The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
FEDERAL RESERVE BANK OF ST. LOUIS DECEMBER 1975 Vol. 57, No. 12 Crowding Out and Its Critics* KEITH M. CARLSON and ROGER W. SPENCER D o e s Government spending displace a near-equal amount of private spending? This notion, popularly known as the “crowding-out” effect of Government expenditures, has recently gained wide-spread at tention at two levels. First, at the policy level, public officials have expressed concern that massive current and projected Federal deficits will have a deleterious effect on private capital expenditures for some time to come. Second, at the academic level, “crowding out” is at least one of the issues which helps to dis tinguish between followers of the two major macroeconomic schools of thought — Keynesians and monetarists. can be described in the context of the standard IS-LM analytic framework. In this framework, which is the cornerstone of most macroeconomics courses taught throughout the western world, the IS curve represents the locus of points (pairs of interest rates and real income) in which the real sector of the economy is in equilibrium, and the LM curve repre sents a similar locus of points for which the demand for money equals the supply. The IS-LM apparatus has distinct limitations, but because of its widespread use as a pedagogical device, it serves a useful func tion in highlighting the issues in the crowding-out controversy.2 This article focuses on “crowding out” from more of an academic than a practical policy point of view. Policy implications can be drawn from this discus sion, but, for the most part, the abstract economic models used in academic circles are not easily adapt able to observable phenomena. Yet the origins of the recent crowding- out controversy at the academic level are traceable to certain empirical results based on U.S. experience. The subject of crowding out is approached by first investigating a number of separate “cases” which pro vide various explanations of how crowding out might occur. Next, the role of stability considerations in the controversy is assessed. Finally, several econometric models are examined to determine what empirical implications they have for the crowding-out issue. New research has been conducted in this area and some old arguments have been revived.1 Many of the developments in the crowding-out controversy To set the stage for the discussion, two matters of a preliminary nature are taken up in this section. First, crowding out is defined for the purposes at hand. Much of the recent discussion of crowding out has been confusing simply because the term has not been carefully defined. Second, since the controversy “The authors acknowledge the helpful comments of James Barth, William Dewald, Dean Dutton, Thomas Havrilesky, Robert Rasche, Paul Smith, Frank Steindl, and William Yohe, none of whom should be held responsible for remaining errors. 1For a survey that includes a discussion of the views of the classical economists on crowding out, see Roger W. Spencer and William P. Yohe, “The ‘Crowding Out’ of Private Ex penditures by Fiscal Policy Actions,” this Review (October 1970), pp. 12-24. Page 2 SOME PRELIMINARIES 2For discussion of the limitations of the IS-LM framework, see Karl Brunner and Allan H. Meltzer, “Monetarism: The Principal Issues, Areas of Agreement and the Work Remain ing,” Monetarism, ed. Jerome L. Stein, (Amsterdam: North Holland Publishing Co., forthcoming), and “Mr. Hicks and the ‘Monetarists,’ ” Econom ica (February 1973), pp. 44-59. F E D E R A L R E S E R V E BANK O F ST. L OUI S has moved through several stages in recent years and has oftentimes involved complex and subtle argu ments, an overview is provided as a guide to the reader. W hat Is Crowding Out? Crowding out generally refers to the economic ef fects of expansionary fiscal actions. If an increase in Government demand, financed by either taxes or debt issuance to the public, fails to stimulate total economic activity, the private sector is said to have been “crowded out” by the Government action. The presumption of a constant money supply insures that the policy action accompanying the increase in Gov ernment demand is fiscal and not monetary. The analysis may be conducted in either real or nominal terms. The crowding-out hypothesis main tains that if prices are held constant, as in typical IS-LM fashion, an increase in real Government de mand financed by real taxes or debt has no lasting effect on real income. Alternatively, crowding out implies that an increase in Government spending, given flexible prices and a constant money supply, has no lasting effect on nominal income. In other words, the steady state Government spending