sheet of a Hypothetical Bank, December 31, 1986 Status of loan losses: loss of $1 in 1986 ASSETS Cash $ 5 Securities 45 Loans Gross loans Reserye for loan and lease losses Net loans $49 Deposits $92 Capital Accounts 3 46 Equity capital Limited life preferred stock Subordinated notes and debentures 3 1 1 1 Goodwill $97 Primary capital: 3 + 3 - 1 LIABILITIES AND CAPITAL Liabilities $97 =5. Primary capital ratio: 5 + (97 + 3 - 1) = .0505. MAY/JUNE 1991 20 Figure 1 Classification of Undercapitalized Banks C lassification o f banks by capital adequacy Status o f undercapitalized banks Reasons banks remain undercapitalized Behavior of undercapitalized banks ten or verbal agreements w ith the bank’s officers and directors.3 Thus, considerable time can pass before supervisors feel the need to resort to m ore severe enforcem ent actions. The terms o f such actions—w hether form al or inform al—depend on the conditions and circum stances at each bank. Most enforcem ent actions require the banks’ officers and directors to sub mit a plan to restore their banks’ capital ratios to adequate levels. Other actions include restric tions on grow th o f total assets, dividends, and loans to officers and directors. Enforcement ac tions may also address violations o f specific regulations. The various restrictions typically imposed on bank behavior are similar to those that would be imposed under the Treasury proposal. This proposal, therefore, does not involve new types 3Spong (1990), pp. 90-93. For descriptions of specific enforcement actions by the Office of the Comptroller of the Currency in recent years, see articles entitled “ Special Supervision and Enforcement Activities” in various issues of the Quarterly Journal of the Comptroller of the Currency. FEDERAL RESERVE BANK OF ST. LOUIS o f restrictions on undercapitalized banks. Rather, it calls fo r m ore rigorous and less discretionary enforcem ent o f these restrictions, facilitated by legislation that would limit the ability o f banks to impede prom pt action by supervisors through litigation. Slow to Close D u e to P roblem s with Finding Buyers Another reason w h y banks may have prim ary capital ratios below 5.5 percent fo r extended periods is if they w e re kept open w hile their supervisors searched fo r other banks to buy them. Such cases w e re especially common in Texas, w here considerable time passed before buyers could be found fo r some troubled bank holding companies.4 4Bovenzi and Muldoon (1990), p. 4. 21 Capital Forbearance Still other undercapitalized banks w ere granted official forbearance by their super visors. Forbearance occurs w hen supervisors decide to forego enforcem ent o f some regula tions, including capital requirements, under special circumstances. As their losses on agricul tural and energy loans rose in the 1980s, many banks turned to Congress fo r relief from capital requirements. In the Competitive Equality Bank ing Act (CEBA) o f 1987, Congress mandated cap ital forbearance fo r agricultural banks. Banks in this program w ere perm itted to defer form al acknowledgem ent o f losses on agricultural loans fo r several years. The typical rationale fo r capital forbearance is that the economic forces respon sible fo r the declines in capital ratios, such as low er farm income and reduced prices o f farm land, are only temporary. In response to this evidence o f congressional intent, the federal supervisory agencies estab lished capital forbearance programs that set broader terms fo r participation than those speci fied in CEBA. Cobos (1989), fo r example, de scribes the capital forbearance program o f the Federal Deposit Insurance Corporation (FDIC) and gives his perspective, as an FDIC official, on the objectives o f the program and the actions that banks granted forbearance are expected to follow . According to Cobos, banks granted fo r bearance should be considered viable by their supervisors; they also should be expected to: 1. limit the grow th o f their total assets and relatively high risk investments. 2. restrict dividends to their shareholders. 3. limit the benefits o f forbearance to insiders, including insider loans.5 Th ere is virtually no w ay to tell which of these three reasons explains w h y any particular bank rem ained undercapitalized fo r an extended period. In fact, m ore than one o f these reasons may apply. Knowing w h y these banks remained undercapitalized is unimportant, how ever, if bank supervisors generally expect them to con form to similar constraints on their behavior. That is, regardless o f w hy they w ere allowed to 5Cobos (1989). 6The banks included in this study are domestically owned commercial banks. Savings banks and foreign owned banks are excluded, as are several special purpose banks, including bankers’ banks. remain undercapitalized fo r so long, these banks should at the very least not have ex perienced rapid asset growth, paid dividends to shareholders or increased their loans to in siders. This paper investigates w hether banks that w e re undercapitalized fo r m ore than a year indeed conform ed to these constraints. THE CHARACTERISTICS OF UNDERCAPITALIZED BANKS Table 1 indicates that 531 federally insured comm ercial banks w ere undercapitalized fo r m ore than a year, about 4 percent o f the average number o f banks operating in the years 1985-89.6 The vast m ajority (87 percent) o f these inade quately capitalized banks w ere relatively small, with total assets o f less than $100 million. Undercapitalized banks whose assets exceeded $100 million w e re concentrated in the energyproducing states o f Louisiana, Oklahoma and Texas (63 percent). Only one undercapitalized bank (located in California) had total assets greater than $1 billion. Outside Texas, a majority o f the undercapital ized banks (60 percent) w ere state nonmem ber banks, supervised by the FDIC. In Texas, in con trast, 73 percent w ere national banks, supervised by the O ffice o f the Com ptroller o f the Currency (OCC).7 As table 1 shows, a sizable proportion o f the 531 banks had prim ary capital ratios below 5.5 percent fo r tw o years or more. In this 20-quarter period (1985-89), 178 banks had prim ary capital ratios below the minimum level fo r eight or m ore consecutive quarters, and six had capital ratios below this level fo r 16 or m ore quarters. Table 1 also shows that banks in 22 states had negative equity capital fo r at least one quarter.8 Some o f these observations may reflect lags in the process b y w hich supervisors get inform a tion on banks and arrange fo r their resolution; they are not necessarily evidence o f supervisory policies that perm it banks w ith negative equity to remain in operation. Indeed, fo r most o f these banks, the period o f negative equity lasted only one or tw o quarters. Some banks, how ever, had under the same federal supervisory authorities in the years 1985-89. 8See the shaded insert for a description of equity capital and the type of accounting entries that can make equity capital negative. HTie undercapitalized banks included in this study remained MAY/JUNE 1991 22 Table 1 Characteristics of Banks with Primary Capital Ratios Below 5.5 Percent for Over Four Consecutive Quarters, 1985-89__________________________________ Number of under capitalized banks Banks with assets greater than $100 million OCC FR FDIC Alaska Arizona California Colorado Connecticut Florida Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Massachusetts Michigan Minnesota Missouri Montana Nebraska New Hampshire New Jersey New Mexico New York North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Dakota Tennessee Texas Virginia West Virginia Wisconsin Wyoming 2 1 25 15 1 9 1 11 8 12 26 4 43 3 3 23 10 6 7 1 3 5 3 1 6 54 6 1 1 2 6 223 4 2 1 2 2 1 5 4 0 1 0 0 0 0 0 0 9 0 0 0 0 0 1 1 1 2 2 0 1 7 2 0 0 1 1 27 0 0 0 0 0 0 5 10 1 4 0 5 3 2 4 0 8 2 0 5 0 3 2 0 3 3 2 1 3 24 0 1 0 2 1 162 2 0 0 1 0 0 2 4 0 4 0 0 0 1 0 0 0 0 1 1 2 3 0 0 0 0 1 0 1 4 0 0 0 0 0 7 1 0 0 0 2 1 18 1 0 1 1 6 5 9 22 4 35 1 2 17 8 0 5 1 0 2 0 0 2 26 6 0 1 0 5 54 1 2 1 1 0 0 11 4 0 1 1 4 3 5 10 0 13 0 1 12 2 2 4 0 1 3 1 0 2 18 3 0 0 0 2 73 2 1 0 0 2 1 4 7 1 6 0 0 3 2 5 0 18 0 0 4 4 2 0 0 0 2 0 0 1 29 0 1 0 1 1 117 1 1 0 0 1 0 1 1 0 1 0 0 0 0 1 0 4 0 0 2 1 1 0 0 0 1 0 0 1 10 0 0 0 0 0 46 1 0 0 0 0 0 10 7 0 4 1 9 4 7 7 4 2 3 2 9 4 2 4 1 2 1 1 0 3 5 4 0 1 1 4 26 1 0 1 0 Totals 531 68 259 32 240 178 213 72 130 State1 Federal supervisory agency Banks with capital ratios below the required level for eight or more consecutive quarters Banks with negative equity capital for: At least one quarter Four or more quarters 'States that are not listed had no banks that were undercapitalized for five or more consecutive quarters during 1985-89. 2Banks that recover are identified as those whose primary capital ratios were consistently above 5.5 percent by IV/1989. Note: Identification of federal supervisory agencies: OCC = Office of the Comptroller of the Currency FR = Federal Reserve FDIC = Federal Deposit Insurance Corporation FEDERAL RESERVE BANK OF ST. LOUIS Banks that recover2 23 negative equity fo r extended periods o f time. O f the 72 banks w ith negative equity fo r fo u r or m ore quarters, 83 percent w ere in Louisiana, Oklahoma and Texas; tw o national banks in Texas had negative equity fo r nine quarters. The last column o f table 1 shows the number o f undercapitalized banks that recovered, that is, had prim ary capital ratios consistently above 5.5 percent, by the fourth quarter o f 1989.9 Only 130 o f the 531 banks had recovered by IV/1989, an average recovery rate o f only 24 percent. The 46 percent rate o f recovery fo r banks out side o f Louisiana, Oklahoma and Texas is much higher than the 10 percent recovery rate fo r banks undercapitalized fo r m ore than a year in these energy-producing states. As might be expected, the recovery rates w ere significantly low er fo r banks w ith negative equity. O f the 213 banks w ith negative equity fo r at least one quarter, the recovery rate was only 6.57 percent, com pared w ith a recovery rate o f 36.48 percent among the remaining 318 banks.1 0 The geographic distribution o f the 531 under capitalized banks is quite uneven. For instance, 14 states and the District o f Columbia had no banks that w ere undercapitalized fo r m ore than a year. W hile these 14 are not clustered in any particular part o f the country, they have one characteristic in comm on—relatively liberal branching laws (see table 2). Eleven o f the 14 states permit statewide branching and the three Table 2 Relationship between Number of Undercapitalized Banks and Bank Failures, 1985-89 States grouped by num ber of banks undercapitalized fo r more than one year Number of undercapitalized banks Number of failed banks None1 1 or 22 3-63 7-264 More than 26s 0 16 49 146 320 17 45 40 258 483 Total 531 843 1 Alabama, Arkansas, Delaware, Georgia, Hawaii, Maine, Maryland, Mississippi, Nevada, North Carolina, South Carolina, Utah, Vermont and Washington. 2 Alaska, Arizona, Connecticut, Idaho, New Hampshire, North Dakota, Pennsylvania, Rhode Island, South Dakota, West Virginia, Wisconsin and Wyoming. 3 Kentucky, Massachusetts, Michigan, Montana, New Jersey, New Mexico, New York, Ohio, Oregon, Tennessee and Virginia. 4 California, Colorado, Florida, Illinois, Indiana, Iowa, Kansas, Minnesota, Missouri and Nebraska. 5 Louisiana, Oklahoma and Texas. 9One objection to this definition of recovery is that it understates the actual recovery rate, because many banks’ capital ratios fell below 5.5 percent only near the end of the 1985-89 period. This possibility is investigated by examining capital ratios in the first three quarters of 1990 for those banks whose primary capital ratios were below 5.5 percent in the fourth quarter of 1989. Reclassify ing these banks as recovered if their primary capital ratios rose consistently above 5.5 percent in the first three quarters of 1990 raises the recovery rate from 10 percent to 12.5 percent in the energy-producing states and from 46 percent to 55.5 percent in the other states. The significance of these increases in recovery rates may be offset by the possibility that some other banks, previously classified as recovered, might be reclassified if their capital ratios for 1990 were examined. Since examination of 1990 data did not change the recovery rates substantial ly, the recovery rates cited in the text are those based on observations through IV/1989. 10The difference between these proportions is significant at the 1 percent level. See, for example, Wonnacott and Wonnacott (1990), pp. 273-75, for the equation to test the statistical significance of the difference between two proportions. MAY/JUNE 1991 24 others perm it limited branching. Only 12 o f the other 36 states are classified as statewide branch ing states.1 1 As table 2 indicates, many o f the banks that failed during 1985-89 w e re not undercapitalized fo r a year or longer. Instead, they w ere closed w ithin a year after their capital ratios fell below the minimum required level. For instance, 17 banks failed in the 14 states that had no banks undercapitalized fo r m ore than a year. In the nation, the num ber o f banks that failed exceed ed the num ber that w ere undercapitalized fo r more than a year. These observations suggest that many bank failures in recent years cannot be attributed to actions taken by banks while undercapitalized f o r extended periods o f time; many banks failed before their prim ary capital ratios had fallen below the minimum required level fo r a year or longer, and many banks that w ere undercapitalized fo r extended periods o f time did not fail. ENFORCEMENT OF CONSTRAINTS ON UNDERCAPITALIZED BANKS The Treasury proposal fo r prom pt corrective action is based on the view that supervisors have delayed too long in imposing constraints on undercapitalized banks. This section investi gates w hether the banks that w ere undercapital ized fo r over m ore than a year violated the types o f constraints that w ould be imposed under the Treasury proposal. T w o constraints on the behavior o f under capitalized banks mentioned in the Treasury proposal are examined here: constraints on asset grow th and dividend payments.1 A third con 2 straint is also investigated: no increase in loans to bank officers and directors (the bank in siders) w hile a bank is undercapitalized. An ex amination o f insider loans is included because supervisory authorities often focus on such loans w hen m onitoring the condition o f under 1 See Conference of State Bank Supervisors (1986), p. 83. 1 This classification is based on state laws as of January 1986. 12Department of the Treasury (1991), p. 39. 13Kummer, Arshadi and Lawrence (1989). 14Twelve banks with asset growth in excess of 10 percent were involved in mergers during the periods in which their primary capital ratios were below 5.5 percent. Mergers distort the measure of asset growth for the purposes of this paper, because they increase capital and assets. Asset growth of banks engaged in mergers does not necessarily reflect greater leverage. Some mergers, for ex FEDERAL RESERVE BANK OF ST. LOUIS capitalized banks. Also, undercapitalized savings and loan associations in the recent past w ere found to have increased the losses to their de posit insurance funds through loans to insiders, and one study has found that banks w ith rela tively high ratios o f insider lending to total assets had low er earnings and higher bank fail ure rates than other banks.1 3 Table 3 presents selected inform ation about the behavior o f undercapitalized banks. The results are divided into regions, except fo r Loui siana, Oklahoma and Texas, which account fo r most o f the undercapitalized banks. This method o f presentation highlights both regional concen trations o f undercapitalized banks and dif ferences in bank behavior. Asset Growth W hen the capital ratio o f a bank falls to a re latively low level, its shareholders have less to lose and, correspondingly, m ore to gain by assuming additional risk, in the hope o f a suffi ciently large turnaround in income to eliminate the capital deficiency. One w ay that a bank assumes additional risk is to increase its assets. Bank supervisors, o f course, p re fer to see under capitalized banks reduce their assets, to raise their capital ratios. Most o f the 531 banks reduced their assets substantially w hile undercapitalized, consistent w ith the desires o f bank supervisors. A sizable number, how ever, actually experienced rapid asset growth. At 84 banks (16 percent o f the total), asset grow th exceeded 10 percent while prim ary capital ratios w ere below 5.5 percent; in fact, asset grow th exceeded 25 percent at 44 undercapitalized banks.1 Most banks whose 4 asset grow th exceeded 25 percent w ere Texas national banks supervised by the OCC—26 o f the 28 Texas banks in this study w ith asset grow th in excess o f 25 percent w ere national banks.1 5 ample, involve subsidiaries of the same holding com panies, which do not change the leverage of these holding companies. For each of these 12 banks, asset growth in the period in which their primary capital ratios were below 5.5 percent is measured as the percentage change in the period before or after the merger, whichever is the longer. 15See O’Keefe (1990) for a thorough discussion of the per formance problems of Texas banks and the problems with bank supervision in Texas in recent years. 25 Table 3 Behavior of Undercapitalized Banks Banks w ith gro w th in assets w hile capital ratios below the required level Census Region New England Middle Atlantic South Atlantic East South Central West South Central1 Louisiana Oklahoma Texas East North Central West North Central Pacific Northwest Pacific Southwest TOTAL Number of banks Asset g ro w th in excess of 10 percent 4 4 3 1 53 6 7 15 10 320 Asset grow th in excess of 25 percent 2 0 2 0 28 0 3 50 43 54 223 Banks that paid dividends w hile undercapitalized Banks w ith the ir highest levels of insider loans when undercapitalized 1 1 3 3 71 2 2 3 3 46 11 7 28 0 0 28 10 14 47 29 81 17 46 2 9 2 6 0 6 2 4 3 14 2 5 9 20 1 17 531 84 44 79 126 ^ h e West South Central Region also includes Arkansas, which had no banks that were undercapitalized for more than four consecutive quarters. Note: States in census regions: New England: Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island and Vermont Middle Atlantic: New Jersey, New York and Pennsylvania South Atlantic: Delaware, Florida, Georgia, Maryland, North Carolina, South Carolina, Virginia and West Virginia East South Central: Alabama, Kentucky, Mississippi and Tennessee East North Central: Illinois, Indiana, Ohio, Michigan and Wisconsin West North Central: Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota and South Dakota Pacific Northwest: Alaska, Idaho, Montana, Oregon, Washington and Wyoming Pacific Southwest: Arizona, California, Colorado, Hawaii, Nevada, New Mexico and Utah. Dividends A bank's capital absorbs its losses, thereby protecting uninsured depositors and the deposit insurance fund from the prospect o f the bank's failure. Dividends reduce this capital cushion. Thus, a bank must obtain permission from its supervisors to pay dividends that exceed its cur rent earnings, and supervisors can restrict the payment o f dividends if a bank's capital ratio is below the required level.1 About 15 percent o f 6 the banks in this study, how ever, paid dividends on their common stock w hile their prim ary capi tal ratios w ere below the minimum level. 16Spong (1990), pp. 64-71. ^D epartm ent of the Treasury (1991), pp. X-12 through X-14. The recent Treasury study reports similar fin dings on dividends paid b y undercapitalized banks. In 1989, fo r instance, about 14 percent o f the 525 banks w ith ratios o f equity capital to assets below 4.5 percent paid dividends.1 7 Loans to Insiders Banks are perm itted to make loans to their o f ficers and directors (or "insiders").1 If a bank’s 8 shareholders and insiders are not exactly the same group, the shareholders have an incentive to limit insider loans, because o f the “moral hazard” o f having insiders assess their ow n 18See Spong (1990), pp. 60-63, for a description of insider lending regulations. MAY/JUNE 1991 26 creditworthiness. If, how ever, a bank’s capital (and, hence, its capital ratio) has fallen to a relatively lo w level, shareholders may exert less constraint on insider lending simply because they have less to lose. Th erefore, w hen banks becom e undercapitalized, supervisory actions to limit insider loans take on greater importance in limiting the deposit insurance fund’s losses. Table 3 displays information on the number o f undercapitalized banks that reported their highest levels o f insider lending w hile under capitalized. This measure is important because it reflects the maximum exposure b y these banks to losses on such loans. A sizable minority of the banks (about 24 percent) reported their highest levels o f insider loans w hen their capital ratios w e re below the minimum required level.1 9 Differences in Constraints Across Supervisors D ifferences in practices among the super visors o f comm ercial banks may explain some o f the variation in behavior among the undercapi talized banks. As noted above, most undercapi talized banks w ith rapid asset grow th w ere na tional banks located in Texas. In Texas, h o w ever, national banks are a majority o f all com m ercial banks in the state. The concentration o f national banks in Texas among the various groups o f undercapitalized banks may reflect simply the relatively large share o f national banks in Texas. Table 4 examines the behavior o f Texas banks by federal supervisory agency. The first ro w presents the distribution o f these banks by their federal supervisory agency. The follow ing row s show the numbers o f undercapitalized banks in various categories by federal supervisory agen cy. Below the numbers o f undercapitalized banks are their percentages o f the total num ber o f Texas banks supervised by the same federal agency. Asterisks indicate w hether the prop or 19The 12 banks involved in mergers while undercapitalized may have had their insider loans rise while undercapitaliz ed because they merged with banks that had insider loans before the mergers. To avoid such biases, these 12 banks are classified among those that did not have their highest level of insider loans while undercapitalized. 20This paper compares recovery rates across banks rather than failure rates because the distinction between failed and surviving banks is rather arbitrary in some cases. For example, banks with negative equity for several con secutive quarters would be classified as survivors simply because they remained in operation. FEDERAL RESERVE BANK OF ST. LOUIS tions fo r national banks are significantly d if ferent from those fo r state-chartered insured banks. Table 4 shows substantial differences in the representation o f national and state-chartered banks among the undercapitalized banks in Texas. Almost 18 percent o f the national banks had prim ary capital ratios below 5.5 percent fo r m ore than four consecutive quarters, compared w ith about 8 percent fo r state-chartered banks. O ver 6 percent o f national banks w ere under capitalized fo r eight or m ore consecutive quar ters, com pared with 1.8 percent fo r statechartered banks. Most o f the banks with negative equity are national banks, especially those w ith negative equity fo r four or m ore quarters. National banks also account fo r most o f the undercapitalized banks w ith rapid asset growth. Table 5 makes the same comparisons among national and state-chartered insured banks out side o f Texas. The only significant differences in proportions fo r banks outside o f Texas involve undercapitalized banks w ith negative equity. Significantly higher proportions o f national banks had negative equity than fo r state-chartered banks. The other differences in proportions are not significantly different from zero. Thus, the relatively high concentrations o f national banks among the various groups o f undercapitalized banks w ere prim arily in Texas. The Behavior o f Undercapitalized Banks and their R e c o v e ry Rates Because some undercapitalized banks violated the constraints that w ould be imposed under the Treasury proposal fo r prom pt corrective ac tion, w e can test w hether these constraints w ould have had a positive effect on bank capital recovery rates. If such constraints tend to in crease the recovery rate, w e should expect low er recovery rates among the banks that violated these constraints.2 0 27 Table 4 Distribution of Federally Insured Commercial Banks in Texas by Their Primary Supervisory Agency Federal supervisory agency OCC FR FDIC Average number of banks 1985-89 Banks with primary capital ratios below 5.5 percent for more than four consecutive quarters 916 162** (17.69%) Total num ber of banks 76 705 1,697 7 (9.21%) 54 (7.66%) 223 Of these undercapitalized banks, number with the following characteristics: Primary capital ratios below 5.5 percent for eight or more consecutive quarters Negative equity for at least one quarter Negative equity for four or more consecutive quarters Asset growth in excess of 10 percent while undercapitalized Asset growth in excess of 25 percent while undercapitalized Paid dividends while undercapitalized Highest level of insider loans while undercapitalized 59** (6.44%) 2 (2.63%) 12 (1.70%) 73 97 ** (10.59%) 2 (2.63%) 18 (2.55%) 117 45** (4.91%) 0 (0.00%) 1 (0.14%) 46 40** (4.37%) 4 (5.26%) 6 (0.85%) 50 26 ** (2.84%) 2 (2.63%) 0 (0.00%) 28 13 (1.42%) 0 (0.00%) 15 (2.13%) 28 30** (3.28%) 3 (3.95%) 14 (1.99%) 47 Note: Figures in parentheses are percentages of the total number of banks supervised by that agency. Single asterisk (*) indicates that the proportion of banks supervised by the OCC is significantly different from the proportion of state-chartered insured banks at the 5 percent level. Double asterisk (**) indicates that the proportion is significantly different at the 1 percent level MAY/JUNE 1991 28 Table 5 Distribution of Federally Insured Commercial Banks Outside Texas by Their Primary Supervisory Agency Federal supervisory agency OCC FR FDIC Average number of banks 1985-89 Banks with primary capital ratios below 5.5 percent for over four consecutive quarters 3,689 97 (2.63%) Total num ber o f banks 1,018 7,261 11,968 25 (2.46%) 186 (2.56%) 308 Of these undercapitalized banks, number with the following characteristics: Primary capital ratios below 5.5 percent for 8 or more consecutive quarters Negative equity for at least one quarter Negative equity for 4 or more consecutive quarters Asset growth in excess of 10 percent while undercapitalized Asset growth in excess of 25 percent while undercapitalized Paid dividends while undercapitalized Highest level of insider loans while undercapitalized 31 (0.84%) 8 (0.79%) 66 (0.91%) 105 47** (1.27%) 7 (0.69%) 42 (0.58%) 96 15* (0.41%) 0 (0.00%) 11 (0.15%) 26 12 (0.33%) 4 (0.39%) 18 (0.25%) 34 3 (0.08%) 3 (0.29%) 10 (0.14%) 16 18 (0.49%) 0 (0.00%) 33 (0.45%) 51 28 (0.76%) 6 (0.59%) 45 (0.62%) 79 Note: Figures in parentheses are percentages of the total number of banks supervised by that agency. Single asterisk (*) indicates that the proportion of banks supervised by the OCC is significantly different from the proportion of state-chartered insured banks at the 5 percent level. Double asterisk (**) indicates that the proportion is significantly different at the 1 percent level. FEDERAL RESERVE BANK OF ST. LOUIS 29 Table 6 Recovery Rates of Undercapitalized Banks Total num ber o f banks Percentage tha t recovered by IV/1989 Banks with asset growth exceeding 10 percent Other banks 84 447 23.81% 24.61 (0.16) Banks with asset growth exceeding 25 percent Other banks 44 487 22.73 24.64 (0.29) Banks that paid dividends while undercapitalized Other banks 79 452 32.91 23.01 (1.75) Banks that increased insider loans while undercapitalized Other banks 126 405 24.60 24.44 (0.04) Note: Absolute values of t-statistics for tests of differences in proportions are in parentheses. Table 6 compares the recovery rates o f banks that violated the constraints w ith those that did not. The recovery rates o f the tw o groups o f banks are not significantly different. The recov ery rate fo r banks that paid dividends is slightly higher than that fo r the other undercapitalized banks, although this difference is not statistical ly significant at the 5 percent level. Similarly, the other observed differences are not significantly different from zero. A comparison o f recovery rates in table 6 shows that asset grow th, dividends and insider loans are not important determinants o f recov ery. These results im ply that imposing constraints on this behavior should not significantly affect the recovery rates o f undercapitalized banks.2 1 CONCLUSIONS Bank supervisory reform is a major com po nent o f the overall plan fo r deposit insurance reform recently proposed by the U.S. Depart ment o f the Treasury. Under this proposal, su pervision w ould be based on the capital ratios 2 Dahl and Spivey (1991) report similar results. They ex 1 amine the characteristics of undercapitalized banks that help distinguish between those that once again have capital ratios above the required level and those that do o f banks. I f a bank's capital ratio fell below the level acceptable to supervisors, it w ould be sub ject to mandatory constraints on its behavior. This proposal fo r prom pt corrective action would limit the discretion o f supervisors in dealing w ith undercapitalized banks. The proposal's objective is to reduce the num ber o f bank failures and the losses by the de posit insurance fund. Advocates o f the proposal assume that imposing sanctions on banks whose capital ratios fall below critical levels w ould give the managers o f healthy banks the incentive to keep their capital ratios above the critical levels at which the sanctions becom e mandatory. By authorizing the closing o f banks with low but positive capital ratios, the proposal gives shareholders o f undercapitalized banks incentive to inject capital promptly, if they wish to retain control o f their banks. Finally, the sanctions are assumed to constrain the types o f behaviors that make undercapitalized banks m ore likely to fail and to increase the losses o f the deposit in surance fund. not recover. They find that asset growth and dividends do not help distinguish between the banks that recover and those that do not. MAY/JUNE 1991 30 In the years 1985-89, many banks remained in operation fo r extended periods o f time w ith prim ary capital ratios below the minimum re quired level. Substantial minorities o f these un dercapitalized banks violated the constraints that w ould be imposed under the proposed policy o f prom pt corrective action. This behavior, p re sumably, w ould not be perm itted under the proposed policy. The evidence does not support the hypothesis that once the capital ratio o f a bank falls below the minimum required level, enforcing the sanc tions specified in the Treasury proposal w ould increase the chances that the undercapitalized bank w ill recover. The recovery rates o f under capitalized banks that violated these constraints w ere no lo w er than the recovery rates o f other undercapitalized banks. Data are not available to test the hypothesis that the failures o f the banks violating the constraints specified in the proposal fo r prom pt corrective action imposed relatively large losses on the deposit insurance fund. Thus, if the proposed policy o f prom pt correc tive action can be expected to reduce the rate o f bank failure, this effect must w ork through the incentives fo r healthy banks to keep their capital ratios relatively high. T o draw conclu sions about the strength o f this incentive, it is necessary to make assumptions about h ow bank management w ould view the penalties to be im posed on banks w ith and without legislation re quiring prom pt corrective action by supervisors. Evidence presented in this paper indicates that the sanctions imposed on undercapitalized banks in recent years have been similar to those to be imposed under the proposed policy. First, several hundred banks w e re closed in recent years shortly after their capital ratios fell below the minimum required level. Th eir failure did not result from violation o f the types o f constraints that w ould be imposed under the Treasury proposal. Resolutions o f these cases appear to have been handled much as they w ould under the policy o f prom pt corrective action. Second, while a large number o f banks had capital ratios below the required level fo r m ore than a year, most o f them did not violate the constraints to be imposed under the policy o f prom pt corrective action. Indeed, the fact that a large share o f the cases in which undercapitalized banks violated these FEDERAL RESERVE BANK OF ST. LOUIS constraints involves banks in one state under the jurisdiction o f one supervisory agency sug gests that such cases are the exception, rather than the norm. Thus, there is some evidence that the nature o f bank supervision in recent years provided banks w ith the incentive to keep their capital ratios above the required level w ith out additional legislation. The evidence on recovery rates o f the banks that w ere undercapitalized fo r m ore than a year provides empirical support fo r one element o f the Treasury proposal: the prom pt closing o f banks w ith low but positive capital ratios unless their shareholders act prom ptly to raise their capital ratios. As this paper shows, only 24 p er cent o f the undercapitalized banks recovered in the period examined. Thus, the Treasury p ro posal w ould not result in prem ature closings of large numbers o f banks that ultimately w ould recover if given enough time. REFERENCES Benston, George J., and George G. Kaufman. Risk and Solvency Regulation of Depository Institutions: Past Policies and Current Options, Monograph Series in Finance and Economics, Graduate School of Business Administration, New York University, Monograph 1988-1. Bovenzi, John F., and Maureen E. Muldoon. “ FailureResolution Methods and Policy Considerations,” FDIC Banking Review (Fall 1990), pp. 1-11. Cobos, Dean Forrester. “ Forbearance: Practices and Pro posed Standards,” FDIC Banking Review (Spring/Summer 1989), pp. 20-28. Conference of State Bank Supervisors. A Profile of StateChartered Banking (1986). Dahl, Drew, and Michael F. Spivey. "Moral Hazard, Equity Is suance and Recoveries of Undercapitalized Banks,” Pro ceedings of a Conference on Bank Structure and Competi tion, Federal Reserve Bank of Chicago, May 1991. Department of the Treasury. Modernizing the Financial System (February 1991). Gilbert, R. Alton, Courtenay C. Stone, and Michael E. Trebing. “ The New Bank Capital Adequacy Standards,” this Review (May 1985), pp. 12-20. Kummer, Donald R., Nasser Arshadi, and Edward C. Lawrence. “ Incentive Problems in Bank Insider Borrowing,” Journal of Financial Services Research (October 1989), pp. 17-31. O’Keefe, John. “ The Texas Banking Crisis: Causes and Con sequences, 1980-89,” FDIC Banking Review (Winter 1990), pp. 1-34. Spong, Kenneth. Banking Regulation: Its Purposes, Imple mentation, and Effects, 3rd ed., (Federal Reserve Bank of Kansas City, 1990). U.S. General Accounting Office. Deposit Insurance: A Strategy for Reform (March 1991). Wonnacott, Thomas H., and Ronald J. Wonnacott. Introduc tory Statistics for Business and Economics (John Wiley and Sons, 1990). 31 James B. Bullard James B. Bullard is an economist at the Federal Reserve Bank of St. Louis. David H. Kelly provided research assistance. The FOMC in 1990: Onset of Recession T HE FOURTH QUARTER OF 1990 marked the end o f a long expansion o f the U.S. economy, extending almost continuously from the final quarter o f 1982. In Novem ber, industrial pro duction plummeted at an annual rate o f 19.8 percent, and civilian em ployment fell by nearly 450,000. The most recent estimate o f real gross national product (GNP) indicates that it fell at an annual rate o f 1.6 percent in the fourth quarter, and the unemployment rate climbed from 5.3 percent in June to 6.1 percent by the end o f the year. By all accounts, recession had arrived.1 Because the U.S. econom y entered the reces sion in the latter portion o f the year, calendar 1990 is an interesting period in which to sum marize the actions o f the Federal Open Market Committee (FOMC).2 By considering FOMC direc tives chronologically, this paper w ill develop a case study o f m onetary policymaking during the onset o f a recession. W ithin the context o f the chronology, emphasis w ill be placed on tw o types o f uncertainty faced by the Committee. First, there is uncertainty about the immediate past, current and future path o f real output, a prim ary measure o f economic activity. Second, 'According to the Federal Reserve Board’s 1991 Monetary Policy Objectives (p. 3), “ The [U.S.] economy . . . fell into recession in the latter part of 1990, and . . . that recession has clearly continued into the early part of 1991.” The Na tional Bureau of Economic Research (NBER), which makes official decisions on dating business cycle peaks and troughs, recently announced that the expansion peak ed in July 1990. there is uncertainty about the thrust o f m onetary policy at a point in time, because o f the various ways the policy stance can be measured. W hile many other considerations enter into FOMC policymaking, as w ill be shown, these tw o fac tors loom large in the Committee's attempts to react swiftly and effectively to economic events. The next section provides the background fo r understanding Committee decision-making in 1990. It introduces the FOMC’s stated objectives and illustrates briefly how the Committee might hope to bring them into balance. This back ground is crucial to an understanding o f the bulk o f the paper, the chronology o f FOMC decision-making contained in the subsequent section. The final portion o f the paper provides summary comments. A FRAMEWORK FOR ANALYZING FOMC POLICY ACTIONS No analysis o f FOMC actions can take place until some context fo r the decision-making is provided.3 T o make sense o f the subsequent chronology, it is essential to understand what 2See the shaded insert, “ The Organization of the FOMC,” for a description of the Committee. 3The terms “ decision-m aking” and “ policy actions” are used interchangeably in this article. MAY/JUNE 1991 32 The Organization o f the F O M C The Federal Reserve Board o f Governors consists o f seven members, and each o f these members has voting rights on the Federal Open Market Committee (FOMC). The presi dent o f the N ew York Federal Reserve Bank also is a permanent voting m em ber o f the FOMC. The remaining 11 Federal Reserve Bank presidents attend meetings and present views, but only four o f these 11 have voting privileges at any one meeting. The voting rights are held fo r terms o f one calendar year and rotate among these presidents annually. Th e Committee typically meets eight times per year, as it did in 1990, and sometimes consults by telephone betw een scheduled meetings. At the end o f each meeting, the Committee agrees on a directive to issue to the Federal Reserve Bank o f N ew York. The directive contains instructions fo r the con the Committee is trying to do and h ow it might hope to achieve its desires. These are matters o f controversy in macroeconomics, and the con troversy w ill not be resolved in this article. In stead, the follow ing fram ew ork fo r understand ing FOMC decision-making relies prim arily on official Committee statements and simple em pirical illustrations.4 Potential interpretations or conclusions are left to the reader. F O M C M onetary P olicy Objectives The FOMC stated its goals fo r m onetary policy in each o f the eight directives it issued in 1990. Specifically, the objectives o f the Committee w ere to foster progress tow ard price stability and to prom ote sustained real output grow th .5 Implementation o f these objectives was general ly achieved by Committee-ordered intervention “The official summary of Committee deliberations is con tained in the ‘‘Record of Policy Actions of the Federal Open Market Committee” for each meeting, released to the press shortly after the subsequent regular meeting and later published in the Federal Reserve Bulletin and the Board’s Annual Report. References to ‘‘the Record” and to press releases in this article refer to this document. 5The following sentence appears in every 1990 directive: “ The [FOMC] seeks monetary and financial conditions that will foster price stability, promote growth in output on a FEDERAL RESERVE BANK OF ST. LOUIS duct o f open market operations until the next regularly scheduled meeting. A summary o f each FOMC meeting is re leased to the press within a fe w business days follow ing the next regularly scheduled meeting and is subsequently published in the Federal Reserve Bulletin. The summary, known as the "Record o f Policy Actions o f the Fed eral Open Market Committee,” is prepared by the Board staff and review ed b y the Commit tee. It typically contains: (1) A synopsis o f re cent econom ic data, (2) A review o f recent open market operations and money market conditions, (3) A Board staff projection o f likely near-term econom ic developments, (4) A summary o f Committee deliberations, (5) The policy directive along w ith a record o f votes and dissenting comments, and (6) A summary o f any other business conducted. in the market fo r reserves or, in Committee parlance, by altering the " . . . degree o f pressure on reserve positions.”6 Committee members sometimes reconcile the tw o policy objectives by view ing price stability as a long-run goal and, correspondingly, sustain ed real output grow th as a short-run goal. For instance, one summary o f a Committee discus sion cites some members arguing that "an eas ing o f short-run policy if such w ere needed to help avert a cumulative deterioration in econom ic activity . . . w ould not be inconsistent with the Committee’s long-term commitment to price stability.”7 Similarly, references are sometimes made to "the Committee’s long-run, anti-inflation strategy.”8 The next section illustrates, via a sim ple empirical exercise, one sense in which price stability is a long-run goal. sustainable basis, and contribute to an improved pattern of international transactions.” The third objective, more am biguous than the first two, also plays a role in the analysis to follow. 6This terminology appears in every FOMC directive in 1990. The market for reserves is discussed in more detail below. 7March Press Release, pp. 12-13. 8August Press Release, p. 13. 33 Controlling Inflation The FOMC’s objective o f controlling inflation arises out o f a generally accepted proposition that the Committee has considerable influence over the long-run rate o f price level change. For instance, at the February 1990 meeting "the Committee recognized that over time . . . slower M2 grow th w ould be compatible w ith price stability . . ,”9 Since all Committee decision making needs to be understood w ith this prop osition in mind, some evidence on long-run inflation w ill be considered h ere.1 0 Figure 1 The Quantity Theory and International Evidence 1970-1989 Average annual CPI inflation Average annual CPI inflation Following Lucas (1980), consider a version o f the quantity theory o f m oney w ith the key im plication stated as follows: over the long term, an increase in the quantity o f money, appro priately defined, is reflected in an equal and proportionate increase in the price level.1 The 1 proposition can be investigated in an atheoretical way, since there is a wealth o f available interna tional cross-section evidence. Th e evidence p re sented below constitutes an updated version of that marshaled by Vogel (1974) and analyzed in Lucas (1987, 1980) and D w yer and Hafer (1988). Figure 1 provides a plot o f 20-year averages o f grow th rates o f M2 and the associated 20year averages o f annual changes in the con sumer price index fo r 23 OECD countries, 11 Latin American countries and M exico.1 The 2 period covered is 1970 to 1989; each country is a single observation in the figure. The quantity theory predicts that the observations w ill lie on a 45° line, that is, that changes in money stocks and price levels w ill be proportional. The 45° line in the diagram is adjusted to pass through the mean o f the data, but it has a slope o f posi tive one. It is not a regression line; no attempt has been made to fit the line to the data. The fact that the observations lie near the 45° line 9March Press Release, p. 12. 10See also the work on money and inflation in the P* model, such as Hallman, Porter and Small (1989). 1 For a discussion of the quantity theory and its variants, 1 see Laidler (1985). 12The countries are: Australia, Austria, Belgium, Brazil, Canada, Denmark, Finland, France, (West) Germany, Greece, Iceland, Ireland, Italy, Japan, Norway, The Netherlands, New Zealand, Portugal, Spain, Sweden, Switzerland, Turkey, United Kingdom, United States, Argentina, Chile, Columbia, Guyana, Suriname, Paraguay, Mexico, Peru, Bolivia, Venezuela, Ecuador and Uruguay. Some observations were missing: Brazil, 1985-1989; Col umbia, 1986 and 1989; Chile, 1985-1989; Guyana, Suriname, Paraguay, Peru, Bolivia and Ecuador, 1989. Average Annual M2 Growth provides some evidence o f the validity o f the quantity theory.1 3 Lucas (1987) was happy enough w ith this type o f evidence to pronounce the inflation problem "successfully solved in a scientific sense.”1 The 4 figure does seem to reflect what many econo mists and market participants have in mind w hen they think about the relationship betw een central bank actions and inflation. The theory appears to w ork surprisingly well, as the figure contains information derived from countries Where data are missing, averages are over available years. M2 is used because the FOMC sets target ranges for this aggregate and because information on this ag gregate is collected for all countries in the sample. 13One possible objection to this evidence is that the relation ships between monetary growth and inflation have chang ed in the 1980s. However, the plots of recalculated averages using only 1980s data tell, by and large, the same story as figure 1. See also Dwyer and Hafer (1988). Recalculating the averages using M1 also does not alter the general conclusion. 14Lucas (1987), p. 221. MAY/JUNE 1991 34 w ith ve ry different social and economic struc tures. Most im portantly fo r the purposes o f analyzing FOMC decisions, the figure provides a basis fo r the Committee’s concern about infla tion because it relates inflation to money grow th, and the Committee sets target ranges fo r certain m onetary aggregates. W hile lo w inflation countries tend to be low m oney grow th countries, the relationship is far from exact. For instance, Japan experienced a 5.5 percent rate o f annual inflation during the period w ith an average M2 grow th rate o f over 11 percent, w hile the U.S. experienced inflation o f 6.1 percent w ith M2 grow th averaging 8.3 percent per annum.1 W hile a fe w percentage 5 points on the inflation rate is substantial by U.S. standards, it is not a lot b y w orld standards. The good fit in the diagram is obtained b y ex amining countries w ith a broad range o f infla tion experiences, from near zero to m ore than 80 percent per year. The point is that the U.S. Federal Reserve, when compared with other cen tral banks worldwide, tends to be in the low m oney g row th —low inflation group. Another important consideration, emphasized by Lucas (1980), is that the FOMC’s influence on inflation appears to be the product o f a great many decisions over a ve ry long time frame. The evidence presented says nothing about the relationship betw een money grow th and infla tion on a quarter-by-quarter basis.1 Thus, even 6 when inflation control is taken very seriously, the FOMC may have considerable room to ma neuver on a meeting-by-meeting basis and still meet its stated long-run inflation objective. Sustaining Real Output Growth The nature o f the relationship betw een m onetary policy and real activity is controver sial and remains an unresolved issue in m acro economics.1 Nevertheless, FOMC meeting sum 7 maries indicate that Committee members believe an easier policy can mitigate declines in real 15Simple measurement error is one possible reason for such discrepancies. 16See Lucas (1980) for a method of recovering the close fit for U.S. quarterly data. 17For a recent survey, see Blanchard (1990). ,8December Press Release, p. 12. 19February Press Release, p. 12. 20February Press Release, p. 12. 21 October Press Release, p. 12. FEDERAL RESERVE BANK OF ST. LOUIS output, at least in the short term. In N ovem ber 1990, fo r instance, "the members agreed that a limited degree o f easing at this juncture w ould provide some insurance against a deep and p ro longed recession. . . .”1 Similarly, in December, 8 “Mem bers noted that m onetary policy had been eased . . . [and that] a limited fu rth er m ove w ould provide some added insurance in cush ioning the econom y against the possibility o f a deepening recession. . . .”1 At the same m eet 9 ing, referen ce is made to “The stimulus p ro vided by the recent easing. . . ,”2 In August 0 1990, Committee discussion noted that “a tight ening m ove . . . might stall an already weak economic expansion.”2 Th erefore, w hile due 1 note is taken o f the theoretical controversy, fo r the purposes o f this article, the real effects o f m onetary policy are simply taken as given. The R o le o f Forecasts in ShortRun P olicy Actions The FOMC’s stated short-run policy objective necessarily emphasizes the role o f forecasting. The Committee must make an assessment o f the likely direction o f the econom y in the near term if it wants to cushion changes in real output w hen warranted. In addition, the Committee also must assess the current and immediate past position o f the economy, since reliable data on real GNP are not available fo r several quarters. As w ill becom e clear in the next section, h ow ever, economists have a difficult time forecasting even a fe w quarters ahead. By proxying the in form ation on real GNP available to the Commit tee w ith the Blue Chip Consensus forecast, and by using only data available at the time o f the meeting, an appreciation o f the uncertainty the Committee faced in 1990 m eeting by meeting w ill be developed.2 2 The Blue Chip Consensus is not the only in dicator o f the perception o f economists regar ding real activity. The Board staff prepares a forecast especially fo r each meeting, and that projection is probably the most relevant one 22The Blue Chip Economic Indicators is a monthly survey of about 50 mostly private sector economic forecasting firms. The Blue Chip forecast for a variety of economic variables is the average forecast of these firms. 35 H o w Much Uncertainty Exists in Forecasts of Quarterly Real G1VP? Forecasts o f quarterly changes in real GNP tend not to be very accurate, as is w ell known. T o get some idea o f the uncertainty involved, consider the forecast errors based on the Blue Chip Consensus fo r the time period IV/1982 to IV/1989. Each month, Blue Chip publishes a consen sus forecast fo r the annualized rate o f real GNP grow th in the current quarter and each quarter ahead fo r at least one year. Although actual data are sometimes revised many years after they are first published (so-called bench mark revisions), changes in real GNP are often considered final w ithin about three months o f the end o f the quarter. Taking the published forecast at the midpoint o f the quarter (e.g., February 10 fo r the first quar ter) and subtracting the final figure gives the forecast erro r at various time horizons. The mean erro r and the mean absolute forecast error at these horizons are as follows (32 observations): Forecast H orizon Contemporaneous 1 quarter ahead 2 quarters ahead 3 quarters ahead 4 quarters ahead Mean Forecast E rro r The difficulty o f aggregate economic fo re casting w ill come as no surprise to many readers, and it is not hard to identify plausi ble reasons fo r the problem. Th ere may simply be considerable random variation in real activity. Similarly, it may be that real output reacts quite quickly to unpredictable economic and political events. In 1990, one such event stands out: the Iraqi invasion o f Kuwait and the subsequent large variations in the price o f oil. In addition, all forecasts are conditional on policy, and actual policy may d iffer from that built in at the time o f the forecast. Mean A bsolu te Forecast E rro r -.43 -.29 -.24 -.48 -.55 O f prim ary concern are the relatively large mean absolute deviations. They suggest that outcomes far from w hat is forecast often oc cur, as an average e rro r at any horizon is at least 1.74 percent. Also interesting is that, fo r this set o f forecasts, there is at least as much uncertainty surrounding the current quarter forecast as the four-quarter-ahead forecast. 2.03 2.04 2.00 1.74 1.76 fo r Committee decision-making.2 Unfortunately, 3 it is not declassified until five years after the meeting. Using the Blue Chip forecasts as a proxy fo r information available to the Commit tee is not much o f a concern fo r the analysis to follow , how ever, because in every case, the qualitative description o f the Board projection given in the Record o f Policy Actions is consis tent with the Blue Chip Consensus. Further more, in only one case did the Committee's assessment d iffer qualitatively from that o f the Board staff; that case (the October meeting) w ill becom e apparent. Finally, the forecasting d iffi culties discussed in detail below are not a mat ter o f decimal points but o f qualitative direction; the Blue Chip forecasts w ill serve to illustrate this point.2 4 23According to Meltzer (1990), p. 31, “ Fed forecasts of GNP are as accurate as the forecasts from other models . . .” See also Karamouzis and Lombra (1989) and Meltzer (1987). 