multi plier, under the above conditions, is approximately zero.3 By approximately zero, we mean that increased Government demand may crowd out exactly the same amount of private demand, slightly less, or slightly more. There is complete crowding out if $1 of Government demand displaces $1 of private de mand, partial crowding out if $1 of Government demand displaces less than $1 of private demand, and over crowding out if $1 of Government demand displaces more than $1 of private demand. The in creased Government demand may increase aggregate demand temporarily, permanently, or not at all, as will be explained below. Overview The origins of the recent controversy are traceable primarily to the empirical results published by Ander sen and Jordan in 1968 and supporting studies by Keran in 1969 and 1970.4 These results indicated 3These definitional issues are explored in more detail in the appendix. 4Leonall C. Andersen and Jerry L. Jordan, “Monetary and Fiscal Actions: A Test of Their Relative Importance in Economic Stabilization,” this Review (November 1968), pp. 11-24; Michael W. Keran, “Monetary and Fiscal Influences DECEMBER 1 9 7 5 that nominal crowding out occurs; that is, a change in Federal spending financed by either borrowing or taxes has only a negligible effect on GNP over a period of about a year. These studies did not suggest that expansionary fiscal actions have no effect, but showed instead that the initial effect, which is posi tive, is followed in later quarters by an approximately off-setting negative effect. The response to these empirical results took place at two levels — statistical and theoretical. At the statistical level, the validity of the results was ques tioned. Were proper statistical procedures followed in their derivation?5 On the theoretical level the ques tion was whether or not the results were consistent with what seemed to be the accumulated evidence on certain theoretical propositions.6 Although all the returns regarding the validity of the Andersen-Jordan empirical procedures are not yet in, this article focuses on the theoretical argu ments that have since evolved. The first theoretical argument offered in response to the crowding out concept was an alleged inconsistency between such results and the prevailing estimates of the interest elasticity of the demand for money.7 The critics charged, on the basis of the IS-LM framework, that in order for crowding out to occur, the proponents of these results must be assuming that the demand for money is nearly perfectly interest-inelastic. This allegation meant acceptance of the proposition that the LM curve is essentially vertical. According to the critics, most empirical estimates do not support a zero interest elasticity of money demand. In answer to this charge of inconsistency, Milton Friedman and others argued that the slope of the LM curve was largely irrelevant to the crowding out on Economic Activity - The Historical Evidence” this R e view (November 1969), pp. 5-24, and “Monetary and Fiscal Influences on Economic Activity; The Foreign Experience” this Review (February 1970), pp. 16-28. 5^re J? ' Jera ld Corrigan, The Measurement and Importance of Fiscal Policy Changes,” Federal Reserve Bank of New York Monthly Review (June 1970), pp. 133-45; Richard G. Davis, “How Much Does Money Matter? A Look at Some Recent Evidence,” Federal Reserve Bank of New York Monthly Review (June 1969), pp. 119-31; and Edward M. Gramlich, “The Usefulness of Monetary and Fiscal Policy as Discretionary Stabilization Tools,” Journal o f Money, Credit and Banking (May 1971), pp. 506-32. 6James Tobin, “Friedman’s Theoretical Framework,” Journal o f Political E conom y (September/October 1972), pp. 852-63; Warren L. Smith, “A Neo-Keynesian View of Monetary Pol icy,” Federal Reserve Bank of Boston, Controlling Monetary Aggregates (June 1969), pp. 105-26; and Ronald L. Teigen, “A Critical Look at Monetarist Economics,” this Review (January 1972), pp. 10-25. ‘Tobin, “Friedman’s Theoretical Framework.” Page 3 F E D E R A L R E S E R V E B A N K O F ST. L O U I S discussion.8 In particular, Friedman pointed out the necessity of distinguishing between initial and subse quent effects of fiscal actions. According to Friedman, an “expansionary” fiscal action might first be reflected in a rise in output, but the financing of the deficit would set in motion contractionary forces which could eventually offset the initial stimulative effect.9 In response to the Friedman explanation, the crit ics developed still another argument, again pointing out an alleged inconsistency. This time the critics attempted to demonstrate that the Friedman argu ment, which stemmed from explicit consideration of the Government’s