24See the shaded insert, “ How Much Uncertainty Exists in Forecasts of Quarterly Real GNP?” for a description of the uncertainty surrounding these forecasts. MAY/JUNE 1991 36 Measuring the P olicy Stance A n y summary o f FOMC m onetary policy ac tions requires some measurement o f the policy stance at a point in time. One o f the problems o f m onetary policymaking is that various mea sures can yield conflicting signals, sometimes making it difficult to discern the thrust o f pol icy. Consideration in this paper w ill be given to four possible measures, or "indicators,” o f the m onetary policy stance: the language o f the di rective, the federal funds rate, the m onetary ag gregate M2, and total reserves.2 O f these, the 3 simplest and most straightforw ard measure, re lied on heavily in the follow ing chronology, is to examine the Record to find out the language o f the directive. Th e other indicators are based on a simple analysis that associates "easing” or "tightening” w ith movements in measured variables. Th e implementation o f m onetary policy typical ly occurs via intervention in the market fo r reserves, w hich are actively traded among banks. The interest rate in this market is the federal funds rate, and the total reserve supply is sub ject to control by the Federal Reserve. For a given downward-sloping demand, the Federal Reserve can increase (decrease) the federal funds rate b y decreasing (increasing) the supply o f total reserves. A common simple analysis relates the sum o f total reserves and currency (the m onetary base) to measures o f m oney such as M2 by a proportional factor greater than one known as the m oney m ultiplier.2 Generally 6 speaking, therefore, a decrease in the federal funds rate, an increase in M2 and an increase in total reserves can be indications o f easier m onetary policy, w hile movements in the op posite directions can be indications o f tighter policy. In practice things are not so clear be cause the demand fo r reserves (and also the 25This list is by no means exhaustive. There are many other indicators that receive attention from economists, including various monetary aggregates, reserve components, in terest rate spreads, commodity prices and so on. Reference to these alternative indicators is suppressed in this article in the interest of streamlining the discussion. 26See Papademos and Modigliani (1990) for a recent exposition. 27ln the Record these fluctuations in demand are sometimes referred to as short-run technical factors. 28The range for the federal funds rate was set primarily for consultative purposes; that is, if the actual rate fluctuated persistently outside the range during an intermeeting period, the Committee agreed to discuss the situation. The Committee did not set a range for the federal funds rate at FEDERAL RESERVE BANK OF ST. LOUIS demand fo r M2) may fluctuate over time, p e r haps swamping the effect o f a change in reserves on the federal funds rate or on M2.2 7 Nevertheless, because total reserves are sub ject to control by the Committee, they constitute a logical indicator o f policy. In addition, because the Committee set target ranges fo r both the federal funds rate and M2 in 1990, they are also logical indicators o f the policy stance.2 Gener 8 ally, how ever, one’s assessment o f the policy stance at a point in time can d iffe r depending on w hich indicator is used. As w ill be shown below, the indicators can give conflicting signals, differin g not only from the language o f the directive but also from each other. The data on indicators referen ced in the subsequent chronology are plotted in figures 2 through 6. The prim ary referen ce fo r the federal funds rate that w ill be used is the w eekly time series fo r 1990 presented in figure 2.2 The 9 1990 w eekly series fo r M2 is plotted in figure 3; the annualized w eekly grow th rates are plotted in figure 4.3 The interpretation o f M2 is typical 0 ly within the context o f the target cone, which is review ed by the Committee tw ice yearly and represents the FOMC’s long-term target. W ithin the target cone, how ever, is some leew ay to alter M2 grow th rates m eeting b y meeting. As can be seen from figure 4, grow th rates o f monetary aggregates tend to be fairly noisy. The time series fo r total reserves in 1990 is given in figu re 5. Unfortunately, these data also tend to be noisy; in addition, the Committee does not set a target grow th cone fo r total re serves. These facts sometimes combine to make interpretation difficult. Figure 6, how ever, plots the annualized interm eeting grow th rates o f total reserves, based on the nearest available data point (since total reserve data are biweekly). its November or December meetings, saying it no longer “ served [any] real purpose.” See the December Press Release, pp. 15-16. Mention of the target ranges for the federal funds rate and M2 is made only to show that the Committee gives these indicators some official status. Total reserves, on the other hand, does not have such a status. 29For an assessment of this interest rate as an indicator of monetary policy and a predictor of future real sector activi ty, see Bernanke and Blinder (1990). 30For discussions of monetary aggregates and their relation ship to real activity, see Christiano and Ljungqvist (1988) and Stock and Watson (1987). 37 Figure 2 The Weekly Federal Funds Rate in 1990 Percent 9.0 Percent 9.0 6.0 2/6 3/27 5/15 7/2 Vertical lines represent FOMC meeting dates 8/21 10/2 11/13 12/12 Figure 3 M2 in 1990 Trillions of dollars 3.36.------------------ Weekly Data in Levels Trillions of dollars 3.36 3.32 3.30 3.26 2/6 3/27 5/15 7/2 ------------------ 13.20 8/21 10/2 11/13 12/12 Source: post 1991 benchmark data Vertical lines represent FOMC meeting dates MAY/JUNE 1991 38 Figure 4 M2 Growth in 1990 Percent change Percent change from previous week Annualized Weekly Rates from previous week 25,---------------------------------------------------------------------- ----------------------------------------------------- 25 20 20 15 10 -5 -10 -1 5 -1 5 2/6 3/27 5/15 7/2 8/21 10/2 11/13 12/12 Source: post 1991 benchmark data Vertical lines represent FOMC meeting dates Figure 5 Total Reserves in 1990 Billions of dollars 63.0 Biweekly Data in Levels Billions of dollars 63.0 ■ ■ II II 1 1 II I I l l 62.5 62.0 I 60.5 \ * \ / \/ * \ 1 \a \ - y i I I I I I I I I i I I I i ,+/ \ a : ' j vV \/ \ \/ 59.0 I I I I I I I I ■ " i 61.5 I I I I I I 61.0 59.5 62.0 I I I I I I 61.5 60.0 62.5 v \/* ; v 61.0 60.5 60.0 59.5 \! ! i < » 59.0 58.5 58.5 1/10 3/7 FEDERAL RESERVE BANK OF ST. LOUIS 5/2 6/27 8/22 10/17 12/12 2/6 Source: post 1991 benchmark data 39 Figure 6 Intermeeting Growth of Total Reserves1 Percent change from previous meeting 30 Annualized Rates of Intermeeting Growth Percent change from previous meeting 30 25 I I I I I I I I I I 1 I I I I I I I I I I I f I I I 20 15 10 5 A A 0 V \ \ -5 -1 0 I I I -1 5 V -20 25 20 15 10 5 0 -5 -1 0 -15 -20 3/21 5/16 6/27 10/3 11/14 12/12 2/6 Source: post 1991 benchmark data Vertical lines represent data point nearest meeting date ’ Data are Board series, adjusted for reserve requirements 2/7 For all indicator data, vertical lines represent FOMC meeting dates. The fram ew ork that w ill be used in this paper to summarize FOMC decision-making is now complete. The Committee states its major objec tives in every directive, and they are to control inflation and to maintain sustained grow th in real output. International evidence suggests that low inflation rates can be achieved by maintain ing low rates o f m oney growth. The real output effects o f monetary policy are theoretically less certain, but summaries o f Committee delibera tions indicate that members believe tem porary easing can mitigate downturns in economic ac tivity. Pursuit o f this objective requires an assess ment o f the current and future time path o f real output, but knowing w hether the incoming data signal a change in direction fo r the econom y is complicated by lags in data releases and er rors in even the best economic forecasts. To summarize FOMC policy actions, some measure o f the m onetary policy stance is required. Since 8/22 various measures sometimes suggest differing interpretations o f the thrust o f m onetary policy, several indicators w ill be employed. CHRONOLOGY OF FOMC DECISION-MAKING IN 1990 Meeting o f February 6-7, 1990 The February m eeting was one o f tw o during 1990 w hen the Committee review ed its long term objectives fo r grow th in the m onetary ag gregates. Much o f the discussion focused on the grow th range fo r M2.3 A staff report suggested 1 that, given the current forecasts fo r nominal GNP, the rate o f grow th o f M2 in 1990 was like ly to be in the "upper end o f the tentative range” o f 3 to 7 percent set the previous July.3 In this 2 view, the Committee could retain "considerable leew ay” to make "faster progress against infla tion without impairing the forw a rd momentum 3 March Press Release, p. 11. 1 32March Press Release, p. 12. MAY/JUNE 1991 40 o f the econom y . . range.3 3 by retaining the tentative A fter contemplating the staff report and other pertinent information, the Committee agreed on a range o f 3 to 7 percent grow th fo r M2 during the year, as computed from the final quarter o f 1989 to the final quarter o f 1990. The range set fo r M3 was 2.5 to 6.5 percent, down from the 3.5-to-7.5 percent tentative range used in 1989.3 4 One interpretation o f these grow th rates is sug gested by the inform ation in figure 1. In par ticular, if maintained over a long period o f time, these grow th rates are consistent w ith low aver age rates o f inflation relative to a w orld stan dard. In this sense the Federal Reserve contin ued to maintain its posture fo r preferring low inflation relative to other central banks. In fact, the Committee hoped to “ signal a [continued] determination to m ove tow ard the objective of price stability."3 5 The Committee also discussed policy fo r the period until the next meeting. The outlook fo r real GNP at the time o f the February m eeting is given in figu re 7, which illustrates the beliefs o f private forecasters on February 10. In the figure, the boxes represent the most recently revised data available at the time fo r points in the past, plus the Blue Chip Consensus forecast at the time fo r points in the future. The crosses rep re sent the evolution o f real GNP based on revised data and the Blue Chip Consensus forecast for 1991 and 1992 available as o f April 1991.3 6 Considering the figure from the perspective at the time, real GNP grow th appeared to be near 3 percent in the third quarter o f 1989, but approached zero in the fourth quarter. The forecast called fo r increasing rates o f grow th throughout 1990. In retrospect, the second and third quarters o f 1989 w ere actually much w eaker than they appeared at the tim e.3 W hile 7 the prediction that the econom y w ould rebound slightly in the first quarter o f 1990 appears by present estimates to have been correct, the pri vate sector forecasts o f a generally strengthen ing econom y throughout 1990 turned out to be erroneous. The Board staff projections w ere qualitatively consistent w ith the private forecasts at the time o f the February meeting, as they predicted that “the econom y was likely to expand relatively slowly over the next several quarters.”3 The 8 FOMC membership generally concurred w ith this view, seeing "continuing grow th in economic activity [as] a reasonable expectation fo r the year ahead” and "some assurance that the ex pansion was no longer weakening and indeed that a m odest acceleration might be under way. . . .”3 9 Am ong the plethora o f other information con sidered by the FOMC in February, the Record indicates that, internationally, Japan was experi encing strong grow th in real GNP and that, while Germany, Italy and France appeared to be gain ing strength, the United Kingdom and Canada rem ained sluggish. Th e trade-weighted value o f the dollar in terms o f foreign currencies had recently fallen, and most o f the depreciation was against the German mark. U.S. civilian un em ployment was unchanged at 5.3 percent.4 0 At the conclusion o f the meeting, the FOMC issued a policy directive to "maintain the exist ing degree o f pressure on reserve positions.”4 1 The policy fo r possible adjustments during the interm eeting period was to be symmetric, with no bias tow ard tightening or easing.4 2 Meeting o f March 27, 1990 Considering figu re 2, policy was indeed steady in the six weeks follow ing the February meeting, as the federal funds rate remained unchanged at about 8.25 percent through late March. W hile figure 3 shows that M2 was slightly above the upper end o f the target cone during 33March Press Release, p. 12. 38March Press Release, p. 6. 34According to the Record, the change in the M3 range was viewed as consistent with an unchanged M2 range for “ technical reasons." See the March Press Release, p. 13. 39March Press Release, p. 7. The Committee also discussed the risk of a downturn. 35March Press Release, p. 12. 4 March Press Release, p. 21. 1 36Blue Chip forecasts are released on the 10th of each month. 42March Press Release, p. 18. The Committee sometimes issues asymmetric directives, which augment the basic directive by stating a direction of bias. In some cases, the Committee ties the direction of bias to data or other infor mation forthcoming during the intermeeting period. 37The problem of data revision is acute, and an important consideration to keep in mind is that the view of the data today is itself subject to revision in the future. See also Mankiw and Shapiro (1986). FEDERAL RESERVE BANK OF ST. LOUIS 40March Press Release, pp. 1-4. 41 Figure 7 Private Forecasters’ View of Real Output, February 1990 Percent change from previous quarter 5 Percent change from previous quarter 5 90 91 1992 Source: Blue Chip Forecasts and Data Revisions February and into March, figure 4 indicates that most o f the w eekly grow th rates w ere within a range consistent w ith a 3 to 7 percent grow th rate fo r the year. It does not appear, therefore, that there was any change in policy according to an M2 measure o f the policy stance during the period immediately follow in g the February meeting. Finally, figure 6 also shows little indi cation o f a change in the thrust o f policy, as reserves continued to grow . All considered, pol icy appeared to be steady as the Committee con vened in late March. Figure 7 indicates that the fou r quarters begin ning April 1990, only a w eek after the March meeting was held, appear from today’s perspec tive to be one o f the weakest sequences o f quarters since 1982. Nevertheless, at the time the FOMC met, there was no indication, accord ing to the Blue Chip Consensus, that the na tional econom y w ould be entering a recession later in the year. The private forecasters’ outlook fo r real GNP grow th changed little betw een the February and March meetings. The Board staff projection was qualitatively consistent w ith the Blue Chip forecast, suggest ing “the econom y was likely to expand at a somewhat faster pace over the next several quarters than in the fourth quarter o f 1989.”4 3 Growth in that quarter was reported to have been less than 1 percent at an annual rate. The Committee concurred, as “ on balance . . . the members view ed sustained grow th in business activity as a reasonable expectation fo r the next several quarters.”4 In addition, the Committee 4 “ expressed a great deal o f concern about the ap parent lack o f im provem ent in underlying infla tion trends.”4 Considering the forecast fo r real 5 output, in addition to other pertinent inform a tion, the majority o f the Committee voted to maintain the "current degree o f pressure on reserve positions.”4 No direction o f interm eeting 6 bias was specified. 43May Press Release, p. 6. 45May Press Release, p. 7. 44May Press Release, p. 7. 46May Press Release, p. 13. MAY/JUNE 1991 42 The Board staff w arned at the M arch meeting that M2 grow th might be slow or non-existent over the spring and early summer, partly fo r special technical reasons and partly because o f the general slowing in the rate o f nominal GNP growth. A number o f Committee members, com menting on the Board staff report, felt a slow ing in the rate o f M2 grow th "w ou ld be a w el come developm ent,” since it w ould put M2 grow th m ore squarely within the Committee's target range.4 7 Meeting o f M ay 15, 1990 Figure 2 suggests that policy was steady dur ing the period immediately follow ing the March FOMC meeting, according to a federal funds rate measure o f the policy stance. As the Board staff report predicted, how ever, the m onetary aggregate measure tells a different story: M2 grow th began to slow in March, m oving tow ard the midpoint o f the Committee’s target cone by July. Figure 4 shows that annualized w eekly M2 grow th rates w e re mostly at or below the 3 percent m ark in the w eeks betw een the March and May meetings. O f course, the March staff report had predicted a slowing in M2 growth, and in addition, the data on m onetary aggregates simply tend to be noisy. Figure 6, how ever, shows that reserve grow th was negative betw een the M arch and M ay meetings, which might be construed as a relatively tight policy immediate ly follow ing the M arch meeting. Th erefore, as the Committee convened in M ay it was not clear by some measures that policy had been con stant during the interm eeting period. By one measure, policy remained steady; by another, policy tightened beginning at about the time o f the March meeting. The outlook fo r real GNP at the time o f the May meeting, as summarized b y the Blue Chip Consensus forecast, was again virtually un changed from February 10. Generally speaking, the view at the time was much m ore optimistic than the view from the present. Private fo re casters at the time view ed real GNP grow th as being faster fo r virtually every quarter in 1989, 1990 and 1991 relative to the view today. The Blue Chip Consensus indicated a virtually flat grow th rate o f 2 percent a quarter through 1990, increasing slightly in 1991. The Board staff projection concurred with private forecasts, suggesting "that the econom y was likely to expand at a m oderate pace over the balance o f the year.”4 In addition, the Com 8 mittee "generally agreed that the current in for mation on business conditions pointed on balance to relatively m oderate but sustained economic expansion.”4 Considering the outlook fo r real 9 GNP as w ell as other economic information, a large majority o f the Committee supported a directive that called fo r unchanged policy with no bias tow ard tightness or ease.5 0 The Board staff explained in an analysis prepared fo r the Committee that M2 grow th was expected to pick up somewhat b efore the next meeting, even under a policy o f "steady reserve pressure.”5 Several m em bers com 1 mented that “ a failure o f such grow th to pick up w ould be a matter o f increasing concern” and might be taken as a reflection, among other things, o f “grow in g constraints on the availabili ty o f credit to potential borrow ers.”5 Generally, 2 how ever, the Committee felt it was too early to reach a definitive conclusion since the observed m oderation might m erely be a manifestation o f the natural volatility in m onetary grow th rates.5 3 Meeting o f July 2-3, 1990 The July meeting was the second o f tw o dur ing the year in which the Committee review ed its long-term goals, including an assessment o f the target cone set at the February meeting fo r M2 growth. According to the Record, “the Com mittee took account o f the much slower than anticipated expansion o f M2 . . . in the first half o f the year. . . 5 Some m em bers noted that any 4 "shortfall from the current ranges should be kept under careful scrutiny to judge w hether 47May Press Release, p. 12. 52July Press Release, p. 11. 48July Press Release, p. 6. 53July Press Release, p. 11. 49July Press Release, p. 7. 54August Press Release, p. 11. s°July Press Release, pp. 10-11. 5 July Press Release, p. 11. This is possible because, while 1 a component of M2 is related to the monetary base by the money multiplier, M2 is a broad aggregate with many other components over which the Federal Reserve has lit tle direct influence. FEDERAL RESERVE BANK OF ST. LOUIS 43 Figure 8 Private Forecasters’ View of Real Output, July 1990 Percent change from previous quarter Percent change from previous quarter Source: Blue Chip Forecasts and Data Revisions policy was indeed tighter than intended or de sired.”5 A fter this review , the Committee re 5 affirm ed its range fo r M2 grow th at 3 to 7 per cent fo r the rem ainder o f 1990.5 One argument 6 that played a role in this decision was that the Committee should avoid making adjustments to target cones, at least at mid-year. Some members suggested, fo r instance, that “the ranges should not be m oved up or dow n to fit special cir cumstances. . . ,5 7 According to the evidence from a num ber o f countries presented in figure 1, the decision to keep the M2 target range at betw een 3 and 7 percent continued to place the U.S. squarely in a group o f relatively low m oney grow th coun tries. This is consistent w ith the Committee’s low inflation strategy because these are the countries that have tended to experience the lowest average inflation rates over the last 20 years. In this sense, the Committee continued to maintain its anti-inflation posture at this meeting. O f course, in this context, “lo w ” is relative to w orld standards, and the relationship betw een m oney grow th and inflation is not exact even over ve ry long time horizons. The Committee also contemplated the policy stance fo r the period until the next meeting. During the interm eeting period, the federal funds rate, M2 and total reserves measures all seemed to indicate a steady policy. The Blue Chip Consensus forecast and the available data fo r real GNP appeared as the path represented by the boxes in figu re 8. A recession was not predicted b y private forecasters at the time, and there was simply no w ay o f knowing that Iraq w ould invade Kuwait during the forthcom ing in term eeting period. Available actual data as w ell as private forecasts continued to be almost uni form ly optimistic relative to the actual outcomes fo r quarterly real GNP grow th as view ed from the present. 55August Press Release, p. 13. 56August Press Release, p. 14. 57August Press Release, p. 14. MAY/JUNE 1991 44 The Board staff forecast fo r the rem ainder o f 1990 was again in general agreem ent w ith the private forecasts, suggesting "the econom y w ould expand . . . at around the rate estimated fo r the first half o f the year.”5 Committee members 8 concurred w ith the Board staff and the private forecasters as they "generally saw sustained but subdued grow th in economic activity as a rea sonable expectation fo r the next several quar ters . . . [T]he econom y as a w hole gave no cur rent indications o f slipping into recession.”5 9 The forecast o f slow but positive grow th was buoyed, according to the Record, by a number o f other factors that the Committee considered in addition to point predictions o f real GNP growth. Unemployment, fo r example, remained at 5.3 percent in M ay and had been at that level fo r m ore than a year. Industrial production was up substantially in May, and economic grow th seemed to be satisfactory in the major indus trialized nations.6 0 Based on the forecasts fo r real GNP and the consideration o f the most recent data on the state o f the economy, the FOMC unanimously endorsed an unchanged policy fo r the seven w eek period until the next meeting. The majori ty o f the Committee also favored a bias tow ard "some slight easing” depending on the interm eet ing data on M2 grow th and inflation. In particu lar, the m ajority wanted to ease slightly unless M2 grow th picked up appreciably or inflation began accelerating faster than expected.6 A c 1 cording to the Record, “the marked slowing in monetary grow th in the second quarter in par ticular suggested the possibility o f more restraint than the Committee intended.”6 2 Figure 9 shows the M2 data available at the time o f each FOMC meeting. The crosses rep re sent the revised data available today, w hile the boxes show the time path as it appeared at the time. A t the July meeting, the Committee saw data suggesting a decline in M2 from the level o f the previous meeting. The revised data avail able today show no such decline, how ever, and indeed generally indicate an increase over the previous 13 weeks. Data revisions can therefore explain, to some extent, the discussion in the 58August Press Release, p. 6. 59August Press Release, p. 7. 60August Press Release, pp. 1-4. 61August Press Release, p. 18. FEDERAL RESERVE BANK OF ST. LOUIS Record o f the "m arked slowing in m onetary g row th ” w hen it appears from figures 3 and 4 that money grow th was picking up in the w eeks before the July meeting. The data revision prob lem fo r M2 does not seem to have been as acute fo r other periods during 1990. Meeting o f August 21, 1990 As it turned out, m oney grow th did not pick up in the w eeks immediately follow in g the July meeting, and in mid-July "pressures on reserve positions w e re eased slightly.”6 Measuring pol 3 icy by the federal funds rate indicates that, ac cording to figure 2, policy did ease slightly on or about July 13, w ith the rate declining to just over 8 percent b y early August. The effective w eekly federal funds rate later rose, how ever, and did not fall below the early August level until mid-October. According to figures 3 and 4, the grow th path o f M2 also seemed to indicate some ease during the interm eeting period. Th e annualized w eekly g row th rates, which are near zero or negative in the month immediately follow in g the July meeting, are greater than 7 percent in the last three weeks leading up to the August meeting. O f course, these data are noisy and interpreta tion is difficult. Total reserves reached a low fo r the year on July 25, reflecting a slight decline overall during the interm eeting period. The invasion o f Kuwait on August 2, 1990, clouded considerably the outlook fo r the U.S. econom y fo r the rem ainder o f the year and through the first half o f 1991. The key economic question was the magnitude and stay ing p ow er o f the resulting crude oil price in creases. According to the Record, at the August meeting the Committee " . . . focused on both the state o f the econom y b e fo re the increase in oil prices and the likely consequences fo r real output and inflation o f that rise.”6 4 A comparison o f the available data and the associated Blue Chip Consensus forecasts fo r Ju ly, August and September, illustrated in figures 8, 10 and 11, respectively, demonstrate the fluidity o f the forecasting situation during this two-month period (July 10-September 10). First 62August Press Release, pp. 18-19. 63October Press Release, p. 4. 64October Press Release, p. 7. 45 Figure 9 M2 Data Available at FOMC Meetings V ertical lin es rep rese nt FOMC m eeting dates o f all, in July, the data fo r all o f 1989 and the first quarter o f 1990 w ere revised downward, revealing greater sluggishness in real GNP g row th than had previously been estimated. Secondly, the situation in the Middle East was evolving on a daily basis, and the eventual out come simply could not be predicted. As figure 11 shows, by September 10, about three weeks after the August FOMC meeting, the Blue Chip survey put the consensus forecast fo r real GNP grow th at zero fo r the fourth quarter o f 1990; this reflects a dow nw ard revision fo r fourth quarter output grow th from over 2 percent as the year began. Once again, the Board staff projection reflec -O c to b e r Press Release, p. 6. 66October Press Release, p. 6. ted that o f the private forecasters. W hile the staff "recognized that the recent steep rise in oil prices could have important adverse effects . . . [It also recognized that] it was not possible to determine w ith any confidence how oil prices might evolve over time. . . ,"6 Nevertheless, it 5 seemed to the staff that slow expansion o f real output was likely over the balance o f the year, albeit at a reduced rate from that previously projected.6 The staff forecast was based, in 6 part, on substantial grow th in exports in the quarters that lay ahead, because foreign indus trial economies w ere view ed as relatively strong and a considerable depreciation had already oc curred in the foreign exchange value o f the dollar.6 7 670 c tober Press Release, pp. 6-7. MAY/JUNE 1991 46 Figure 10 Private Forecasters’ View of Real Output, August 1990 Percent change from previous quarter 5 Percent change from previous quarter 5 -3 1988 90 91 1992 Source: Blue Chip Forecasts and Data Revisions Figure 11 Private Forecasters’ View of Real Output, September 1990 Percent change from previous quarter FEDERAL RESERVE BANK OF ST. LOUIS Percent change from previous quarter Source: Blue Chip Forecasts and Data Revisions 47 FOMC members again concurred with the staff and private forecasters that “limited grow th in economic activity rem ained a reasonable ex pectation. 1 8 The Committee recognized that ,6 most o f the available data on the econom y per tained to a period b efore the Iraqi invasion. The Record notes that unemployment rose from 5.3 percent to 5.5 percent in July and that domestic industrial production was flat. Overseas, real output grow th in both Japan and Germany re mained robust. In addition, the trade-weighted value o f the dollar had fallen substantially dur ing the interm eeting period in terms o f other G-10 currencies.6 9 In the Committee’s view, the oil price shock w ould tend to "w eaken economic activity while also intensifying inflationary pressures.”7 Com 0 mittee members tended to see changes in policy as counterproductive, an easing probably fu el ing inflation, a tightening probably stalling a weak econom y.7 Accordingly, the Committee 1 elected to maintain the current policy stance, “ fostering a stable policy environm ent.” But sev eral members stated that they wanted "to avoid any paralysis o f policy . . . in the weeks ahead.” Some saw a likely need to ease "at some point,” w eighed against continuing decline in the dollar in markets fo r foreign exchange.7 Therefore, 2 while there w ere "some differences in view s,” the majority o f the Committee membership sup ported a bias tow ard ease in the interm eeting period.7 3 Meeting o f O ctober 2, 1990 According to the Record, the bias in the direc tive was not acted on in the interm eeting period because inflation was “ not abating and the econ om y [was] continuing to advance, albeit slow ly. . . .” 7 Measures o f the policy stance in the 4 weeks immediately follow ing the August meeting, in general, seemed to indicate a steady policy without any bias tow ard ease. As figure 2 in dicates, federal funds traded at 8-8.25 percent over the period, which represented no change from the previous interm eeting period. Total “ October Press Release, p. 8. “ October Press Release, pp. 1-3. 70October Press Release, p. 11. 7 October Press Release, pp. 11-12. 1 reserves g re w somewhat during the interm eet ing period. M oney by an M2 measure displayed, according to figure 4, considerable volatility in annualized w eekly grow th rates in the weeks follow ing the August meeting, to the point w h ere an assessment o f the interm eeting policy stance by this measure is quite difficult. The private forecasts fo r July, August and September indicate rapidly deteriorating expec tations fo r real output. Still, no recession was forecast at the time o f the August m eeting—or even three w eeks later on September 10—and the private forecasters appeared to view the slow grow th as tem porary, predicting annual ized gains in real GNP o f m ore than 2 percent by the third and fourth quarters o f 1991. The October meeting o f the FOMC occurred eight days before the Blue Chip Consensus forecast il lustrated in figu re 12 was officially released. October, the first month o f the fourth quarter, was the first time this set o f forecasters envi sioned negative grow th on the horizon. By Octo ber 10, the Blue Chip Consensus forecast was actually tw o consecutive quarters o f negative grow th in real GNP—but just barely. A recession was not definitively predicted by Blue Chip until Novem ber.7 5 Figure 13 illustrates the dramatic change in the outlook fo r real GNP as forecast b y the Blue Chip Consensus from July 1990 to October 1990. In the space o f only three months, the forecast changed from one o f sluggish but increasing rates o f real grow th to near zero and even neg ative grow th rates. In terms o f time fo r policy reaction, this change was quite fast. If one ac cepts research evidence that m onetary actions affect real activity only with a lag o f several quarters, this rapid deterioration in the ex pected perform ance o f real output underscores the difficulty o f making timely short-run adjust ments in the stance o f policy. The Board staff, acknowledging a great deal o f uncertainty linked to developments in the Middle East, projected “ a mild downturn in eco nomic activity . . . fo r the near term.’’7 The staff 6 75See the Blue Chip Economic Indicators, October 10 and November 10, 1990. The October consensus forecast call ed for two consecutive quarters of negative real GNP growth, but the second of these quarters was nearly flat. 76November Press Release, p. 6. 72October Press Release, p. 12. 73October Press Release, pp. 13-14 and p. 16. 74November Press Release, p. 4. MAY/JUNE 1991 48 Figure 12 Private Forecasters’ View of Real Output, October 1990 Percent change from previous quarter Percent change from previous quarter Source: Blue Chip Forecasts and Data Revisions Figure 13 Private Forecasters’ Changing Perceptions Percent change in data and forecasts from July to October 0.4 Extent of Downward Revision, 7/90-10/90 Percent change in data and forecasts from July to October 0.4 -0.4 - 0.8 - 1.2 - - 1.6 2.0 -2 .4 - 1988 FEDERAL RESERVE BANK OF ST. LOUIS 89 90 1991 Source: Blue Chip Forecasts and Data Revisions 2.8 49 continued to see strong exports as a mitigating factor, propelled by the projection o f continuing grow th in several other major industrialized na tions, especially Germany and Japan. The staff forecast also relied to some extent on a drop in oil prices during the first half o f 1991.7 7 For the first time in 1990, the FOMC m em ber ship dissented qualitatively from both private forecasters and the Board staff in their view of the likely future path o f real output. The Record reports a sense o f the m eeting concluding that, while the Committee felt the risk o f negative real output grow th had increased, “overall eco nomic activity appeared to be continuing to ex pand, although at a slow pace. . . . [T]he avail able data did not point to cumulating weakness and the onset o f a recession.”7 Many members 8 w ere concerned that surveys o f business confi dence seemed to indicate a declining faith in a continued expansion, w hile traditional indicators continued to suggest sluggish grow th .7 Some 9 members suggested that inflation was getting w orse even after accounting fo r the effect o f higher crude oil prices.8 According to the 0 Record, "A major concern was that the rise in oil prices w ould becom e . . . firm ly entrenched in the cost structure o f the econom y . . . and [delay] progress tow ard price stability.”8 N ever 1 theless, most members felt that "an easing move was warranted in light o f the indications that there was a significant risk o f a much w eaker econom y.”8 2 The Committee also expressed concern about an easing in response to the impending budget deal being crafted by Congress and the W hite House. The timing o f any m ove needed to be considered carefully, as action before any bud get accord might create the expectation o f more action after the deal was struck.8 In the discus 3 sion, some members suggested that "associating an easing m ove too closely w ith a fiscal policy action might set an undesirable precedent in terms o f producing expectations o f similar mon etary policy adjustments in the future.”8 The 4 advocates o f easing on the Committee agreed that the "reasons fo r the easing w ere not keyed to the enactment o f the new federal budget alone but m ore broadly to developments in credit markets and the economy. . . .”8 The 5 crux, according to the Record, was simply that "market participants expected a m onetary policy response to the fiscal policy actions. . . .”8 6 The Committee issued an unusual directive in response to the concerns about declining out put, accelerating inflation and fiscal policy. The directive called fo r no change in the degree o f pressure on reserve positions initially, but as sumed “ some slight easing w ould be implement ed in the interm eeting period, assuming passage o f a federal budget resolution . . . and the ab sence o f major unexpected . . . developments.”8 7 Thus, the directive was biased tow ard ease. However, the Record indicates that an addi tional proviso was added; in particular, the Committee agreed to ease further if real output showed further signs o f deterioration.8 The 8 Record also notes that “ some slight firm ing” was not ruled out, should inflation appear to pick up.8 9 Meeting o f N o v e m b e r 13, 1990 During the interm eeting period, policy was in itially unchanged. Th e contingencies in the directive w e re exercised late in October, when “ pressures on reserve conditions w ere eased slightly.” The Record cites both the “background o f a weakening econom y” and "the conclusion o f a budget agreem ent” as factors influencing the decision and the timing o f the easing.9 0 Other indicators o f the thrust o f policy, h o w ever, do not provide evidence o f an easing dur ing the w eeks leading up to the Novem ber m eet ing. The w eekly average federal funds rate, plotted in figure 2, had drifted up to a level near 8.25 percent by the time o f the October meeting. As the figu re shows, federal funds had been trading near 8.25 percent fo r most o f the year, except fo r the period immediately fo llo w ing the mid-July easing. The rate had fallen to a 77November Press Release, pp. 6-7. 85November Press Release, p. 13. 78November Press Release, p. 7. 86November Press Release, p. 14. 79November Press Release, p. 10. 87November Press Release, p. 15. 80November Press Release, pp. 11-12. 88November Press Release, p. 15. 8 November Press Release, p. 12. 1 89November Press Release, p. 15. 82November Press Release, p. 12. 90December Press Release, p. 5. 83November Press Release, pp. 13-14. 84November Press Release, p. 14. MAY/JUNE 1991 50 point just below its early August level during the interm eeting period. Based on a cursory look at the level o f the federal funds rate, therefore, the policy stance seemed to be about the same as it was after the mid-July easing.9 1 Total reserves fell substantially betw een the Oc tober and N ovem ber meetings, suggesting a tight policy instead o f an easy one. Finally, beginning at about the time o f the October meeting, M2 grow th nearly slowed to a stand still, reaching a level it w ould not again attain until the final w eeks o f 1990. Figure 4 indicates that most o f the annualized w eekly grow th rates fo r the rem ainder o f the year w ere below 3 percent, and many w ere negative. By an M2 indicator, then, policy tightened considerably in the fourth quarter.9 2 Private forecasters, as surveyed by Blue Chip, reached a consensus view that the econom y was entering a recession in Novem ber, accor ding to the forecast illustrated in figure 14. Re lative to current projections, how ever, the fo re cast rem ained optimistic about the depth o f the downturn. The Board staff also projected a mild recession w ith recovery occurring in the first half o f 1991. The staff assumed a drop in crude oil prices early in 1991 and export grow th driven by the expansion in foreign industrial ized nations. Th e staff forecast also emphasized the uncertain environment prodded by the mili tary standoff betw een the U.S. and Iraq on the Kuwaiti b ord er.9 3 The FOMC membership saw a relatively mild recession ahead, thus establishing general quali tative agreem ent w ith private forecasters and the Board staff. They also view ed a slow expan sion in 1991 as a reasonable expectation.9 A c 4 cordingly, the Committee agreed to some slight easing immediately and to some bias tow ard further easing during the interm eeting period. W hether the option to ease further was exer cised depended in part on "market reactions to the initial action. . . ,”9 5 The Novem ber directive o f the FOMC is the first to indicate a substantial commitment to ease. At the time o f the Novem ber meeting, the econom y was in the middle o f w hat appeared to be the onset o f recession. No amount o f easing was likely to change fourth-quarter real output —industrial production was already in the midst o f dropping 19.8 percent on an annualized basis in Novem ber. Instead, according to the Record, the Committee view ed the easing as providing “ some insurance against a deep and prolonged recession. . . ,”9 6 Meeting o f D ecem b er 12-13, 1990 As the Committee convened in mid-December, the indicators o f policy w ere again sending con flicting signals. In the period betw een the No vem ber and D ecem ber meetings, the federal funds rate dropped substantially, suggesting dramatic easing relative to earlier actions (see figure 2). As reflected in figure 6, how ever, the data on total reserves pointed instead to a fu r ther tightening o f policy, as the previous nega tive interm eeting grow th rate is follow ed by a steeper decline in reserves after the Novem ber meeting. The annualized w eek ly grow th rates of M2 plotted in figure 4 also do not o ffe r evi dence o f substantial ease during this period. M oney grow th simply continued at a near zero pace, on average, through to the Decem ber meeting. Figure 15 reflects the outlook fo r real GNP at the time o f the FOMC's Decem ber meeting. The assessment o f the private forecasters in the Blue Chip survey continued to g row m ore pessimistic by the month. In fact, one o f the striking fea tures o f the evolution o f Blue Chip forecasts in 1990 is that they fairly consistently overpredicted real output growth. The Board staff continued to project a mild recession w ith a rebound before mid-year 1991.9 7 Committee members concurred that a short downturn follow ed b y modest recovery seemed reasonable, but they emphasized the risks o f a prolonged downturn. Some Committee members also recognized, how ever, the possibility o f pro- 9 According to the Record, the reason federal funds traded 1 at 8.25 percent early in the intermeeting period was “ more cautious reserve management policies at some banks and some carryover of end-of-quarter pressures. . . .” See the December Press Release, p. 5. 93December Press Release, pp. 6-7. 92By the November meeting, “ the recent weakness in monetary growth was becoming a matter of increasing concern and was an important consideration for some members in their support of some easing of reserve condi tions.” See the December Press Release, p. 12. 97February Press Release, p. 6. FEDERAL RESERVE BANK OF ST. LOUIS 94December Press Release, p. 8. 95December Press Release, p. 13. 96December Press Release, p. 12. 51 Figure 14 Private Forecasters’ View of Real Output, November 1990 Percent change from previous quarter 5 1988 Percent change from previous quarter 90 91 1992 Source: Blue Chip Forecasts and Data Revisions Figure 15 Private Forecasters’ View of Real Output, December 1990 Percent change from previous quarter Percent change from previous quarter Source: Blue Chip Forecasts and Data Revisions MAY/JUNE 1991 52 cyclical policy, noting that because o f "the lags involved, there was some risk o f overdoing the easing o f policy at some point. . . .”9 N everthe 8 less, members unanimously supported additional easing in the directive in order to "provide some added insurance in cushioning the econ omy against the possibility o f a deepening reces sion and an inadequate rebound.”9 9 In the w eeks follow ing the Decem ber meeting, policy indicators suggested that further easing occurred, if one considers a federal funds rate measure o f the policy stance. In particular, the rate dropped to 6.25 percent by the time o f the February 1991 meeting. In addition, the inter meeting grow th rate o f total reserves plotted in figure 6 seems also to indicate dramatic easing, as reserves increased nearly 25 percent on an annualized basis from the Decem ber meeting. Th ere was little evidence o f M2 grow th in the w eeks immediately follow ing the Decem ber meeting, but then the aggregate began to show substantial grow th beginning in late January 1991.1 0 By an M2 measure, policy remained 0 tight through the first weeks o f 1991 before in dicating signs o f ease. All measures seemed to indicate ease by the first w eek o f February 1991. SUMMARY This article has presented a case study o f FOMC actions during a year in w hich a reces sion began. The Committee states its goals in each directive, and they are to provide fo r stable prices and to prom ote sustained real out put growth. The article has emphasized, within the context o f a chronology o f 1990 FOMC pol icymaking, some problems o f implementing a policy to meet these stated objectives. Evidence is presented early in the article regarding the relation betw een money grow th rates and inflation rates across countries and 98February Press Release, p. 12. "F eb rua ry Press Release, p. 12. 1C 0Committee members showed concern for the flat growth in M2 at the December meeting. The Record suggests that while “ the behavior of M2 was not fully understood,” FEDERAL RESERVE BANK OF ST. LOUIS time. An argument is made that over very long time periods, average inflation tends to reflect average rates o f m oney growth. I f this evidence is used as a guide, the Federal Reserve in 1990 recorded stellar success in maintaining grow th rates o f m oney stocks that are much low er than those achieved by many other central banks. Coupled w ith other similar decisions over a long period, this w ill continue to place the U.S. in a small group o f countries that w ill likely con tinue to experience, relative to other countries in the w orld, very low inflation rates. This arti cle has provided, therefore, one interpretation o f the FOMC’s "long-run, anti-inflation strategy” sometimes cited in the Record.1 1 0 The FOMC’s ability to achieve its real output objective is hampered, how ever, by the d iffi culty o f forecasting real output changes far enough ahead to take corrective action. The ar ticle assessed the information available on the projected evolution o f real output at the m eet ings by presenting the unrevised data available at the time along w ith the most current Blue Chip Consensus forecast. According to the Record, the Committee’s views rarely differed substantially, at least qualitatively, from the private sector forecast. This, along with evi dence in the Record, indicates that a negative quarter o f real output grow th was not antici pated until October, and a recession forecast did not come until Novem ber, already w ell into the first quarter o f the downturn. Another feature o f short-run policymaking is that it requires some measurement o f the m one tary policy stance. This article has emphasized h ow different indicators on some occasions im ply different assessments o f the thrust o f m one tary policy. In particular, the total reserves and M2 indicators suggested that policy was tight in the fourth quarter, w hile an interest rate indi cator suggested the opposite. By the first quar ter o f 1991, how ever, all measures suggested that the policy stance had shifted tow ard ease in response to the onset o f recession. it might be due to caps on credit availability as well as the weak growth of the economy. See the February Press Release, p. 13. 10'August Press Release, p. 13. 53 REFERENCES Bernanke, Ben, and Alan Blinder. “ The Federal Funds Rate and the Channels of Monetary Transmission,” NBER Work ing Paper No. 3487 (October 1990). Lucas, Robert E., Jr. “ Two Illustrations of the Quantity Theory of Money,” American Economic Review (December 1980), pp. 1005-14. Blanchard, Olivier Jean. “ Why Does Money Affect Output? A Survey,” in Benjamin M. Friedman and Frank H. Hahn, eds., Handbook of Monetary Economics, Volume II (NorthHolland, 1990), pp. 780-835. ________“Adaptive Behavior and Economic Theory,” in Robin M. Hogarth and Melvin W. Reder, eds., Rational Choice: The Contrast Between Economics and Psychology, (University of Chicago Press, 1987), pp. 217-42. Blue Chip Economic Indicators, 1990 Issues. Mankiw, N. Gregory, and M. G. Shapiro. “ News or Noise? An Analysis of GNP Revisions,” Survey of Current Business (May 1986), pp. 20-25. Christiano, Lawrence J., and Lars J. Ljungqvist. “ Money Does Granger-Cause Output in the Bivariate Money-Output Relation,” Journal of Monetary Economics (1988), pp. 217-35. Dwyer, Gerald P., Jr. and R. W. Hafer. “ Is Money Irrelevant?” this Review, (May/June 1988), pp. 3-17. Federal Reserve Press Releases, Record of Policy Actions of the Federal Open Market Committee, dated March 30, 1990; May 18, 1990; July 6, 1990; August 24, 1990; October 5, 1990; November 16, 1990; December 21, 1990; and February 8, 1991. Hallman, Jeffrey J., Richard D. Porter, and David H. Small. “ M2 Per Unit of Potential GNP as an Anchor for the Price Level,” Staff Study 157 (Board of Governors of the Federal Reserve System, April 1989). Karamouzis, Nicholas, and Raymond Lombra.“ Federal Reserve Policymaking: An Overview and Analysis of the Policy Process,” Carnegie-Rochester Conference Series on Public Policy (Spring 1989), pp. 7-62. Laidler, David E. W. The Demand for Money: Theories, Evidence, and Problems (Harper and Row, 1985). Meltzer, Allan H. “ Limits of Short-Run Stabilization Policy,” Economic Inquiry (January 1987), pp. 1-14. _______ .“ The Fed at Seventy-Five,” in Michael T. Belongia, ed., Monetary Policy on the 75th Anniversary of the Federal Reserve System (Kluwer Academic Publishers, 1991), pp. 3-104. 1991 Monetary Policy Objectives: A Summary Report of the Federal Reserve Board. Papademos, Lucas, and Franco Modigliani. “ The Supply of Money and the Control of Nominal Income,” in Benjamin M. Friedman and Frank H. Hahn, eds., Handbook of Monetary Economics, Volume I, (North-Holland, 1990), pp. 399-494. Stock, James H., and Mark W. Watson. “ Interpreting the Evidence on Money-lncome Causality,” NBER Working Paper No. 2228 (April 1987). Vogel, Robert C. “ The Dynamics of Inflation in Latin America, 1950-1969,” American Economic Review (March 1974), pp. 102-14. MAY/JUNE 1991 54 Allan H. Meltzer Allan H. Meltzer is a professor of political economy and public policy at Carnegie Mellon University and is a visiting scholar at the American Enterprise Institute. This paper, the fifth annual Homer Jones Memorial Lecture, was delivered at Washington University in St. Louis on April 8, 1991. Jeffrey Liang provided assistance in preparing this paper. The views expressed in this paper are those of Mr. Meltzer and do not necessarily reflect official positions of the Federal Reserve System or the Federal Reserve Bank of St. Louis. U.S. Policy in the Bretton Woods Era I t is a SPECIAL PLEASURE fo r me to give the Hom er Jones lecture before this distinguish ed audience, many o f them H om er’s friends. I first met Hom er in 1964 w hen he invited me to give a seminar at the Bank. At the time, I was a visiting professor at the University o f Chicago, on leave from Carnegie-Mellon. Karl Brunner and I had just completed a study o f the Federal Reserve’s monetary policy operations fo r Con gressman Patman’s House Banking Committee. Given its auspices, the study caught the atten tion o f many within the Federal Reserve. It was not surprising, then, that Hom er invited me to visit. The report had raised issues in which Homer had a long-standing interest. One o f these was the issue o f m onetary control procedures. Although Hom er was sympathetic to our criticisms, he was not easily persuaded about our proposals—such as monetary base control. Later, knowing him much better, I w ould say he was not easily persuaded about ve ry much. You had to convince Homer with facts. He re spected facts much more than clever arguments. H om er’s concern fo r facts never left him. It is not an accident that under his leadership, the economic staff at St. Louis began publishing those data triangles that economists all over the FEDERAL RESERVE BANK OF ST. LOUIS w orld now rely on w hen they want to know what has happened to m onetary grow th and the grow th o f other non-monetary aggregates. I am persuaded that the publication and w ide dissemination o f these facts in the 1960s and 1970s did much m ore to get the monetarist case accepted than w e usually recognize. I don’t think Homer was surprised at that outcome. He be lieved in the p o w er o f ideas, but he believed that ideas w ere made pow erfu l b y their cor respondence to facts. W hen Karl Brunner and I started the Shadow Open Market Committee, w e invited Hom er to be a member. He was a valuable and conscien tious m em ber w ho came to the meetings fo r many years armed with the kind o f penetrating questions that one learned to expect from him. W hen he believed that his energy had declined and he could not contribute as fully and fo rce fully as in the past, he o ffere d to resign. W e persuaded him to stay on. He rem ained through the first ten years, leaving after the September 1983 meeting, only a fe w years before his death in 1986. One o f the facts about m onetary policy during H om er’s years at the St. Louis Federal Reserve Bank is that the United States was part o f the 55 Bretton W oods system, in fact at the center of the system. Bretton W oods and international monetary policy w ere not major concerns o f the Federal Reserve how ever, despite the form al commitment to the system and the responsibili ty implied by the role o f the dollar. The failure to honor the commitment is one part o f the in flationary policies o f that period. I am pleased to review and interpret the main facts about that experience in this lecture in honor o f Hom er Jones. In the 45 years follow ing W orld W a r II, there was a remarkable transform ation o f the interna tional m onetary system. At the w a r’s end, the dollar was the dominant currency fo r interna tional transactions and was universally held as a reserve asset or store o f value. The Bretton W oods system recognized this role by making the dollar the principal reserve currency o f the international system w ith the British pound as a second reserve currency. Exchange rates o f other currencies w ere fixed to the dollar but w ere adjustable under conditions defined by the agreement. But, by 1971 the Bretton W oods system was in shambles, and in 1973 major countries agreed to experim ent with fluctuating exchange rates. This paper is about the history o f U.S. inter national economic policy under Bretton W oods from 1959 to 1973. The period begins w ith the recognition o f a problem that was to become the central problem o f the international m one tary system fo r the next decade. At first, the problem was seen as a tem porary balance o f payments problem —the inability o f the United States to balance its trade and payments at the prevailing fixed exchange rates. The end o f this historical era is fixed b y the decision in March 1973 to abandon fixed exchange rates betw een principal currencies. The starting point, 1959, is the year major currencies became convertible, subject in many cases to restrictions on capital movements that w e re increased or relaxed as reserve positions changed. Soon after the start o f the period, and at the end, policymakers expressed concern about the