financing requirements, is not con sistent with generally accepted assumptions concern ing stability of the economic system (as represented by the IS-LM apparatus).10 In particular, a debtfinanced increase in Government spending in a world where crowding out occurs does not set in motion a set of forces that will drive the IS-LM model to a new equilibrium once it is disturbed from an initial equilibrium. All of these arguments are reviewed in some detail in this article. Several alternative explanations are offered as to how crowding out might occur regard less of the slope of the LM curve. A number of shortcomings of the recently advanced arguments based on stability analysis are discussed. Finally, returning to the empirical level, the results of some well-known econometric models are examined to see what light they shed on the crowding-out controversy. CROWDING OUT AND THE SLOPE OF THE LM CURVE Until recently, it was suggested by a number of analysts that contemporary monetarists view the sMilton Friedman, “Comments on the Critics,” Journal of Political Econom y (September/October 1972), pp. 906-50; and Karl Brunner and Allan H. Meltzer, “Money, Debt, and Economic Activity,” Journal o f Political E conom y (Septem ber/October 1972), pp. 951-77. 9For further discussion of the role of the Government financ ing constraint, see Spencer and Yohe, “The ‘Crowding Out’ of Private Expenditures;” Carl F. Christ, “A Short-Run Aggregate-Demand Model of the Interdependence and Effects of Monetary and Fiscal Policies with Keynesian and Classical Interest Elasticities,” T he American Econom ic Review (May 1967), pp. 434-43, and “A Simple Macroeconomic Model with a Government Budget Restraint,” Journal o f Politi cal Econom y (lanuary/February 1968), pp. 53-67; and W il liam L. Silber, “Fiscal Policy in IS-LM Analysis; A Correc tion,” Journal o f M oney, C redit and Banking (November 1970), pp. 461-72. I0Alan S. Blinder and Robert M. Solow, “Does Fiscal Policy Matter?” Journal o f Public Econom ics (November 1973), pp. 319-37; and lames Tobin and Willem Buiter, “Long Page 4 DECEMBER 1 9 7 5 vertical LM curve as a requirement for the existence of crowding out. James Tobin, for example, observed that a vertical LM curve leads to the “characteristic monetarist” proposition that “a shift of the IS locus, whether due to fiscal policy or to exogenous change in consumption and investment behavior, cannot alter y” n William Branson, in his popular macroeconomics textbook, noted that T h e m onetarist position is th at the interest elastici ties of the demand for and supply of money are zero, so that the L M curve is vertical. In this case fiscal policy changes the composition, but not the level of national output, while monetary policy, shift ing a vertical L M curve, can change the level of output.12 Similar statements can be found in other texts. This classical case of crowding out is examined in some detail because of its presumed importance in the crowding-out discussion. Following discussion of this classical case, several alternative explanations are offered as to how crowding out can occur in the IS-LM framework, even if the interest elasticity of money demand is not zero. The Classical C ase: A Vertical LM Curve In order for Government spending to stimulate economic activity, it must either foster increases in the money stock (however defined) or increases in the rate at which the existing money stock turns over. Because the former possibility does not involve net debt purchases by the private sector or increases in taxes, there is no reason to think that private spend ing would be crowded out. However, if the money stock does not increase, Government spending must be financed by debt issuance or increased tax rev enue, either of which could result in a reduction in private spending. If private spending is not curbed by such actions, total spending rises, which implies a rise in velocity — the rate at which the money stock turns over. Run Effects of Fiscal and Monetary Policy on Aggregate Demand,” Cowles Foundation Discussion Paper No. 384 (December 13, 1974). 1'Tobin, “Friedman’s Theoretical Framework,” p. 853. 