competitive position o f the U.S. economy. This concern about competitiveness returns again and again in the next four decades, although the focus o f concern and the principal manifes tation o f the alleged problem shift. The alleged cause in 1959-60 was trade discrimination, w hich had been accepted by the United States at the end o f the w ar to assist in the recovery from w artim e destruction abroad. Soon after, the costs o f foreign assistance and foreign military expenditures w ere added as causes. By the late 1960s these concerns and concern about foreign investment led successive adminis trations to restrict payments to foreigners by means such as the interest equalization tax, taxes on tourist expenditures, "buy Am erica” programs, and "tem p orary” controls o f foreign investment. Inflationary financing o f the w ar in Vietnam and o f domestic social spending more than offset any effect these programs may have had on the equilibrium value o f the fixed, nominal exchange rate. Increasingly, the p ro blem came to be seen as an exchange rate pro blem, specifically an overvalued dollar. As the Bretton W oods system ended, the dollar was first devalued against gold and major curren cies, then allowed to fluctuate. In the U.S. system, principal responsibility fo r international econom ic policy rests w ith the Treasury. The Federal Reserve is form ally o f secondary importance. Under Bretton W oods the Federal Reserve’s main responsibility was to conduct m onetary policy so as to maintain the fixed exchange rate agreed to by the administra tion. Th ere is no specific legislative authoriza tion fo r the Federal Reserve to buy and sell foreign currencies (Schwartz 1991). But the Federal Reserve had a larger, informal role. O f ficials and staff participated in international meetings, gave advice and counsel on what w ere seen to be the principal problems o f the Bretton W oo d system, and proposed solutions. Th ey participated, as observers, at the regular meetings o f the Bank fo r International Set tlements, w here central bankers held regular discussions and review s o f U.S. policies. There is little evidence, how ever, o f any systematic e f fo rt by the Federal Reserve to conduct m onetary policy in a manner consistent with the requirements o f a fixed exchange rate system. And, there is no evidence that any o f the administrations objected to this neglect. On the contrary, from the Kennedy to the Nixon administrations, domestic econom ic policy objec tives, though frequ ently changed, w e re o f over riding interest. THE UNITED STATES IN THE BRETTON WOODS SYSTEM The Bretton W oods agreem ent o f 1944 established a system o f fixed exchange rates based on gold valued at $35 per ounce. The MAY/JUNE 1991 56 agreem ent was the product o f extensive negotia tions, w ith much o f the w ork done by the United States and British Treasuries. The intention o f the drafters, principally John Maynard Keynes in Britain and Harry Dexter W hite in the United States, was to establish a set o f rules to replace the rules o f the international gold standard and to avoid the rigidity o f that system. Because the British feared that the United States w ould re turn to the protectionist and deflationary policies o f the interw ar years, there w ere safeguards against that occurrence. U.S. infla tion was considered unlikely or, m ore accurate ly, was not considered at all, so there w ere no rules fo r adjustment to prevent inflation from spreading to countries in the fixed exchange rate system. The agreem ent obligated countries to in tervene to keep their currencies within 1 p er cent o f their fixed but adjustable (dollar) parities. As the principal reserve currency, the United States was obligated to buy and sell gold fo r dollars (or convertible currency) at the $35 price. W hen the system started, the United States held about % o f the w orld ’s m onetary gold stock. Currencies other than the dollar w ere inconver tible. By 1960, the U.S. gold stock had fallen, but the U.S. still held $20 billion, almost half of all m onetary gold. In the early years, the United States’s loss o f gold was looked on favorably as a step tow ard convertibility. By the end o f 1958, major currencies had becom e convertible fo r current transactions.1 Th e strengthening o f foreign economies was a major aim o f early postwar U.S. economic policy. At first, balance o f payments deficits, foreign accumulation o f dollars, and the redistri bution o f the gold stock w ere seen as desirable steps tow ard a viable international monetary system. By 1960, official concern about con tinued U.S. payments deficits began to be ex pressed.2 The President’s Economic Report fo r 1960, the last report prepared b y the Eisenhower administration, discusses the competitive pro blems o f the steel and automobile industries in w orld markets during 1959 and the grow th o f U.S. investment abroad, problems that w e re to remain fo r years to come (ERP, I960).3 Sug gested rem edies are limited to pro-competitive 'Germany permitted convertibility on capital account at the same time. The Japanese yen did not become convertible on current account until 1964. 2For the year 1959 as a whole, the U.S. balance on current account was negative, -$1.3 billion, for the first time since 1953. FEDERAL RESERVE BANK OF ST. LOUIS policies, such as the rem oval o f quantitative restrictions against imports from the United States, and to recommendations that foreign governm ents increase lending to developing countries. The major problem at the time was not a U.S. current account deficit. Throughout the 1960s, the United States typically had a surplus on cur rent account. The problem was that the trade and current account surpluses w ere not large enough to finance net private investment abroad plus military, travel, and foreign aid spending abroad. T o settle the balance, the United States either had to sell gold or accumulate dollar lia bilities to foreigners. As the gold reserve declined and liabilities rose, concern increased that the liabilities w ould becom e too large relative to the gold reserve to maintain confidence that the gold price w ould remain fixed. Under Bretton W oods rules, foreigners had the option o f con verting dollars into gold; the United States had responsibility fo r keeping the gold price fixed by perm itting conversions and, at a m ore basic level, adjusting the production o f dollars to main tain confidence in future gold convertibility. This part o f the agreem ent was an early casual ty. As foreign liabilities rose, restrictions w ere placed on gold sales to private holders and pres sure or persuasion was used to discourage cen tral banks and governm ents from converting dollars into gold. Deficits and foreign dollar accumulation was not the only problem in the system as seen by U.S. policymakers at the time. A steady surplus in the U.S. balance o f payments w ould have transferred gold and dollars to the United States. Since dollar balances w ere part o f foreign reserves, but not U.S. reserves, total w orld reserves w ould fall. A U.S. surplus was seen as undesirable, therefore. Under a U.S. payments deficit, conversion o f dollars into gold left w orld reserves unchanged but low ered the gold re serves behind the principal reserve currency. W ith grow in g foreign trade, and an implicit assumption that imbalances increase w ith trade, reserves w ould prove inadequate to finance im balances at fixed exchange rates and a fixed gold price. 3I will refer to these reports as ERP (year), 57 Figure 1 U.S. Gold Reserves and Liabilities to Foreign Central Banks and Governments Billions of dollars 70 Billions of dollars 70 60 1960 61 62 63 64 65 Figure 1 shows U.S. gold reserves and dollar liabilities to foreign central banks and govern ments fo r the period as a whole. The dollar liabilities shown in the figure w ere generally about half the size o f total U.S. liabilities to foreigners. Dollar liabilities mounted, at first gradually, then m ore rapidly. By 1964, liabilities to central banks w ere equal to the U.S. gold reserve and by 1970 w ere tw ice the level o f the reserve. It seems clear in retrospect, as it did to some at the time, that the Bretton W oods system had a serious flaw. If foreign trade and payments imbalances rose w ith the grow th o f w orld in come, there w ould be grow th in the demand fo r w orld reserves. For a time, the demand could be met, as it had been, by increased foreign holdings o f dollars and a decline in U.S. gold reserves. By the time recovery from wartim e destruction abroad reached the point that re 66 67 68 69 70 71 1972 strictions on convertibility could be removed, the United States was losing gold and accumu lating liabilities at rates that threatened the system’s long-term stability. Under the Marshall Plan (1948-52) U.S. policy encouraged foreign governm ents to rebuild reserves as a step tow ard convertibility. Coun tries w ere permitted, and even encouraged, to export to the U.S. w hile restricting imports from the United States. W hen the Marshall Plan ended, military expenditures to support military commitments and aid to developing countries maintained a flo w o f dollars equal to m ore than $2 billion a year. This helped w estern European countries as a group to achieve a current ac count surplus b y the mid-1950s, despite a conti nuing trade deficit. Convertibility fo r current account transactions at the end o f 1958 (as well as fo r capital account transactions in Germany) MAY/JUNE 1991 58 plus the commitment to reduce internal trade barriers as part o f a common market stimulated U.S. investment in Europe.4 Hence, a capital in flo w augmented the stock o f foreign exchange reserves in countries outside the United States. In the tw o years 1958 and 1959, gold and dollar reserves o f the principal European countries in creased by $5 billion, about 25 percent o f total U.S. reserve assets at the end o f the period.5 Th ere w e re four possible solutions (Friedman, 1953): (1) devaluation against gold and major currencies, (2) deflation, (3) b o rro w as long as foreigners w ould lend, and (4) impose controls o f various kinds. Several o f these solutions could be achieved in different ways. For exam ple, foreigners could revalue against the dollar. Or, foreigners could inflate faster than the United States, thereby changing the relative prices o f domestic and foreign goods at fixed exchange rates. Policies o f the K ennedy and Johnson Administrations Under Bretton W oods rules countries w ere perm itted to devalue up to 10 percent without consultation w hen faced w ith “ fundamental disequilibrium." The precise conditions char acterizing fundamental disequilibrium w e re not spelled out, and the International M onetary Fund (IMF) did nothing to clarify the conditions. The drafters had w anted to avoid both the inflex ibility o f the classical gold standard and the competitive devaluations o f the interw ar period. The language may have been intended to permit devaluation if the alternative was deflation w hile avoiding devaluation if a country could expect to restore payments’ balance and repay a shortor medium-term adjustment loan. Devaluations by major countries occurred. In the early years several countries follow ed the United Kingdom in a 30% devaluation in 1949. France devalued by 29 percent in 1958, and the United Kingdom devalued again in 1967. The United States chose to regard its problem as less than “ fundamental.” President Kennedy came in 4See ERP (1959). 5Total U.S. reserve assets include the total U.S. gold stock and its reserve position in the International Monetary Fund. At the end of 1959, these balances were $19.5 and $2.0 billion respectively. ERP (1971). 6The Kennedy Treasury led by Douglas Dillon and Robert Roosa was firmly opposed to devaluation. Dillon was an in vestment banker, and a Republican, who had served as FEDERAL RESERVE BANK OF ST. LOUIS to office committed to maintain the $35 gold price but also to “ get the econom y m oving” after the relatively slow grow th and tw o reces sions in the previous four years. The comm it ment to a fixed nominal gold price ruled out devaluation o f the dollar against gold and all other currencies; the commitment to higher economic grow th rem oved the classical remedy, deflation o f domestic prices and costs o f p ro duction to raise the real dollar value o f the U.S. gold stock and low er the relative price o f U.S. exports. That left borrow ing, controls and fo r eign inflation as the principal options. The decision to avoid devaluation and defla tion reflects some strongly held views o f the period. The $35 gold price was seen as a firm commitment under the Bretton W oods A gree ment. If the United States devalued once, it could do so again, w ith costs to the stability that Bretton W oods was supposed to provide. Avoiding repetition o f the experience w ith defla tion in the early 1930s and the decade-long depression was a major factor in the passage o f the Employment Act and the Bretton W oods agreement. Few wished to repeat the prew ar experience even in m ilder form . Hence, the Kennedy administration met little opposition from business or political groups in excluding the price options, devaluation and deflation.6 Kennedy’s main domestic campaign theme had been getting the econom y "m oving” after the relatively slow average grow th rate o f the se cond Eisenhower administration. The Kennedy administration policies emphasized domestic growth, full employment and price stability as their major aims and relied on a so-called fiscal m onetary mix to stimulate output w hile reduc ing the capital outflow. In practice, this trans lated into a series o f tax measures—faster de preciation fo r capital, an investment tax credit and, later, reductions in personal and corporate tax rates. T o limit the capital outflow, the ad ministration tried to prevent a sharp counter cyclical decline in interest rates during the early months o f the recovery from the 1960-61 reces- ambassador to France. Roosa had been a senior econo mist at the N.Y. Fed and was very attracted to activist policies whether in domestic credit markets or international markets. At the Council of Economic Advisers (CEA), Walter Heller was mainly interested in domestic policy. Heller (1966) says very little about the dollar problem other than noting that the balance of payments required higher short-term interest rates. Kennedy saw the problem as a matter of prestige. Sorensen (1965), pp. 405-12. 59 Figure 2a Federal Funds Rate During the 1960s sion.7 In cooperation w ith the Federal Reserve, the Treasury attempted to "tw ist” the yield curve by buying long-term bonds and selling short term Treasury bills.8 Since the market fo r governm ent securities is ve ry active and highly competitive, participants w e re able to reverse any tem porary change in interest rates achieved by the twist. puted as the percentage rate o f change from the corresponding month o f the preceding year. Shortly after the Kennedy administration came into office, grow th o f the base rose to about 1-1/2 to 2 percent. By mid-1963 the grow th rate o f the base was consistently above the long term grow th o f output, 3 percent per annum. Growth o f the base continued to rise in 1964. Neither m oney grow th nor interest rates shows evidence o f "tight m oney” in the early 1960s. Figure 2a shows the federal funds rate fo r that decade. The funds rate is the rate most tightly controlled b y the Federal Reserve. From 1961 to 1966, the rate rose slowly, and it did not exceed 3 percent until late in 1963. Figure 2b shows the grow th rate o f the m onetary base fo r the same period. The grow th rate is com W ith the possible exception o f 1961-62, base grow th shows no evidence that m onetary policy was relatively restrictive. In fact, base growth, far from being deflationary, was inconsistent w ith continuation o f the relatively low rate o f inflation inherited from the past. To prevent higher inflation, the Kennedy administration in troduced inform al guidelines fo r prices and wages. The prevailing view was set out in the 7See Heller (1966), p. 5. 8The Treasury also auctioned “ strips” of bills that required dealers to buy more than one issue at a time on the im plausible assumption that the additional distribution cost would be treated by the market as a rise in the effective interest rate instead of a fee for service. MAY/JUNE 1991 60 Figure 2b Annual Growth of the Monetary Base During the 1960s Percent Change Percent Change 8 8 1962 Econom ic Report o f the President. In this view , a relatively stable Phillips curve permitted policymakers to trade higher inflation fo r low er unemployment. The price and w age guidelines w ere supposed to im prove the trade-off by reducing w age and price increases as the econom y approached full employment. monthly until 1966. CPIs are relatively com prehensive measures but, as is w ell known, not perfect measures o f traded goods and services. Doubtless there are temporal differences bet w een the price ratio in figure 3 and ratios com puted using other prices, but the pattern o f per sistent decline w ould be little affected. A slow recovery gave w ay in 1963 to robust real growth. Inflation remained low. Until 1965, the fixed w eight deflator rose betw een 1 per cent and 1-1/2 percent annually and the con sumer price index betw een 3/4 percent and 2 percent. Inflation was generally higher abroad, so relative prices o f U.S. goods declined. Figure 3 shows the ratio o f the U.S. consumer price in dex to a trade-weighted average o f consumer prices abroad, based on the weights in Federal Reserve index o f nominal exchange rates. The index shows the sustained relative decline in the U.S. price level during the first half o f the decade. W ith fe w exceptions, the ratio declined Under the impact o f relative price changes and other factors, the merchandise trade balance increased. U.S. capital investment abroad con tinued to rise and domestic and foreign b o r row ers used the U.S. market to raise funds fo r investment abroad, so the capital outflow continued. 9Various definitions were used but the official settlements balance appears to have been the main concern. This balance consists of current account plus long-term capital flows plus net private short-term capital flows. I report this balance from time to time in the text, but I base my judg ments much more on the current account balance. The FEDERAL RESERVE BANK OF ST. LOUIS Special Measures From 1960 on, each administration sustained or strengthened controls on trade and payments intended to reduce the balance o f payments deficit.9 At first, the measures consisted o f concern about profitable investment abroad puzzles me; investment of this kind produces a subsequent inflow. Also, the current account balance is more useful for com parison of the fixed and fluctuating rate periods. On the role of the current account see Corden (1990). 61 Figure 3 Index of the Ratio of U.S. CPI to Trade-Weighted CPI 1973 = 100 120 1973 = 100 ---------------------------------------------------------------------------------------------------------------------------- — — — I 120 100 relatively modest steps to encourage exports (by subsidized loans from the Export-Import Bank) or to purchase military equipment and supplies in the United States even if it imposed higher costs. W ithin a fe w years, the list o f controls and their efficiency cost increased. Development aid was tied to dollar purchases. An "interest equalization” tax was put on foreign borrow ing and later raised. Most tourist expenditures w ere made dutiable. And guidelines w ere used to fu r ther limit the expansion o f bank lending to fo r eigners and to limit grow th o f foreign direct in vestment. Table 1 lists some o f the measures proposed or adopted. The efficacy o f the controls varied with the opportunity or ability to substitute uncontrolled fo r controlled transactions. Tieing military expen ditures is inefficient to the extent that it raises costs, but substitution is probably limited. Re quiring foreigners to pay the cost o f U.S. troops abroad increases exports and the current ac count and reduces the capital outflow. Tieing foreign aid or military spending reduces the real value o f the spending but may lead to high er appropriations, so that over time the actual dollar outflow is not much affected. Restrictions on foreign borrow ing in the U.S. market are circumvented if foreigners or domestic citizens sell their holdings o f dollar securities to pur chase new issues sold abroad. Further, restric tions o f foreign lending and investment affect relative rates o f interest at home and abroad thereby reducing the effect o f the restrictions. And foreign firms im port fe w e r goods from the United States. MAY/JUNE 1991 62 Table 1 Selected Balance of Payments Measures (Actual and Proposed) 1960 Expansion of Export-lmport Bank lending and guarantees of non-commercial risks. Reduction in military dependents abroad (repealed in 1961). Reduction in defense and non-defense government purchases abroad. 1961 Offsets for military expenditure in Europe and additional procurement at home. Tieing development aid to dollar purchases. Increased taxes on foreign earnings of U.S. corporations. Reduced allowance for tourist purchases abroad from $500 to $100. Treasury intervention in foreign exchange markets. Repayment of German loans. 1962 Expansion of earlier programs. Offset purchases by Germany and Italy. Increased borrowing authority for the IMF. Beginning of Federal Reserve “ swap” arrangements for currencies. Treasury issues foreign denominated securities. 1963 An interest equalization tax of 1% on foreign borrowers in U.S. market. Additional tieing of foreign aid to domestic purchases. 1965 Interest Equalization Tax on bank loans with duration of one year or more made to bor rowers in developed countries (except Canada). Limits on growth of bank lending to foreigners. Encourage private companies to increase exports and repatriate earnings. Guidelines for direct investment by non-financial corporations to limit growth of foreign direct investment. 1967 Permit higher tax rates (up to 2%) for interest equalization tax. Expansion of lending authority of Export-lmport Bank. Source: Economic Report of the President, various years. Controls and restrictions w ere, at best, a short term solution. Most o f the controls and restric tions in table 1 w e re introduced as tem porary measures, although several w ere extended and strengthened w hen renewed. Even if these mea sures had succeeded in stemming the balance o f payments deficit, they did not o ffe r a perm a nent solution at a new equilibrium without con trols. The problem w ould have returned when the controls w ere removed. T o smooth fluctuations in the gold price and short-term capital movements, the Treasury in troduced several measures. Eight countries join ed a gold pool in 1961 to stabilize the London gold market. Reciprocal credit agreements (call ed “ swaps”) w ith foreign central banks and the Bank fo r International Settlements provided 10See Solomon (1982), p. 42. "C anada, Japan, Belgium, France, Germany, Italy, Netherlands, Sweden, United Kingdom, United States. FEDERAL RESERVE BANK OF ST. LOUIS loans o f foreign currencies and dollars. Typical ly the Federal Reserve b orrow ed to purchase dollars held abroad instead o f selling gold.1 0 Swaps are a short-term accommodation. T o re pay the swaps, the Treasury began borrow in g from foreign central banks at longer terms us ing bonds denominated in foreign currencies. The proceeds from the bond sales (called Roosa bonds) helped to repay the swaps without re ducing the U.S. gold stock, again postponing the problem. Lending facilities o f the IMF w e re ex panded under the General Agreem ents to Bor row . The agreements provided that 10 countries would lend under specified conditions to aug ment the Fund's resources. This is the origin o f the group o f 10 (G-10).1 Again, these w ere 1 mainly short-term measures. Later, Switzerland joined the G-10, but the name remained. 63 The United States supplied 60 percent o f the gold sold in the London gold market; the other members o f G-10 w ere supposed to provide the rest. Foreign countries replaced some o f their sales by purchases from the United States, so the U.S. contribution to the pool, direct and in direct, became a major cause o f the decline in the U.S. gold stock. In March 1968, w ith U.S. gold reserves under $11 billion, the gold pool was abandoned. The price o f gold fo r official transactions rem ained at $35, but the G-10 gov ernments did not attempt to control the price fo r private transactions. To prevent arbitrage, foreign central banks agreed not to sell in the gold market. For the years 1960-67 as a whole, the non-U.S. members o f the G-10 (including Switzerland) ac quired 150 million ounces o f gold, an increase o f one-third over their holdings at the end o f I960.1 Every country except Britain and Canada 2 added to its stocks. Britain sold 38 million ounces, the U.S. 164 million ounces. France acquired m ore than 100 million ounces, two-thirds o f the total acquisition by G-10 countries.1 3 The year 1966 is the peak year fo r gold hold ings o f the G-10, excluding the United States, and 1965 is the peak year fo r the eleven coun tries, measured in ounces o f gold. In these years, the value o f the stock at $35 per ounce was ap proximately $15 billion. The value o f the U.S. stock was approximately $13 billion, about equal to the non-gold foreign exchange holdings of the other members o f the G-10. A fter 1968, nonU.S. members o f the G-10 as a group reduced their stocks slightly until 1971, but several ac quired gold in the market. The embargo can be said to have succeeded in this limited sense. Also in 1961, the Treasury's Exchange Stabili zation Fund (ESF) began operations in foreign currencies fo r the first time since the 1930s. The Federal Reserve joined in these operations in 1962, and in 1963 the Fed began lending dollars to the Treasury secured by Treasury holdings o f foreign exchange. These so-called “warehousing” operations perm itted the T rea sury to expand its purchases o f foreign ex change without seeking Congressional appropri ations to support the activity. Warehousing re mained small in the 1960s but increased sub stantially in the 1980s. Proposals f o r Long-Term Adjustment The policies o f the Kennedy and Johnson ad ministrations may have stabilized the level o f foreign official holdings in the years 1963-66 but, as shown in figure 1 above, U.S. gold re serves continued to decline. A popular analogy at the time treated the United States as a bank fo r the international m onetary system. Foreign dollar holdings w ere considered the analogue o f bank deposits and gold the analogue o f bank reserves.1 The analogy suggests that the series 4 o f mainly short-term, one-time or allegedly tem porary measures, such as the interest equaliza tion tax, had not solved the problem o f a future run on the bank. The gold reserve continued to fall absolutely and relative to official (or total) dollar claims. The Kennedy administration was aware o f the long-term problem. In 1962, the administration asked economists at the Brookings Institution to study the longer-term prospects fo r balance of payments adjustment. The report took the "basic balance"—balance on goods and services, govern ment payments plus long-term capital flo w —as its standard.1 It projected that by 1968 this bal 5 ance w ould be betw een a surplus o f $1.9 billion and a deficit o f $600 million, depending on the assumptions about growth, prices and costs at home and abroad. This section o f the report was greeted warm ly by the administration. The projected im provem ent reflected the assumption o f a rise in foreign relative to domestic prices and a slowing o f U.S. investment abroad as p ro fit rates rose in the United States relative to abroad. The expected rise in domestic profit rates reflected the direct effect o f the Adminis tration's proposed reduction in corporate tax rates and the indirect, stimulative effect o f tax rate reductions fo r households and businesses.1 6 As shown in figure 3, a relative increase in prices abroad occurred, but the projections pro ved optimistic. The recorded 1967 balance was -$2.1 billion.1 7 12AII data on gold are from IMF (1990), p. 65. 15See Salant et.al. (1963). 131967 is the peak for France’s accumulation of gold. The following year France sold 40 million ounces to defend the franc’s parity following the riots and disturbances. 16For the tax reduction to improve the basic balance, the rise in expected real returns in the U.S. had to overcome the expected positive effect of tax reduction on imports. 14The analogy neglects the role of the pound as an alter native reserve currency, but its role was small and of declining importance. 17See ERP (1964), p. 131. MAY/JUNE 1991 64 The Brookings report also considered the e f fects o f a U.S. surplus in its basic balance on the supply o f w orld reserves and concluded that either a new source o f w orld reserves w ould have to be found, or there w ould have to be greater flexibility o f exchange rates. The report discussed a dollar-pound bloc and a continental European bloc w ith fixed rates inside the bloc and fluctuating rates betw een the blocs.1 The 8 Council o f Economic Advisers summary o f the Brookings study has no reference to this discussion. The Council o f Economic Advisers argued that, although the Bretton W oods agreement permitted exchange rate adjustments, "fo r a reserve currency country, this alternative is not available.”1 And, they added, "fo r other major 9 industrial countries, even occasional recourse to such adjustments w ould induce serious specula tive capital movements, thereby accentuating imbalances.”2 0 W ith exchange rates adjustments ruled out, only tw o alternatives w ere considered. One was increased fiscal expansion by surplus countries and less expansive policies fo r countries in defi cit. The other was introduction o f some type of new reserve asset. The latter proposal led even tually to the creation o f special drawing rights (SDRs). This was the heyday o f Keynesian policy, so it is not surprising that Keynesian policies have a prom inent place in administration proposals. The administration favored a policy mix and what later became known as policy coordina tion. Under the fixed exchange rate system, countries w e re expected to buy and sell dollars to maintain their exchange rate. The economic reports o f the President fo r the period assumed, how ever, that countries can adjust capital flow by varying the mix o f fiscal and monetary poli cies. The ERP argues that “flexible changes in the mix o f fiscal and m onetary policies can serve to reconcile internal and external policy goals.”2 1 For the United States, the prescription was tax reduction to expand domestic spending w hile holding short-term interest rates high to reduce short-term capital outflow. Surplus countries w ith strong domestic demand w e re called upon to raise tax rates or low er governm ent spending and expand money grow th to low er interest rates. The idea was that the inflationary conse quences o f domestic m onetary expansion would be reduced or avoided by the restrictive fiscal policy. The International M onetary Fund’s annual report fo r the period offers similar advice, but it warns o f an inflationary bias. Surplus coun tries are subject to upw ard adjustment o f wages and prices, but deficit countries are not sub ject to dow nw ard adjustments.2 The IMF recog 2 nized that w orld trade had grow n faster than the gold stock, but they limited their recom m en dation to a study o f possible future needs.2 3 The Germans, to w hom the recommendation fo r m onetary expansion and fiscal restraint was often directed, w ere skeptical about the policy mix proposals. Their expressed concern was the inflationary consequences o f U.S. m oney growth. They held to a m ore classical view that the probblem was expansionary U.S. m onetary policy, so it must be solved by restrictive policies in the United States, not expansive German policies. Their skepticism about "coordination” becam e a persistent feature o f the policy dialogue under both fixed and fluctuating exchange rates. Initially the German response was to discourage capital inflows. For example, German banks w ere required to hold relatively high reserve requ ire ments against foreigners' deposits. The Germans argued, against the spirit o f the Bretton W oods agreement, that deficit countries should adjust. They opposed revaluation o f the mark even m ore strongly than they opposed domestic ex pansion, since they could avoid revaluation but could avoid expansive m onetary policy only by imposing severe restrictions on capital inflows. A fter much delay, and many denials, Germany revalued the mark b y 9.3 percent in October 1969.2 W ith the revaluation, Germany rem oved 4 many o f the border taxes and special reserve requirements on foreign deposits in German banks that had been used to limit capital in- 18See Salanl et.al. (1963). 23lbid., p. 32. 19See ERP (1964), p. 139. 24The mark had been revalued by 5 percent in 1961. Solomon (1982), p. 162, reports the May 1969 statement given by a German official that the decision to not revalue was “ final, unequivocal and for eternity.” This is one of many strong denials during the period. 20lbid. 2 See ERP (1964), p. 143. 1 22See IMF (1964), p. 28. FEDERAL RESERVE BANK OF ST. LOUIS 65 flows to Germany.2 A fter the revaluation Ger 5 man prices rose m ore slowly than U.S. prices. The experience o f the late 1960s had a lasting effect on German m onetary policy. A fter the mark re-entered a fixed exchange rate system w ith the principal continental European coun tries in the late 1970s, Germany revalued more frequently to avoid inflation and exchange con trols. In the late 1960s, how ever, revaluation was delayed too long and was much too small to offset the effects o f inflationary U.S. monetary policy.2 6 W hile urging German revaluation, the United States increased m oney grow th in 1968. A fter the German revaluation, the Federal Reserve shifted to a m ore restrictive policy, raising the Federal funds rate (figure 2a) and slowing the grow th o f the m onetary base. The sharp con traction in the grow th o f the base was follow ed by the start o f a recession in the fourth quarter. The reduction in m oney growth, the increase in U.S. real rates o f interest, and the recession helped to shift the current account balance to w ard surplus. By the middle o f 1970 the quar terly balance had returned to a level not reached since the latter part o f 1965. Proposals to Increase Liquidity “Adjustment” was one o f a triad o f topics dis cussed at numerous official and unofficial inter national meetings. The other topics w ere “li quidity” and “ confidence.” Liquidity received the most attention. Proposals fo r additional liquidity differed. The French position was one extreme, the U.S. posi tion the other. The International M onetary Fund took a position close to that o f the United States. Positions did not remain fixed, but they w ere never fully reconciled.2 7 The U.S. position was that a new reserve asset was needed to supplement the stock o f gold and dollars. Sales from official holdings in the London gold market had reduced official hold ings during the middle 1960s, and dollar liabilities had continued to rise relative to the U.S. gold stock. The United States argued that it expected to bring its payment deficits to an end. W hen this happened, the w orld trading system w ould lack an adequate supply o f reserves to finance future demand fo r reserve assets at the fixed gold price. Hence, the United States favored creation o f a new reserve asset that could be increased w ith w orld trade or w orld demand fo r reserves. This argument is, at best, incomplete. I f the United States had a payments surplus, other countries w ould have deficits. The United States could add to reserves by buying other stable currencies, just as these countries bought dol lars. The Econom ic Report recognizes that this argument is correct. Even if each country was in balance, the United States could buy foreign exchange fo r dollars to augment its reserves.2 8 The IMF combined the liquidity and adjust ment issues. They argued that the creation o f a new reserve asset and additional reserves re duced the need fo r deficit countries to adjust and increased the pressure on surplus countries to adjust.2 This is, o f course, an argument fo r 9 inflation as a solution to the adjustment prob lem fo r the deficit countries and revaluation as the rem edy fo r the surplus countries. This p ro gram was asymmetric; w orld inflation w ould in crease but not decline. Data in the IMF report, how ever, do not show a general problem o f liquidity at the time. The IMF used reserves as a percentage o f imports to measure liquidity—on the usual assumptions that reserves are used to finance imbalances and imbalances increase with trade. The data show that there was no general shortage o f li quidity on this measure. The problem was lim ited mainly to the United States and the United Kingdom. Table 2 shows these data. Reserves in clude gold, foreign exchange and reserve posi tion at the IMF. 25See Solomon (1982), p. 164. 28See Economic Report (1964), p. 145. 260th er parity changes during the second half of the 1960s include an 11.1 percent devaluation of the French franc in 1968 and a 14.3 percent devaluation of the British pound in 1967. Many of the sterling bloc devalued following Britain. 29See IMF (1966), p. 10. 27Solomon (1982) gives a thorough account of the discus sions, proposals and the meetings of various official groups. Solomon was a senior civil servant at the Federal Reserve with responsibility for international finance and an active participant or observer at most of the discussions. MAY/JUNE 1991 66 Table 2 Ratio of Reserves to Imports 1951 All countries1 All countries except U.S. G-10 Japan Germany United Kingdom United States 1959 1965 67% 39 58% 44 43% 49 73 14 22 204 63 40 34 25 126 43 26 42 19 67 ’ approximately 60 countries Source: IMF, 1966, p. 13 The United States aside, w orld reserves w ere a larger percentage o f imports in 1965 than in 1951 and not very different in 1965 than in 1959. The IMF notes that the ratio o f reserves to imports fluctuated around a constant value.3 0 Am ong major countries, only the United King dom shows a relatively low ratio. For many countries, the ratio had converged to 40 to 50 percent.3 1 The French complained about the special role o f the dollar, the opportunity given to the United States to use domestic inflation to ac quire foreign assets (at fixed exchange rates), and what they called U.S. hegemony. As a first step, they proposed to limit the size o f U.S. payments deficits that other countries w ere obligated to finance, but they also sought a per manent arrangement under which reserve assets would be tied to gold. Later, they urged an in crease in the price o f gold. Jacques Rueff (1967) proposed a doubling o f the price o f gold accom panied by commitments by the United States and the United Kingdom to use part o f the p ro fit from revaluation to retire some o f the dollar and sterling reserves held by foreign central banks.3 2 In its official proposals, the French governm ent did not at first go as far as R u eff in favoring an increase in the gold price. Nor did it insist on a return to the gold standard. It favored a larger role fo r gold, restrictions on the financing o f U.S. deficits, and a refunding o f the sterling and dollar balances, particularly the latter.3 In early 3 discussions, France wanted to circum vent the IMF by creating a reserve asset fo r use b y the Group o f 10. The new asset w ould have a p er manently fixed relation to gold. The effect o f this proposal was to increase the effective gold stock and to devalue the dollar against gold. The French proposal was not accepted. The alternative chosen was to create a new asset. In September 1967, at the Rio de Janeiro meeting o f the International M onetary Fund, agreem ent was reached on the general principles govern ing creation o f a supplementary reserve asset called Special D raw ing Rights (SDRs). The new asset was to be a supplement to gold and dol lars. The United States was not required to redeem dollar balances, and the gold price re mained fixed. SDRs could be issued only if an 85 percent majority approved, and they w ould be held only by official holders, central banks and international m onetary institutions. Finally, in July 1969, the amendments to the IMF agree ment w ere ratified by a sufficient number o f members to come into effect. A t the Rio meeting o f the IMF, the first allocations w ere agreed to but not issued. In the IMF view, reserves w e re “ less than ade quate."3 The report acknowledged that “the 4 signals w e re conflicting.”3 The principal argu 5 ment fo r m ore reserves is that non-tariff bar riers, aid-tying, domestic preference and other trade restrictions had increased. At the time, no argument was made about the relation o f re serves to trade or imbalances. And the argument about trade restrictions makes no effo rt to link trade restrictions to the liquidity problem. In fact, the restrictions continued in many coun tries after 1973. SDRs w ere issued in 1970-72 and again in 1979-81. The 1970 issue added $3.1 billion to reserves. In the same year foreign exchange reserves increased by $14 billion, and total reserves reached $91 billion, a 50 percent in crease fo r the decade and a 22 percent increase fo r 19 70.3 In total, 21.4 billion SDRs (valued in 6 SDR units) w ere issued through 1985 w hen new 30See IMF (1966), p. 12. 34See IMF (1969), p. 27. 3 Ibid., p. 14. 1 35lbid., p. 26. 32Rueff was Economic Adviser to President DeGaulle. 36See IMF (1971), p. 19. 33See Solomon (1982), p. 73. FEDERAL RESERVE BANK OF ST. LOUIS 67 issues ceased. SDRs never became an important means o f settlement. By the time the agreement to create SDRs had been reached, the Bretton W oods system was in its last days. It seems doubtful that the SDR w ould have becom e a dominant medium o f exchange or store o f international reserves if the fixed ex change rate system had survived. The SDR was a specialized money but did not dominate alter natives as a means o f payment or store o f value. Gold is an established store o f value with a long history. SDRs had to compete also with the dol lar and later the mark, the yen and other cur rencies as a reserve asset. Balances held in each o f these assets earn interest. At first, SDR bal ances did not earn interest, so they w ere less attractive than balances held in short-term gov ernment securities o f the principal countries. Th ere was no source o f revenue or earnings; interest payments could only be made by crea ting additional SDRs. A fter 1973, flexible exchange rates rem oved any need fo r a large stock o f reserves fo r settl ing balances betw een principal countries, although countries continued to accumulate reserves and to intervene in the foreign ex change markets. The SDR could not be held by private wealthowners, so it could not be used fo r intervention. Further, the introduction of SDRs did not adjust relative prices or real ex change rates. Failure to solve the adjustment problem meant that the major e ffo rt to sustain the system by producing a supplementary reserve asset was largely wasted effort. Table 3 Rates of Growth of Money, 1969-71, Selected Countries Country 1969 1970 1971 United States France Germany Italy Japan Netherlands United Kingdom 5.9 6.1 8.2 15.0 18.4 9.4 - 0 .3 3.8 - 1 .3 6.4 21.7 18.3 10.6 6.8 6.8 13.7 12.0 22.9 25.5 16.7 12.9 Source: International Economic Conditions, Federal Reserve Bank of St. Louis, June 1982. inflation to the rest o f the world. Countries w ere obligated to buy all dollars offered at a fixed price. Sterilization o f the inflow could be successful fo r short-periods, but as Switzerland, Germany and others discovered at the time, the fixed exchange rate gave speculators an oppor tunity to invest in one-way gambles with low risk and relatively high expected return. The Swiss franc or the m ark w ere unlikely to depre ciate, more likely to appreciate. Investors and speculators knew this. Hence, flow s into these currencies became difficult to curtail. Additional creation o f SDRs in 1972 was again divorced from events. W orld reserves (net o f gold) had doubled in tw o years (from SDR 56 in 1970 to SDR 112 in 1972). Reserves in relation to imports w ere at the highest level in the post w ar period before or after; countries held reserves equal to m ore than 14 w eeks o f im ports at the end o f 1972. In 1963, at about the time that the discussion o f additional reserves began, the ratio was equal to nine weeks. In the same period, 1963-72, total reserves (net o f gold) quadrupled in nominal value.3 7 Table 3 shows the pattern o f money grow th fo r a sample o f countries. Many countries show a decline in m oney grow th from 1969 to 1970 and all show a rise from 1970 to 1971, co rre sponding to the pattern in the United States, as France claimed. The size o f the changes differs by country. All countries did not have the same trade pattern, so they did not receive the same proportional increase in base money. Some ster ilized part o f the increase fo r a time, and some countries adopted controls to reduce the inflow. In the w inter o f 1971, Germany allowed its ex change rate to appreciate relative to the dollar, slowing the in flow o f dollars. This action recog nized, as France has insisted, that countries could not prevent inflation w hile maintaining their dollar parities. The French w ere correct on tw o points that U.S. officials (and others) refused to acknow ledge. First, the Bretton W oods system based on the dollar permitted the United States to export The second point on w hich the French posi tion was correct was that revaluation o f gold w ould solve the liquidity problem. Their various proposals w ould have devalued the dollar against 37Data are from International Financial Statistics, Yearbook 1990. MAY/JUNE 1991 68 other currencies, thereby providing the addition to the stock o f reserves that the SDR was sup posed to provide. French proposals did not limit future changes in the price o f gold, so they are open to the charge that expectations o f future devaluation w ould lead to a run on the dollar once it had been devalued. Much earlier, Keynes (1923) had proposed a type o f commodity stan dard in w hich gold served as a medium o f ex changes. In this proposal, the gold price was tied to an index o f commodity prices. Had a scheme o f this kind been adopted, it seems like ly that the Bretton W oods system w ould have lasted longer. A devaluation o f the dollar against gold, with other currency values unchanged, w ould have rem oved the liquidity problem in the 1960s. At the end o f 1968, the U.S. price level was ap proxim ately 2-1/2 times the 1929 level. I f the gold price had been raised proportionally from its 1929 value ($20.67), the 1968 price w ould have been approximately $52. At that price, the U.S. gold reserve w ould have been $17.6 billion, $1.6 billion m ore than U.S. liabilities to central banks and governments. Although adjustments in the price o f gold w ould have extended the life o f the Bretton W oods system, it is unclear w hether the system would have survived fo r m ore than a fe w addi tional years without some restriction on U.S. m onetary policy, restrictions that the United States was unlikely to accept. Inflationary pol icies in the United States continued through the 1970s w ith only b rie f interruptions. Countries that chose to low er inflation in the 1970s would have had to leave the system. Further, inflation was not the only problem. The oil shocks o f 1974 and 1979 changed the terms o f trade, re quiring changes in exchange rates that Bretton W oods system found difficult to accommodate. At best, devaluation o f the dollar against gold w ould have corrected fo r differences in produc tivity grow th and changing costs o f production betw een the United States and the principal surplus countries, Germany and Japan. U.S. devaluation and the oil shocks w ould have changed the relative positions o f other coun tries. Some w ould have found their trade bal ances in persistent deficit, requiring adjustment o f their relative prices and costs or devaluations and revaluations o f bilateral and multilateral 38The use of reserve currencies was not part of the Bretton Woods plan. Britain's role evolved from its prewar position and the holding of sterling balances by countries in the FEDERAL RESERVE BANK OF ST. LOUIS rates, adjustments that w ere difficult to make and which would, in turn, have required fu r ther adjustments. There is a plausible case to be made on the other side—that devaluation o f the dollar w ould have prolonged the life o f the Bretton W oods system. The key assumption is that the oil pro ducing countries raised the price o f oil in re sponse to the decline in their real incomes after the dollar floated. A modest devaluation to a new fixed parity might have avoided the first oil shock. If so, the mistaken policies in the United States, attempting to offset the real effects o f the oil price rise by inflation, w ould have been avoided. Inflation w ould have been low er and U.S. nominal gold reserves larger. It seems unlikely, how ever, that other countries would have accepted a U.S. policy o f inflation and repeated, periodic devaluation against gold. Without a low er rate o f inflation in the United States, the Bretton W oods system w ould have failed sooner or later. Nevertheless, the French proposal was a straightforw ard solution to the liquidity prob lem o f the 1960s. It w ould have resolved the li quidity problem at least fo r a time but w ould not have resolved the m ore difficult "adjustment problem s” arising from changes in countries’ prices, productivity, and costs o f production. This was not the main reason fo r rejecting the proposal, how ever. Representatives o f the g o v ernments and central banks claimed that any change in the $35 gold price or devaluation by a reserve currency country w ould damage "confidence.” Confidence Throughout the 1960s, there w ere concerns about w hether the United States could avoid default on its obligation to convert dollars into gold at the $35 gold price. Expressions o f lack o f confidence in the dollar often brought forth speeches by Presidents, Treasury secretaries and others to bolster "confidence.” W ords w ere not the only response. Actions w ere taken to strengthen or restore confidence. The dollar and, to a lesser extent, the pound had a special role in the Bretton W oods system. They w ere “reserve currencies.”3 Central banks 8 held dollar or pound securities as reserves in sterling bloc. The dollar’s role evolved from the dollar bloc and the unique position of the United States in the early postwar years. 69 place o f gold to earn interest on their balances. Devaluation reduces the real value o f these re serves, so anticipation o f a devaluation could lead to a run on the dollar or the pound. Once total claims against the reserve currencies ex ceeded the gold reserves held by the United States and Britain, discussion o f the confidence problem intensified. The first evidence o f a lack o f confidence in the dollar was a tem porary rise in the gold price in 1960. In October, the gold price on the Lon don market m oved above the official inter vention price, $35.20 an ounce. Speculators said that the market was concerned about the possi ble election o f John F. Kennedy as president and the continued capital outflow from the United States. Kennedy’s talk o f "getting the econom y m oving” may have seemed inflationary. T o counter these concerns, Kennedy made a strong commitment to maintain the gold value o f the dollar, and the Eisenhower administration took the first steps (discussed above) to reduce the U.S. payments deficit. Th e tem porary increase in the London gold price reflected both a rising demand fo r gold from European central banks and private hold ers and a refusal by the United States to supply gold to the market. The London gold price was set to clear trading. To maintain the price within its band, the Bank o f England bought or sold gold fo r dollars, replacing the dollars or gold in an exchange w ith the U.S. Treasury. In October 1960, the Treasury appeared unwilling to restore the Bank’s gold holdings, so the Bank refused to buy dollars fo r gold. W ith the resid ual buyer o f dollars inactive, actual and p ro spective supply was reduced; the price o f gold rose to $40 on October 27. The Treasury re sumed sales, and the price returned to $35. T w o changes w ere introduced as a result o f this experience. European central banks agreed not to buy gold on the London market if the price rose above the U.S. price plus shipping cost, $35.20. In October 1961, seven European governments and the United States created a gold pool. Each m em ber o f the pool agreed to let the Bank o f England buy and sell fo r the group, and each received a pro-rata share o f any gold purchases and supplied a share o f gold 39See Schwartz (1987), p. 342. 40Unit labor costs (ULC) are available for Canada, Japan, Germany and the United Kingdom from 1963 to date. The index of foreign ULC is based on these countries. All data sales. During the years that the pool functioned, m em ber countries sold gold w orth (net) $2.5 billion on the London market. The U.S. share was $1.6 billion.3 As shown in figure 1, during 9 approximately the same period, 1961-67, U.S. gold reserves fell m ore than $5 billion, the dif ference reflecting direct sales from the U.S. gold stock. But, as noted earlier, during the same period, the countries in the G-10, and especially France, added m ore than 100 million ounces ($3.5 billion) to their official gold reserves. The amount added by the G-10 represents 97 percent o f the sales by the United States outside the gold pool. These data suggest that the pool did not function as intended; the G-10 replaced their sales from U.S. stocks. The years 1962-64 saw a substantial increase in the U.S. current account surplus and reduc tion in the payments imbalance. The gold out flow, figure 1, slowed. Part o f the im provem ent resulted from the restrictions on military and other purchases abroad, but part was the result o f rising exports, achieved despite the relatively robust economic expansion in 1963 and 1964. Figure 4 shows quarterly data on the current account balance (in billion o f dollars) from 1960 to the end o f the Bretton W oods system. A fter the increase in the surplus, to 1964, there is a steady decline interrupted b y the recession o f 1969-70. The tem porary surplus o f 1970 was soon replaced by a deficit that eliminated the e f fects o f the surplus; the observations fo r 1972 are on a straight line fitted to the data fo r the second half o f the 1960s. To slow the grow th o f dollar reserves abroad, the United States had to reverse the current ac count balance sufficiently to cover private invest ment abroad, transfers, and other capital flows, not in any particular year, but over time. By this standard, policy can be said to have failed to offset the negative trend in the current ac count balance after 1964. One reason fo r the rising surplus in the U.S. current account balance in the early 1960s is that foreign costs rose relative to U.S. costs. Figure 5 shows the unit labor costs (ULC) for the United States relative to unit labor costs abroad.4 The ULC ratio reaches a trough in 0 are from OECD, Main Economic Indicators, Historical Statistics, 1990. The weights are Federal Reserve trade weights normalized to sum to unity. 