12William H. Branson, M acroeconom ic Theory and Policy (New York: Harper & Row, Publishers, 1972), p. 281. It is of interest to note that Tobin labels the case in which only monetary policy can affect income as characteristically mon etarist and the situation in which both monetary and fiscal policies can alter income as characteristically neo-Keynesian. Branson symmetrically views the vertical LM case as “extreme” monetarist, and the vertical IS case as “extreme” neo-Keynesian (or “fiscalist” ). F E D E R A L R E S E R V E BANK O F ST. L OUI S It is an axiom of classical economics that velocity is virtually constant and cannot be increased by Government actions. In particular, the rise in interest rates, which is associated with the issuance of Gov ernment debt, does not induce the private sector to attempt to hold less money balances because the demand for money is not sensitive to interest rate changes. This idea can be illustrated graphically with the Hicksian IS-LM apparatus in Figure 1. DECEMBER 1 9 7 5 F ig u re 1 Classical Case The LM curve is vertical (drawn for a given price level, P0) in the classical case, reflecting a zero inter est elasticity of the demand for (and supply of) money. Thus, an increase in Government spending which shifts the IS curve to the right can only in crease the interest rate, but does not stimulate veloc ity. Consequently, aggregate demand, as shown in the bottom half of Figure 1, does not shift.13 One or more components of private spending are crowded out by an amount equal to the amount of the Govern ment spending increase. As a result, with aggregate demand failing to shift in response to the increase in Government spending, crowding out occurs in both real and nominal terms. Alternative Cases: Crowding O ut W ithout a Vertical LM Curve Five cases are presented which represent eco nomic situations conducive to Government displace ment of private spending without the requirement of a vertical LM curve. The architects of these frame works range from such disparate figures as the Chi cago economists, Frank Knight and Milton Friedman, to John Maynard Keynes. T h e Keynes C ase: Expectations Effects — John Maynard Keynes in 1936 provided the thrust for the proposition that Government spending does not crowd out private spending in his landmark book, The G en eral Theory o f Em ploym ent, Interest and M oney.14 It is ironic that certain passages in that book provide strong support for the opposite contention. Keynes, throughout his General Theory, was much concerned with expectations and confidence. He did not overlook the possibility, even in those times of relatively small budget deficits, that Government 13Although shown as a straight line, the true spirit of the classical case would be better preserved if aggregate de mand were drawn as a rectangular hyperbola. 14John Maynard Keynes, The General Theory o f Em ploy ment, Interest and Money (New York: Harcourt, Brace and Company, 1936), pp. 119-20. spending could adversely affect the confidence of the private sector in its economic future. W ith the confused psychology which often prevails, the Government programme may, through its effect on ‘confidence’, increase liquidity-preference or di minish the marginal efficiency of capital, which, again, may retard other investment unless measures are taken to offset it.15 An induced increase in liquidity preference, subse quent to an increase in Government spending from Gn to G,, is depicted in the IS-LM framework (see Figure 2) by a leftward shift of the LM curve, and a diminished marginal efficiency of investment schedule is reflected by the subsequent backward shift of the 15Ibid., p. 120. For an algebraic analysis that takes into ac count some of the relevant aspects of this Keynes case, see Richard J. Cebula, “Deficit Spending, Expectations, and Fiscal Policy Effectiveness,” Public Finance (3-4/1973), pp. 362-70. Page 5 F E D E R A L R E S E R V E BAN K O F ST. LO UIS DECEMBER 1 9 7 5 A continued experience with deficits which do not produce sustained recovery, as in this country, or a recent inflation and collapse, as in continental Eu ropean countries, is likely to make a deficit a matter for concern and anxiety. And, if there is disbelief in the benefits of a deficit, then the new money spent by the government may well be more than offset by additional withdrawals of private money which would otherwise be spent. Likewise, if consumer in comes do increase immediately as a result of the def icit, business may anticipate that the increase is tem porary and refrain from long-term commitments.16 T he Knight C ase: A H orizontal I S Curve — This case is constructed on the basis of the writings of Frank Knight.17 The analysis does not do justice to the complex theories of Knight, but is offered as being roughly consistent with the spirit of his theory of capital and interest.18 Though Knight certainly did not conduct his analysis within an IS-LM framework, an attempt is made to translate his ideas into such terms. According to Knight, we should expect no dimin ishing returns from investment. One reason for a nearly perfectly interest-elastic investment function is that the quantity of capital is so large relative to the additions to it that these additions should not