1985 is taken as the base year for United States and the trade weighted ULC. MAY/JUNE 1991 70 Figure 4 Current Account Balance Billions of dollars 1960 61 Quarterly Data 62 63 64 65 third quarter 1963, then reverses to reach a local peak in fourth quarter 1968. The current account (figure 4) has a similar movement; al though its local peak is a bit earlier, the balance remains relatively higher until second quarter 1965. The trough o f the current account is also in second quarter 1968. The next swing contin ues the negative relation. The current account rises until early 1970 while the ULC ratio falls. Thereafter, the tw o charts m ove together, fall ing until the end o f Bretton Woods. A sharp decline in the ULC ratio accompanied the decline in the trade balance until 1972. The comparison suggests that until 1970 or 1971, changes in the current account balance are consistent w ith the movements o f relative 41The Federal Reserve uses shares of world trade to set in dividual country weights. The real interest rate index for foreign countries is based on Treasury bills for Canada and the United Kingdom and call money rates for Ger many and Japan. The Federal Reserve trade weights are FEDERAL RESERVE BANK OF ST. LOUIS 66 67 Billions of dollars 68 69 70 71 1972 costs. A fter 1970 the situation changed. The fall in the current account is not the result o f a worsening o f the competitive position o f the United States as reflected in relative costs o f production. The fall in relative costs may have continued to increase the current account bal ance, but their influence was m ore than offset by pressures in the opposite direction. One possible explanation o f the change in 1970-71 is that relative real rates o f return to capital m oved against the dollar. Such m ove ments should be reflected in relative real rates o f interest. Figure 6 shows the ratio o f the ex post U.S. real interest rate to a trade w eighted average o f real interest rates using Federal Reserve trade weights.4 The figure shows the 1 standardized for the four countries to sum to unity. Infla tion is measured by the consumer price index for each country. The ratio shown uses three-month moving averages (not centered) for both series. 71 Figure 5 Ratio of U.S. Unit Labor Costs to Trade-Weighted Unit Labor Costs Quarterly Data 1.36 1.36 1.32 1.32 1.28 1.28 1.24 1.24 1.20 1.20 1.16 1.16 1.12 1.12 1.08 1963 64 65 66 67 ratio o f a three-month, non-centered, moving average o f the U.S. short-term interest rate to a three-month m oving average o f short-term rates abroad. A ll rates have been adjusted fo r infla tion in the particular country. These data show that from 1960 to 1969, U.S. real interest rates rose on average relative to rates abroad. The sharp rise in U.S. rates from 1962 to 1964 contributed to the reduction in the net capital outflow and is reflected in the rise in current account surplus. The movement o f relative interest rates, on average, reinforced the effect o f falling relative costs o f production during this b rief period. A fter 1965, the relative 68 69 1.08 rate o f interest continued to rise, but the rise was too small to reverse the falling current ac count balance. And the increase in relative in terest rates in the United States was apparently too small to prevent the very large capital out flo w in that period. Th e path o f relative interest rates does not explain the collapse o f the cur rent account balance and the large capital out flo w in the early 1970s. Relative interest rates give little evidence o f a grow in g lack o f confidence in the dollar in the late 1960s. A flight from the dollar w ould have pushed real U.S. interest rates above rates in w orld markets to compensate fo r the risk o f MAY/JUNE 1991 72 Figure 6 Ratio of U.S. Real Interest Rate to Trade-Weighted Real Interest Rate 3-Month Moving Average devaluation and a run on the dollar. W ith few exceptions real rates in the United States w ere below rates abroad, on average about 10 per cent below from 1966 to 1972. There is no evi dence o f a sustained rise in U.S. rates. The Federal Reserve pumped out m onetary base to hold dow n interest rates. The rest o f the w orld absorbed the dollar ou tflow w ith little change in real U.S. rates relative to rates abroad. Th e main exception is a spike in 1968-69 that mainly re flects a decline in the w eighted average o f real rates abroad. The data on relative real rates o f interest sug gest that the Federal Reserve made little effo rt to slow or stop the capital flow . During the w in FEDERAL RESERVE BANK OF ST. LOUIS ter o f 1971 relative real rates in the United States fell until M arch along w ith U.S. real rates, despite the large dollar outflow. The rise in rel ative rates in the spring and summer reflects both a decline in foreign rates and a rise in U.S. rates. The gold and foreign exchange markets also give little evidence o f a lack o f confidence lead ing to a flight from the dollar during the 1960s. An exception is the w inter o f 1968 w hen the gold price rose and the tw o-tier market began. By 1969, the gold price in the fre e London mar ket had fallen back to $35.2 per ounce. The same cannot be said fo r the other reserve cur rency, the pound sterling. By 1964, the pound was subject to market pressure to devalue rela 73 tive to gold and the dollar. Repeated attempts to reduce the pressure each succeeded fo r a short time, then failed. Finally in N ovem ber 1967, Bri tain devalued by 14.3 percent (from $2.80 to $2.40). Mem bers o f the old sterling bloc followed. Pressure to devalue sterling occurred against a w orsening problem o f U.S. domestic inflation and a deteriorating relative cost position (figure 5). The Johnson administration’s main efforts to slow price and w age increases w ere limited to exhortation (jawboning). These efforts extended to interest rates. W ith short-term interest rates fixed by the Federal Reserve (figure 2a), rising demand fo r goods and services and anticipations o f higher inflation encouraged increased bor row in g and higher m oney growth. As shown in figure 2b, the annual grow th o f the m onetary base rose in 1965. Base grow th reached the highest rate experienced in the postwar years to that time. Efforts to hold dow n rates paid on time deposits under Regulation Q ceilings added to the pressure on the banks. Depositors drew on balances to purchase securities in the open markets at home or abroad. Early in Decem ber 1965, the Federal Reserve raised the discount rate from 4 to 4-1/2 percent and raised ceiling rates on time deposits. A l though President Johnson criticized the change publicly, the higher rates rem ained in effect. As is often the case, how ever, the increase was too little and too late. Annual grow th o f the m one tary base from the same month a year earlier did not decline until the second half o f 1966, as shown in figure 2b. Inflation rose early in 1966. U.S. interest rates, after adjusting fo r inflation, fell relative to foreign rates (figure 6). Declines in relative unit labor costs and relative consumer prices ended. Reflecting these changes, the nom inal current account balance plunged in the five quarters follow ing the Decem ber 1965 decision, eliminating most o f the increase achieved in the previous six years. The Federal Reserve made a short-lived effort to slow the inflation in 1966. The federal funds rate rose and the grow th o f the m onetary base contracted in the second half o f the year. Re sponding to the less inflationary policy, sensitive measures o f prices, such as the producer price index, reversed direction o f change, falling from 42See Solomon (1982), p. 102. 43Haberler (1965) was one of the first to emphasize the greater importance of the adjustment problem relative to the local peak in third quarter 1966 to a local trough in second quarter 1967. A six-month aver age o f consumer prices fell from a 4 percent annual rate in January 1966 to 1.3 percent in February 1967. Given the upw ard bias in con sumer prices, resulting from the heavy weight on service prices (that are not adjusted fo r pro ductivity and quality changes in inputs and out puts), it appears that the Federal Reserve had stopped the inflation. But, as figures 2a and 2b show, the Federal Reserve did not continue the policy. Federal funds rem ained at 4 percent fo r most o f 1967 despite clear evidence o f recovery. By early 1968, consumer prices w ere rising at a 3 to 4 percent annual rate and, m ore im por tantly fo r the payments problem, rising relative to a w eighted average o f foreign prices. Relative unit labor costs rose sharply. The effects o f changes in relative costs and prices on the U.S. payments position w ere partially hidden at first by a capital in flow from Europe; U.S. banks had began borrow in g from the Eurodollar market. In part, this reflected the rise in relative rates o f interest by 0.25 in the United States (figure 6), in part efforts to circum vent ceilings on time deposits including, at the time, all certificates o f deposit (regardless o f denomination). The result was a payments surplus in 1966, the first since 1957, and repaym ent o f earlier Treasury and Federal Reserve borrow in g from central banks and governm ents.4 2 Confidence in the administration's ability to maintain convertibility into gold or avoid deval uation reached a tem porary low early in 1968. The immediate problem began with a run on gold w hen Britain devalued. The gold pool sold $800 million in Novem ber 1967. The run subsid ed in January, follow ing the announcement that President Johnson had placed new controls on foreign investment by businesses, banks and financial institutions. Demands fo r gold rose again in March. Rumors that the gold pool w ould end and the low cost o f speculating against a price that could fall very little encouraged renew ed speculation. A fter sales o f $400 million on March 14, the London gold pool closed the next day. That marked the end o f the gold pool. The market did not reopen until April. W hen it did, central banks no the liquidity problem, but neither his remonstrance nor others had much effect on official proposals. See also Friedman (1953). MAY/JUNE 1991 74 longer supported the market price. Th ere was now a two-tier market. Private transactors could buy and sell at the market determined price, although the 1934 ban on U.S. citizens’ ownership o f gold remained. Transactions bet w een central banks w ere placed outside the m arket and continued at the $35 price. Also, the governm ents agreed that gold w ould not be sold by the members o f the fo rm er pool to replace any central bank sales to the private market. The central bank governors' communique’ put a positive interpretation on their announcement, repeated their intention to maintain existing parities, and referred to the then forthcom ing agreem ent to establish the SDR. It made no mention o f adjustment o f parities.4 The central 3 banks agreed to "no longer supply gold to the London gold market or any other gold market,” but they hedged their statement to retain the possibility o f buying gold.4 The two-tier agree 4 ment remained in effect until N ovem ber 1973. The free market gold price w ent to $38 per ounce, suggesting that the market w ould have been satisfied by 10 to 15 percent devaluation o f the dollar. For the rest o f the year, the free market price rem ained betw een $38 and $43. Then the price fell back to $35 to absorb an increased supply o f South A frican gold.4 5 For the year 1968 as a whole, the United States had a surplus in the balance o f payments. The reason is that banks continued to b orrow in the Eurodollar market and, follow ing the Soviet invasion o f Czechoslovakia, foreign inves tors purchased U.S. assets.4 These decisions 6 and events produced a payments surplus in 1968 despite a fu rth er decline o f $2 billion to $0.6 billion in the current account surplus. One lasting effect o f the w in ter’s events was the elimination o f the gold reserve requirem ent fo r Federal Reserve notes. On March 12, Con gress abolished the 25 percent gold reserve behind Federal Reserve notes. The legislation rem oved one o f the last links betw een gold and the dollar in the original Federal Reserve Act. The initial requirem ent ratios, or backing, fo r bank reserves and currency had first been re duced, then eliminated fo r bank reserves and, finally, eliminated fo r currency. The stated pur pose was to make the gold stock available to de- fend the $35 price. In fact, central banks did not again convert dollars into gold until 1971. The two-tier system and the decision b y cen tral banks to refrain from converting dollars in to gold had an unanticipated effect on the soon to be created SDRs. The United States had spon sored SDRs and urged their use as a substitute fo r gold in central bank reserves. Since central banks refrained from selling gold, and bought new ly mined gold from South A frica and the Soviet Union, issues o f SDRs served as a substi tute fo r dollars in central bank reserves. Looking back after the fact, it is surprising how small was the loss o f confidence in the dol lar as a reserve currency b efore 1971. Central banks and governm ents p referred to absorb dollars rather than revalue their currencies. Some private speculators purchased gold or other assets, but the purchases w ere not large enough to m ove the equilibrium gold price per sistently above the fixed price until 1970. Meeting after meeting during the last five years o f Bretton W oods mentioned the three problems: confidence, liquidity, and adjustment. Central bank governors, ministers and their staffs gave most o f their attention to liquidity and then to confidence. Aside from a fe w re latively small changes in parities, little was done about adjustment. The attitude o f the IMF is representative o f the period. Their 1969 report mentions adjustment o f par values, follow ing the French devaluation. The discussion reached only one conclusion: the par value system should be retained.4 7 THE END OF BRETTON WOODS Concern about the rising budget deficit, the payments problem and inflation led Congress in June 1968 to accept the Johnson administration’s proposed 10 percent income tax surcharge and, in return, to require the administration to reduce the grow th o f governm ent spending. The response o f the Federal Reserve was almost immediate; they reduced the federal funds rate in an attempt to mix easier m onetary policy w ith tighter fiscal policy. G row th o f the m one tary base declined briefly, then rose to a 7 per cent annual rate o f increase (figure 2b). The six month average rate o f increase o f consumer 44See Solomon (1982), p. 122. 46lbid., p. 105. 45lbid., p. 124. 47See IMF (1969), p. 32. FEDERAL RESERVE BANK OF ST. LOUIS 75 prices rose at a 4 percent rate in 1968 but the annualized rate o f increase was 6.5 percent at the end o f the year. less than 4 percent the follow ing year. Growth o f the m onetary base rose from 4 to 8 percent annual rate in the same period. U.S. consumer prices now began to rise rela tive to a trade w eighted average o f prices abroad (figure 3), reversing the trend decline o f the early 1960s. Unit labor costs rose sharply rela tive to costs abroad (figure 5). Th e trade and payments position deteriorated; real net exports (1982 dollars) fell to -$ 3 0 billion in 1968 and -$ 3 5 billion in 1969 from -$ 1 7 billion in 1967 after a $6 billion surplus in 1964. As shown in Figure 4, the nominal current account balance continued to fall in 1968, reversed briefly dur ing the 1969-70 recession, then resumed its decline. To any outside observer, it must have been clear that the United States did not intend to follow the classical rules or the policies o f a country that intended to maintain a fixed ex change rate. The run from the dollar began. Foreign central banks experienced large in creases in dollar reserves. Germany added $3.1 billion in the first six months o f the year. Ger man money grow th rose from 6.4 percent in 1970 to 12.0 percent in 1971. German consumer prices, which had increased 1.8 percent in 1969 rose 5.3 percent in 1971. Despite rigid exchange controls, Japan could not escape the direct e f fect o f U.S. money grow th through the trade account. Japan’s reserves nearly tripled in the first nine months o f 1971, rising $8.6 billion. The Japanese money stock (MJ rose m ore than 25 percent in 1971. Other countries had qualita tively similar experience. The official settlements measure o f the balance o f payments shows the United States in surplus in 1968 and 1969 fo r the first time in the decade.4 This was misleading, much o f it the 8 proximate result o f Regulation Q ceilings on personal and corporate time deposits. As U.S. interest rates rose above the legal ceiling rates, commercial banks lost time deposits to the Euro dollar market.4 The U.S. banks then borrow ed 9 in the Eurodollar market acquiring many o f the deposits they had lost and some additional funds. The effect was to have a large inflow o f short term capital, $4.3 billion on the official settle ments basis fo r the tw o years. Interest rates fell during the 1970 recession, real rates declined on average relative to rates abroad, and the capital flo w reversed w ith a vengeance. The official settlements deficit o f -$9.8 billion was by far the largest to that time. But this flo w was soon dw arfed by the -$ 3 1 billion outflow in the first three quarters o f 1971.5 0 Figure 1 shows the surge in liabilities to foreign central banks and governments. These liabilities m ore than doubled to $50 billion in 1970, then rose another $11 billion in 1971. The classic response to a capital ou tflow fo r a coun try on a fixed exchange rate is to raise interest rates and reduce m oney growth. The Federal Reserve did the opposite, the federal funds rate fell from a peak o f 9 percent early in 1970 to 48The official settlements balance measures the change in reserve assets minus the change in short- and long-term liabilities to foreign official institutions (central banks or in ternational agencies). 49A Eurodollar is a dollar deposit liability of a European based bank, including European branches of U.S. banks. In 1969, after much delay, Germany had closed the foreign exchange market, allowed the mark to float, then revalued by 9.3 percent on Octo ber 24. W ithin tw o years, three major curren cies—the British pound, the French franc and the German m ark—had been forced to change exchange rates.5 Th e belief that economic 1 stability required exchange rate stability began to erode. The 1971 capital flo w dw arfed previous ex perience. Th e U.S. deficit on capital account was almost $30 billion fo r the full year 1971 and $42 billion fo r the tw o years 1970-71. O f this amount, $40 billion becam e dollar reserves o f other countries. Japan and Germany accumu lated $11 billion each, m ore than half the total, and the United Kingdom acquired nearly $10 billion.5 On May 5, seven European countries 2 closed their foreign exchange markets. Th e Ger man Finance Minister, Schiller, tried to persuade principal European countries to agree to a joint float, but France and Italy opposed. Four days later the mark and the Dutch guilder began to float. Switzerland revalued by 7 percent and Austria by 5 percent.5 Belgium, w ith a split ex3 50See ERP (1972), p. 150. 5 France devalued by 11.1 percent on August 10, 1969. 1 Canada floated its currency against gold in May 1970. 52See IMF (1972), p. 15. 53See Solomon (1982), p. 180. MAY/JUNE 1991 76 change rate, allowed the financial rate to float up. Early in August, faced w ith slow recovery from the recession, rising inflation, a persistent payments deficit, and fifteen months to election day, President Nixon decided to change economic policy. A fter m eeting at Camp David on the w eekend o f August 13 to 15, he ended the fixed exchange rate system by suspending "tem porar ily the convertibility o f the dollar into gold or other reserve assets.” The plan announced on August 15 included much m ore than the suspension o f gold conver tibility. W ages and prices w ere frozen fo r 90 days, allegedly to stop inflation then running at an annual rate o f 3 percent fo r the six months ending in July. Tax credits to increase em ploy ment and investment w ere introduced to in crease demands fo r output and labor and to reduce unemployment below 6 percent. A 10 percent surcharge was put on imports.5 By the 4 end o f August, all major currencies except the French franc floated against the dollar.5 The 5 last remaining tie o f the dollar to gold had been severed, at least tem porarily. Many o f the changes announced on August 15 had been discussed fo r some time in advance. Chairman Burns o f the Federal Reserve had ad vocated a price-wage policy fo r months. John Connolly, the secretary o f the Treasury, favored strong action on trade and inflation. Stein reports that President Nixon and Connolly had agreed in the spring that they w ould impose price and w age controls if foreign demand fo r gold required them to close the gold w in d ow .5 6 The triggering event was renew ed demands fo r gold from France and Britain. On August 8, the press reported that France w ould ask fo r $191 million in gold to make a scheduled repay ment to the IMF. Later in the same week, on August 13, Britain also requested to exchange dollars fo r gold. The combined requests w ere 54More precisely, the surtax was used to raise import duties no higher than their statutory level from which tariff reduc tions had been made. For autos, the increase was, there fore, 6.5 percent (ERP, 1972, p. 148). Shultz and Dam (1977), p. 115 explain that the surcharge was to be used as a bargaining chip to keep other countries from following the U.S. devaluation against gold. U.S. policy was to devalue against gold and other currencies. The surcharge raised the price of U.S. imports, so devalued the dollar against other currencies until they agreed to a formal devaluation. 55France adopted a dual exchange rate with financial tran sactions at a floating rate. 56See Stein (1988), p. 166. Herbert Stein was a member and later chairman of President Nixon’s Council of Economic FEDERAL RESERVE BANK OF ST. LOUIS small in comparison w ith the $12 billion in crease in foreign holdings o f dollars in the first nine months o f 1971. Reports o f demands fo r gold, how ever, generated fears o f a run against the remaining U.S. gold reserve.5 President 7 Nixon and his principal advisers met at Camp David on the w eekend o f August 13 to 15 to adopt the program based on the agreement reached by Nixon and Connolly in the spring.5 8 The earlier policy had been called "steady as you go.” In fact, U.S. m onetary policy had been far from steady in 1970-71. The federal funds rate was driven dow n from 9 percent early in 1970 to 3.7 percent in March 1971, then in creased 5.3 percent in July. During the same period, grow th o f the m onetary base increased from 4 percent to 8 percent. The effect, given policies abroad, was to low er U.S. short-term in terest rates relative to a trade-weighted average o f rate abroad as shown in figure 6.5 The de 9 cline in relative real interest rates ended in March, then reversed but, by March, the dollar's fixed exchange rate was falling, reflecting g ro w ing fears o f devaluation. These fears strength ened as currencies began to float or revalue against the dollar. W hen asset markets opened on Monday August 16, traders greeted the new economic policy enthusiastically. U.S. interest rates fell and stock prices rose. Many o f the foreign ex change markets abroad w ere closed, but in the United States and, later in markets overseas, the dollar depreciated against most currencies. An exception is the Canadian dollar w hich rem ain ed in a narrow band fo r the rest o f the year. The Japanese yen was at the other extreme; it revalued b y about 5 percent initially and rose 10 percent by the end o f September despite much intervention by the Bank o f Japan. Be tw een D ecem ber 1970 and July 1971, the trade w eighted index o f the real U.S. exchange rate Advisers. The decision was told only to George Shultz, director of the Office of Management and Budget and Paul McCracken, chairman of the Council of Economic Advisers. 57See Shultz and Dam (1977), p. 110; Stein (1988), p. 166. 58Arthur Burns, chairman of the Board of Governors, par ticipated in the meeting despite the independence of the Federal Reserve. Burns also participated in the administra tion’s program as chairman of the Committee on Interest and Dividends. 59The local peak in the relative rate is 1.27 in January 1969. By March 1971 the relative rate reached 0.6. 77 Figure 7 Dollar’s Real Exchange Rate Index declined by approximately 3 percent; betw een July and the end o f Decem ber 1971 the rate fell an additional 6.5 percent, so real devaluation fo r the year was approximately 9.4 percent. Figure 7 shows the real exchange rate fo r the dollar against a trade-weighted basket o f cur rencies using Federal Reserve weights. Since there are fe w parity changes p rior to 1971, the real exchange rate reflects mainly relative price changes. Hence, figure 7 fo r the most part dup licates figure 3 until 1971. Thereafter, the tw o charts differ; the real exchange rate reflects both the devaluation o f the nominal exchange rate and the change in relative prices. What Next? The price-wage freeze, the surtax on imports and the floating dollar w ere thought to be tem porary measures. Th ere is no evidence that the administration had developed a long-term pro gram by August 15, but they began to do so. The 1972 Econom ic Report summarizes some o f their thinking by discussing three questions. Should realignment occur through market adjust ment or negotiation? H ow large is the structural or permanent deficit? H ow should the reduction in the U.S. payments deficit be distributed among other trading countries? In practice, the distribution w ould be determ ined by the choice MAY/JUNE 1991 78 o f exchange rates, so the issue was the size o f relative revaluations against the dollar.6 Other 0 countries, particularly France, wanted acknowl edgment by the United States o f its past infla tion in the form o f a devaluation against gold. Perhaps m ore important was the effect on country wealth. A devaluation by the United States raised the value o f country gold stocks w hile a devaluation by other countries reduced the value o f their dollar assets. In the first nine months o f 1971 the U.S. short-term capital ou tflow rose to $23 billion, and the basic balance declined to -$10.2 billion. At an annual rate, these outflows w e re equiva lent to the entire short-term capital ou tflow fo r 1960-69. The ou tflow includes capital flight from the United States in anticipation o f devaluation and a deteriorating current account balance. In part, deterioration reflected the w orsening rela tion in U.S. prices relative to foreign prices. The ratio o f U.S. consumer prices to trade weighted consumer prices in figure 3 shows an increase o f fou r percentage points betw een 1966 and the end o f 1970. Although Treasury Undersecretary Paul Volcker had testified in June that the basic im balance was about $2.5 to $3 billion, Volcker now argued fo r a $13 billion adjustment to reach equilibrium.6 The United States favored a 1 period o f fre e floating and offered to rem ove the 10 percent surcharge on imports if other countries w ould rem ove barriers to trade. In stead central banks intervened, and govern ments imposed exchange controls. Exchange rates w ere not perm itted to adjust freely; new rates w ere negotiated. The Smithsonian Agreement In December, finance ministers and central bank governors met at the Smithsonian Institu tion in Washington to agree on a new set o f exchange rates. Gold was repriced at $38 per ounce, and bands on exchange rates w ere rais ed from 1 percent to 2.25 percent o f central rates. The United States eliminated the 10 per cent surcharge on imports, and the principal countries agreed to discuss reductions o f trade barriers.6 2 60See ERP (1972), p. 149. The agreem ent on devaluation o f the dollar against gold raised, or m ore accurately, con tinued a problem. The official gold price r e mained below the open market price that had now reached $42. Central banks, therefore, had an incentive to convert dollars into gold and sell the gold on the open market. The "solution” was to raise the official price but not require the dollar to be convertible into gold! The new exchange rates revalued the mark by 13.6 percent against the dollar, the yen by 16.9 percent, the pound and the French franc b y 8.6 percent, and most other European cur rencies b y 7.5 to 11.6 percent. Th e Federal Reserve’s calculation showed a trade weighted devaluation o f the dollar o f 6.5 percent against all currencies and 10 percent against the cur rencies o f the Group o f 10. The devaluation was estimated to produce an $8 billion swing in the U.S. trade balance in tw o to three years.6 The 3 IMF estimated the dollar devaluation as 7.9 p er cent o f its form er par value.6 Th e effect o f all 4 the parity changes was to raise the w orld im port prices by about 7.5 percent.6 5 Prior to the agreement, inflexibility and lack o f an adjustment mechanism had been major problems. Surplus countries had been reluctant to revalue because o f the effects o f their exports on domestic employment. The United States had been unwilling to devalue against gold. Many countries had relied on exchange controls to strengthen their currencies. The Smithsonian agreem ent took tw o steps to im prove adjustment. The gold price changed, opening the possibility o f fu rth er changes, and the dollar was devalued against the leading cur rencies. In addition, cross rates o f exchange w ere altered to reflect, partially, the changes in the w orld financial system. Also, the 2-1/4 per cent band on exchange rates perm itted diver gence o f up to 4.5 percent. But, nothing was done to provide an orderly procedure fo r chang ing parities w hen there w ere persistent deficits or surpluses. The system rem ained relatively in flexible and poorly designed fo r the event which it soon faced. sion of trade barriers produced very few changes. (Solomon, 1982), p. 191. 6 See Solomon (1982), p. 192-93; ERP (1972), p. 154. The 1 $13 billion included a $6 billion surplus to provide for len ding to developing countries. 63See Solomon (1982), p. 211. 62Secretary Connally’s initial position included renegotiation of defense costs, but this issue was dropped. The discus “ ibid., p. 22. FEDERAL RESERVE BANK OF ST. LOUIS 64See IMF (1972), p. 38. 79 President Nixon called the agreem ent “the most significant m onetary agreement in the history o f the w orld .”6 In fact, it was a modest 6 agreement that lasted less than fifteen months. W ithin a fe w months stresses reappeared in the international m onetary system. In June, Britain decided to let the pound float. W ithin a month 17 members o f the sterling bloc followed. In the same month, Germany imposed controls on cap ital inflows fo r the first time since the 1950s.6 7 The underlying problem was that U.S. policy remained inconsistent with maintenance o f a fixed exchange rate system. Despite the price controls introduced in August 1971, the rep ort ed rate o f change in consumer prices in 1972 was only 1 percent low er than in 1971. M ore importantly, the future did not look promising. The Federal Reserve made no e ffo rt to restrict m oney growth. Despite a surge in aggregate de mand and industrial production, the federal funds rate was reduced to below 3-1/2 percent early in the year and was held below 5 percent until the N ovem ber election. Growth o f the m onetary base rem ained above 6-1/2 percent, and grow th o f M, increased to 7 percent. Nomi nal imports surged, reflecting the devaluation and the strong economic expansion. For the year as a whole, real net exports w ere -$4 9 billion, a 20 percent rise in a single year and the largest deficit to that time. Th e nominal cur rent account balance rem ained negative, -$5.8 billion, m ore than fou r times the deficit o f the previous year. For a fe w months in the summer and fall, the dollar stabilized. Prices in the United States de clined relative to trade w eighted prices abroad (figure 3) and, until September, U.S. real inter est rates rose relative to trade w eighted real rates abroad (figure 6). The rise in the relative real interest rate during the fall and early w in ter was small, how ever, in relation to the capi tal outflow. In late January 1973, a new foreign exchange crisis began. Italy announced a two-tier exchange market to discourage capital outflows. Sw itzer land floated to reduce the flo w from Italy and to control money growth. Pressure shifted to Germany and Japan. W ithin tw o weeks, Japan floated, and the Europeans closed their foreign exchange markets to halt the inflow o f dollars.6 8 The United States made its last attempt to re tain the par value system. On February 12, the dollar was devalued by 10 percent (to $42.22). Since the United States did not intervene to maintain the new price, the action was more symbol than substance. Secretary Shultz (who had replaced Connally the previous summer) also announced an end to U.S. exchange con trols, including the interest equalization tax and restrictions on foreign loans and investments scheduled fo r Decem ber 1974.6 9 Within a fe w weeks, there was a renew ed flight from the dollar, requiring additional pur chases by foreign central banks under the rules. During the first quarter o f the year, foreign central banks, mainly in the G-10, bought an ad ditional $10 billion. The addition was m ore than 17 percent o f total G-10 foreign exchange bal ances at the end o f 1972.7 The new purchases 0 w ere sufficient to convince most countries to bring the Bretton W oods system to an end. The Europeans agreed on a joint float against the dollar and other currencies. Th e yen had floated earlier. De facto fluctuating exchanges rates became the norm fo r major currencies. W HY BRETTON WOODS FAILED In retrospect, the Bretton W oods system o f fixed but adjustable exchange rates appears to have failed fo r tw o main reasons. First, the sys tem was poorly designed, and the flaws became m ore apparent as tim e passed. Second, the United States did not pursue the monetary pol icy necessary to maintain a fixed exchange rate. On a fe w occasions, interest rates may have been raised to support the exchange rates (or to slow the capital flow ), but m onetary policy con centrated almost exclusively on a variety o f domestic objectives. This was particularly true w hen the climax came in 1970-72. The Flaws in Bretton Woods The designers o f the Bretton W oods system wanted to reduce the role o f gold and make ad- 66See the Wall Street Journal, December 19, 1971. 69See Pauls (1990), p. 897-98. 67Previously Germany used reserve requirements on foreign deposits to reduce capital inflows. In June 1973 sales of German securities to foreigners were prohibited. 70IMF Yearbook (1990). 68The Bundesbank purchased $5 billion in the week ending February 9, 1973. MAY/JUNE 1991 80 justment by deficit and surplus countries more nearly symmetrical. One o f the designers, John Maynard Keynes, believed that w ith fixed but adjustable exchange rates and adjustment by both deficit and surplus countries, fluctuations in economic activity would be reduced; deficit countries w ould not be forced to contract, or w ould contract less, w hen faced with a tem porary loss o f reserves; instead, surplus coun tries w ould lend to deficit countries. In this way, fluctuations in output w ould be damped. There w ere tw o problems w ith this plan fo r adjustment. First, surplus countries had no in centive to adjust, and they w ere generally reluc tant to do so. Keynes had proposed a penalty on surplus countries that accumulated reserves, but this proposal was eliminated early in the developm ent o f the Bretton W oods system.7 Se 1 cond, policymakers could not distinguish a tem porary disequilibrium, to be resolved b y b o r row ing and lending, from a permanent disequil ibrium requiring a change in par values. In practice, the system became m ore rigid w ith the passage o f time. Britain delayed devaluation fo r several years before 1967. France delayed de valuation in 1968. Germany, Japan and other surplus countries delayed revaluations. Japan supported the yen fo r a few w eeks even after August 15, 1971, by intervening sizably to slow the yen's appreciation.7 2 The flaws in the system appeared quickly, al though they w ere not always recognized as such. A starting point fo r the full operation o f the system is 1959, w hen currencies became convertible. By 1968, the dollar was de fa cto inconvertible into gold. Although the Bretton W oods system stumbled through the next several years, foreign central banks and governments, after M arch 1968, w ere discouraged from con verting dollars into gold and did not do so. W hen some tried to convert, in August 1971, the U.S. form alized the restriction that had been in effect fo r m ore than three years by refusing to sell gold. During the 10 years, 1959-68, from the m ove to convertibility to the effective em bargo on U.S. gold, restrictions on trade and payments grew . The United States paid considerable costs to avoid buying supplies abroad fo r its troops in 7 See Meltzer (1988). 1 72The European exchange rate mechanism had much greater flexibility in its early years, and Germany and the FEDERAL RESERVE BANK OF ST. LOUIS Europe and Asia. Much foreign aid was tied to purchases from the United States. Restriction o f capital movements b y Britain, the United States and other countries made the payments system highly illiberal and complex. The ideal o f Bretton W oods was a system o f fixed but adjustable exchange rates among con vertible currencies. During its short life, the goal became increasingly one o f maintaining fixed rates. Adjustment o f exchange rates and conver tibility o f the dollar into gold w e re lost. Presi dent Nixon's August 1971 decision, in this light, should be seen as a choice o f adjustment over gold convertibility. U.S. P olicy The professed principal aims o f U.S. interna tional economic policy included maintaining con vertibility into gold and sustaining the Bretton W oods system. The policy failed in part because it was often short-sighted or w rong, in part be cause the United States placed much m ore w eight on domestic concerns than on the main tenance o f the w orld m onetary system. Throughout the 1960s, France urged, and the United States opposed, a revaluation o f gold. The French argument was correct insofar as it recognized that a devaluation o f the dollar against gold w ould increase the supply o f w orld reserves and reduce dependence on the dollar as a reserve currency. The French w ere correct also in insisting that gold had a long history as an international money, but this argument con fused rather than clarified the issues under dis cussion. Devaluation o f the dollar against gold did not presuppose a change in the Bretton W oods system. That system was based on the dollar, a point that should have been completely clear after March 1968 when, de facto, coun tries could no longer convert dollars into gold. Had the United States agreed to revalue gold in 1965 or shortly after, it seems entirely possible that the many discussions leading to the issuance o f SDRs w ould have been avoided and an ade quate solution found fo r the liquidity problem much earlier. The U.S. policy delayed the solu tion until long past the time w hen it should have been apparent that a solution to the liquid ity problem w ould not sustain the system. Netherlands revalued several times. By the late 1980s, however, countries tried to avoid exchange rate changes. 81 Gold w ould have served as an effective means o f payment betw een central banks, a role that the SDR did not acquire. M ore im portantly fo r the Bretton W oods system, a revalued gold stock, if agreed to before 1968, could have imposed some discipline on the United States to pay in gold. Discipline was lacking once the de facto em bargo on gold was in place after March 1968. Devaluation was not a panacea, how ever. Coun tries w ould have been unlikely to accept the cost o f high U.S. inflation and frequent large devaluations against gold, so the system’s sur vival w ould have required some restriction on U.S. policy. SDRs provided no discipline at all. The arguments against gold revaluation w ere weak. The principal arguments w ere: (1) a de valuation o f the dollar against gold w ould not solve the adjustment problem if other countries devalued against the dollar; (2) an increase in the gold price w ould benefit South Africa or the Soviet Union; (3) the gold standard was too rigid. It is true that if all countries had devalued against the dollar, exchange rates w ould have remained fixed. But, the "liquidity" problem w ould have been solved and perhaps part of the "confidence” problem as well. Countries w ould have taken losses on their dollar reserves, but devaluation o f the dollar w ould have reduc ed the risk that the system w ould collapse w ith a run on the dollar. Simultaneous devaluation w ould have focused attention on the adjustment problem by rem oving the liquidity problem that occupied so much time and attention. Arguments about South A frica and the Soviet Union addressed domestic political concerns. These arguments had no practical relevance. In the end, the revaluation o f gold was not avoid ed. South Africa and the Soviet Union continued to sell gold. The IMF established rules fo r pur chasing South African gold that accepted South Africa's right to sell gold in the market and to m em ber governments. liabilities.7 Th ey w ro te as if they did not want 3 a m ore flexible system o f adjustment; they wanted greater discipline. They talked much m ore about the losses on holdings o f dollars than the gain from a m ore stable, adjustable m onetary system. Although a return to a full gold standard was not the consistent aim o f French policy tow ard the international m one tary system, their proposals and arguments made it easy fo r others to dismiss their argu ments. Neither France nor other governments offered alternative proposals under which U.S. monetary policy w ould be subject to discipline.7 4 Perhaps it was inevitable in the 1960s that any alternative to U.S. policy w ould be dismissed. The U.S. argument that the dollar could not be devalued was, at most, a half truth. The restrictions placed on various types o f transac tions w ere partial devaluations that changed the relative prices o f those transactions. The most obvious example was the interest equalization tax which raised the cost o f borrow in g dollars. Im port-export bank subsidies low ered the cost o f favored U.S. exports. Other restrictions w o rk ed in much the same w ay to selectively devalue the dollar. Some o f the restrictions w ere so short-sighted as to raise doubts about the purpose o f the poli cy and the objectives o f the policymakers. Re strictions on investment abroad by U.S. firms are an obvious example. Absent the restrictions, investment abroad w ould have been higher and the capital ou tflow larger. But, profitable invest ment abroad w ould have produced a return flow , increasing the current account surplus in the future. W hatever short-term gain the re strictions achieved, they had long-term, negative consequences. The problem was not a short term problem o f maintaining the Bretton W oods system fo r a fe w months or years. From a longer perspective the restrictions on invest ment w ere counter-productive. Th ere is little reason to doubt that public opi nion in many countries wanted to avoid a return to the gold standard. The gold standard was w idely view ed as an excessively rigid, 19th cen tury system. Some o f the French, w ho favored a greater role fo r gold, emphasized the "discipline o f gold” and the repaym ent o f dollar and sterling One reason fo r the investment restrictions may have been that policymakers often focused on the basic balance or broader measures such as the official settlements balance. For these bal ances, U.S. long-term investment abroad is equivalent to an im port o f goods and services. Greater attention to the current account balance 73See Rueff (1969). refers to such proposals as endowing “ the creation of their fantasy with perfect foresight, infinite wisdom...” 74There were many other private proposals including pro posals for a world central bank. Haberler (1966), p. 9 MAY/JUNE 1991 82 w ould have shown the steady decline in that balance, thereby concentrating attention on rela tive costs and prices. This focus w ould have posed m ore sharply the basic issue: deflation or devaluation. Reliance on investment controls partially obscured this basic issue as w ell as sacrificing the future fo r small, costly improvemments in the present. The interest equalization tax was no less short sighted than the restrictions on investment. The effect o f the tax was to raise the cost o f trading in U.S. markets, thereby encouraging part o f the financial service industry to m ove abroad. A long-term result was loss o f the export o f some financial services. By far the major fla w in U.S. policy, and the most damaging feature o f the Bretton W oods system, was the failure to prevent U.S. inflation. As the system developed, the United States was able to choose domestic over international goals w henever a choice had to be made. At times, particularly in the early 1960s, U.S. nominal in terest rates w ere kept higher fo r a fe w weeks or months to reduce the capital outflow. But such choices usually w ere reversed if unem ploy ment rose. U.S. policymakers typically chose ex pansion to deflation and used controls o f vari ous kinds to get tem porary reductions in the capital outflow. And, after 1966, policymakers adopted m ore inflationary policies than before. Given the priority placed on em ployment and other domestic goods, such as housing, price stability was ruled out. All o f the responsibility fo r failure does not fall on the United States, Germany, Japan and others pursued export grow th as a holy grail and either made fe w efforts to adjust their ex change rates or none at all. Spokesmen fo r Ger many could point correctly to the inflationary policies o f the United States, and the failure o f the United States to adjust domestic policies so as to honor international commitments, but they w ere less forthright about the adjustment o f countries w ith sustained surpluses that had un dervalued currencies.7 A fter the Bretton W oods 5 experience, Germany changed its policies tow ard adjustment. In the European M onetary System, Germany revalued frequently to keep that 75See Emminger (1967) as an example. Otmar Emminger was a director and later president of the Deutsche Bundesbank. 76See Shultz and Dam (1971), p. 111. Milton Friedman sent a long memo to President-elect Nixon in December 1968 FEDERAL RESERVE BANK OF ST. LOUIS system from experiencing the adjustment p ro blems o f Bretton Woods. It is surprising how little attention was paid to the adjustment problem. The Kennedy, Johnson and Nixon administrations did not propose a permanent solution.7 Each so-called crisis left 6 m ore controls on capital m ovements or other transactions, but these efforts w e re not follow ed by internal discussions o f policies leading to a long-term solution in which controls w ould be rem oved. The Econom ic Reports o f Presidents Kennedy, Johnson and Nixon have lengthy sec tions each year on the international monetary system, but the reports say little about adjust ment. The typical comment was that surplus countries should revalue or expand demand. The explanation fo r neglect o f adjustment is not that such discussions w ere sensitive. Solomon (1982) reports on the many meetings that w e re held during these years. Th ere are pages on p ro posals and negotiations about liquidity, ve ry lit tle discussion o f adjustment. Th e G-10 and the International M onetary Fund are no better.7 7 They, too, avoided the issue or limited their comments to suggestions that surplus countries expand without inflating. Even the devaluations by Britain and France w ere not follow ed by discussions o f the effect o f the devaluation on the United States. As the Bretton W oods system developed, it ac quired some o f the characteristics its designers had hoped to avoid. Major countries w ere reluc tant to change parities. Surplus countries argued that adjustment was the responsibility o f deficit countries. Deficit countries made opposite argu ments, appealing to the need fo r symmetry. The 1960s witnessed the beginning o f efforts to solve international m onetary or economic problem s by coordinating policy actions. In practice, this usually meant that surplus and deficit countries w e re supposed to agree on dif ferent mixes o f m onetary and fiscal policy ac tions. Discussions produced fe w concrete steps. Foreign countries accepted the expansions im plied by the flow s o f U.S. dollars, but they did not systematically reduce governm ent spending or raise taxes to slow their expansions and low er domestic interest rates. And, as discussed earlier, the United States focused mainly on domestic urging a prompt, decisive change in policy. Nothing was done. See Friedman (1988). 77See Solomon (1982), p. 173. 83 objectives. The dialogue about coordination, once started, was hard to stop. It continued into the 1970s and 1980s. Countries that faced a balance o f payments deficit usually favored coordinated action. Countries in surplus usually w ere opposed. The end o f the Bretton W oods system was follow ed by diverse predictions. Some saw fluc tuating exchange rates as a means o f increasing stability or domestic policy. Some w arned about the instability that w ould follow . It is n ow clear that neither was correct. The warnings about the consequences o f the collapse o f Bretton Woods proved to be wrong. The inconvertible dollar continued to function as an international medium o f exchange and store o f value. The ac cumulation o f dollar assets b y foreign central banks and governm ents continued to rise. By the end o f the 1970s, nominal dollar reserves o f the G-10 countries, excluding the United States, had doubled from the level at the end o f 1972. In the next decade, these reserves doubled again to m ore than $300 billion. Central banks and governm ents continued to be m ore willing to acquire additional dollars than to allow the dollar to devalue. The end o f Bretton W oods im proved the ad justment mechanism, but did not quickly elimi nate inflation or inflationary policies. Countries gained the opportunity to pursue independent policies. Inflation differed betw een major cur rencies but both governm ents and some private individuals complained about the variability o f nominal and real exchange rates. During most o f the next 20 years, the principal currencies continued to fluctuate. H ow ever, countries did not float freely. Dirty floating, managed exchange rates, intervention, exchange controls and trade restrictions w ere retained or introduced. Despite these policies and complaints about excessive variability o f ex change rates, there was no interest in a return to Bretton Woods. Economic Report of the President (Government Printing Of fice, various years). Emminger, Otmar. “ Practical Aspects of the Problem of Balance-of-Payments Adjustment,” Journal of Political Economy, Vol. 75 (August 1967, Part II), pp. 512-22. Friedman, Milton. “ The Case for Flexible Exchange Rates,” in M. Friedman, ed., Essays in Positive Economics (Univer sity of Chicago Press, 1953), pp. 157-203. _______ . “A Proposal for Resolving the U.S. Balance of Payments Problem,” in Leo Melamed, ed., The Merits of Flexible Exchange Rates: An Anthology (George Mason University Press, 1988), pp. 429-38. Haberler, Gottfried. Money in the International Economy (Har vard University Press, 1965). _______ . “ The International Payments System: Postwar Trends and Prospects,” in International Payments Problems: A Symposium Sponsored by the American Enterprise In stitute (Washington: American Enterprise Institute, 1965), pp. 1-16. Heller, Walter. New Dimensions of Political Economy (Harvard University Press, 1966). International Monetary Fund. Annual Report, various years. Keynes, J.M. A Tract on Monetary Reform (London: Mac millan, 1923), reprinted as vol. 4 of The Collected Writings of John Maynard Keynes (London: Macmillan, 1971). Meltzer, Allan H. Keynes’s Monetary Theory: A Different Interpretation (Cambridge University Press, 1988). Okun, Arthur M. The Political Economy of Prosperity. (Washington: The Brookings Institution, 1970). Pauls, B. Dianne. “ U.S. Exchange Rate Policy: Bretton Woods to Present,” Federal Reserve Bulletin, vol. 76 (November 1990), pp. 891-908. Rueff, Jacques. “ The Rueff Approach,” in R. Hinshaw, ed., Monetary Reform and the Price of Gold. (The Johns Hopkins Press, 1967), pp. 37-46. Salant, Walter, and others. The United States Balance of Payments in 1968. (Washington: The Brookings Institution, 1963). Schwartz, Anna J. “ The Postwar Institutional Evolution of the International Monetary System,” in A.J. Schwartz, ed., Money in Historical Perspective (University of Chicago Press for the National Bureau of Economic Research, 1987), pp. 333-63. _______ . “ The Performance of the Federal Reserve in Pur suing International Monetary Objectives,” Money and Bank ing: The American Experience (Durrell Foundation, 1991). Shultz, George P., and Kenneth W. Dam. Economic Policy Beyond the Headlines. (Stanford University Press, 1977). Solomon, Robert. The International Monetary System, 1945-1981 (Harper & Row, 1982). Sorensen, Theodore. Kennedy. (Harper & Row, 1965), pp. 405-12. REFERENCES Corden, Max. “ Does the Current Account Matter? The Old View and the New,” unpublished paper (Johns Hopkins University, 1990). Stein, Herbert. Presidential Economics, 2nd rev. ed. (Washington: American Enterprise Institute for Public Policy Research, 1988). The Wall Street Journal, December 19, 1971. MAY/JUNE 1991