be expected to have much of an effect on the yield of capital.18 Another reason, according to Knight, is IS curve to the position denoted as IS (G i). If these shifts in the IS and LM curves result in no change in aggregate demand at the given price level P0, both nominal and real crowding out will occur. However, the actual shift in aggregate demand could be posi tive, negative, or negligible, depending on the rela tive shifts of the IS and LM curves. A number of analysts have recently invoked the Keynes case to explain the sluggishness of capital expenditures in recent years. They, however, are not the first since Keynes to attribute lackluster invest ment plans to stepped-up Government spending. De scribing a situation with some similarities to the present, Daniel Throop Smith observed (in 1939) that: Page 6 16Daniel Throop Smith, “Is Deficit Spending Practical?” Har vard Business Review (Autumn 1939), p. 38. 17No attempt is made to cite all of Knight’s articles on in terest and capital, but a summary is contained in Frank H. Knight, “Capital and Interest,” in Readings in the Theory o f In com e Distribution, The American Economic Association (Philadelphia: The Blakiston Company, 1949), pp. 384417. The Knight case was suggested to the authors by Wil liam Dewald of Ohio State University, but he is absolved of any responsibility for the particular analysis here. 18The difficulty of interpreting Knight’s writing is illustrated by Friedrich A. Lutz, T he Theory o f Interest (Chicago: Aldine Publishing Co., 1968), p. 104, where he intro duces his chapter on Knight as follows: It is not easy to give an exposition of Knight’s theory of capital and interest. Over a number of years Knight devoted many papers to the subject; and, as anyone who ever attempted to work his way through Knight’s theory knows, these writings have passages which are very difficult to understand and also, either apparently or really, contradictory. 19For a discussion of the relationship between stocks and flows in the market for capital goods, see James G. Witte, Jr., “The Microfoundations of the Social Investment Func tion,” T he Journal o f Political E conom y (October 1963), pp. 441-56. To add to the confusion relating to the interpretation of Knight’s writings, it should be noted that Knight did not accept the three-part division of resources into land, labor and capital. His interpretation, rather, was that anyone who has control over productive capacity will employ any or all sources in such a way as to maximize the return for their use. For an analysis that preserves this broad interpretation of capital, see Milton Friedman, Price Theory: A Provisional Text (Chicago: Aldine Publishing Company, 1962), pp. 244-63. F E D E R A L R E S E R V E BAN K O F ST. L OUI S DECEMBER 1 9 7 5 that investment carries with it an investment in knowledge, including research and development. As a result, a declining marginal product of capital is approximately offset by technological advances so that an aggregate investment curve is drawn as nearly horizontal with respect to the yield on capital. When translated into an IS-LM frame of reference, the Knight case introduces an interesting element to the crowding-out controversy. A perfectly flat IS curve (see Figure 3) means that fiscal actions are incapable of shifting the IS curve. An increase in Government spending, for example, absorbs saving and reduces the amount available for private invest ment (any increase in Government spending shows up as a one-for-one displacement of private invest ment). Combining the flat IS curve with the LM curve provides a case where monetary policy domin ates the determination of output. Fiscal actions have no effect on either output or the interest rate.20 It is of interest to note that monetary policy has no effect on the interest rate either, an implication which runs counter to some statements by Knight.21 But because fiscal actions do not shift aggregate de mand for this so-called Knight case, the implication is that both nominal and real crowding out occur.22 T h e Ultrarational C ase: D irect Substitution Effects — Recently, Professors Paul David and John Scadding developed some arguments for crowding out that are derived from an assumption of ultrarationality on the part of households.23 The notion of ultraration ality is based on the assumption that households re gard the corporate and Government sectors as exten sions of themselves — as instruments of their private interests. This fundamental behavorial assumption is offered as an explanation for Denison’s Law — the -°It is surprising that this case has not received more atten tion in the literature, because it is every bit as monetarist as the vertical LM case. For an example of one writer who does mention this case, see Martin Bronfenbrenner, Incom e Distribution Theory (Chicago: Aldine-Atherton, 1971), pp 339-40. However, Bronfenbrenner dismisses it as a long-run case with little short-run significance. - 1See Knight, “Capital and Interest,” p. 406. --Though the Knight case has not been empirically tested, it has implications which are consistent with the results of a number of empirical studies. The Andersen-Jordan results relating changes in GNP to monetary and fiscal actions are consistent with such a case. The inability to find a stable relationship between interest rates and various measures of fiscal action is also consistent. And finally, the stability of real interest rates over time —at least to the extent real rates have been measured — provides indirect evidence in support of the Knight model. -■ ‘Paul A. David and John L. Scadding, “Private Savings: Ultrarationality, Aggregation, and ‘Denison’s Law,’ ” Journal o f Political Econom y (March/April 1974), pp. 225-49. observed stability of the ratio of gross private saving to GNP in the United States.24 The David-Scadding article is of relevance to the crowding-out controversy because of its fiscal policy implications. The assumption of ultrarationality im plies displacement effects of Government spending which the authors call “ex ante crowding out.” They argue that stability of the gross private saving ratio - 4Edward F. Denison, “A Note on Private Saving,” The R e view o f Econom ics and Statistics (August 1958), pp. 261-67. David and Scadding suggest that if Government and cor porate activity simply substitute for, rather than augment, household activity, there should be virtually no change in such broad aggregates as the ratio of gross private saving to GNP. Page 7 F E D E R A L R E S E R V E BAN K O F ST. LOUIS DECEMBER 1 9 7 5 in the face of substantial variation in the Government deficit suggests that private debt and public debt are close substitutes. An extra dollar of Government deficit displaces a dollar of private investment ex penditure because deficit financing is viewed as pub lic investment and substitutes for private investment in that households tend to classify both in terms of future consumption benefits. This case is shown in Figure 4, where an increase in Government spending financed by borrowing induces an offsetting change in private investment so that the IS curve does not shift on balance. Similarly, tax-financed expenditures have a dis placement effect on private consumption since they are viewed in terms of their present consumption benefits and substitute perfectly for private con sumption. With an increase in Government spending for consumption financed by increased taxes, the in crease in taxes reduces private consumption with no effect on private saving. As a result, there is a shift in the composition of output from the private sector to the Government, but there is no shift in aggregate demand. Consequently, with tax-financed Government ex penditures displacing private consumption and Gov ernment bond issues (deficit financing) displacing private debt issues dollar for dollar, there is no way that fiscal actions can affect total demand for goods and services. In the parlance of the IS-LM framework, fiscal actions (defined as either tax- or debt-financed Government expenditures) have no net effect on the IS curve or on aggregate demand, which implies both nominal and real crowding out. Also, for this case, fiscal actions have no influence on interest rates. Whether the David-Scadding ultrarational case is to be taken as a serious explanation of crowding out is an open question. Yet it is important to note the implications of this model, because it represents a departure from the severe restrictions implicit in the IS-LM model. In particular, the IS-LM model allows for no substitution between private spending and public spending; David-Scadding have shown that moving away from these restrictive assumptions acts in the direction of reducing the fiscal policy multipliers. Furthermore, by way of Denison’s Law, they conclude that the evidence leans more toward the extreme of ultrarationality than the extreme of the IS-LM model. T he E xtended IS-LM C ase: Price Flexibility — All cases discussed thus far have not presented any con flicts with respect to the nominal versus real crowding Page 8 not shift. There is, however, another way in which crowding out might occur, reflecting a response of the price level to a step-up in Government spending. This case argues that crowding out is possible even without the assumption that aggregate demand does not shift. The implication for nominal versus real crowding out is ambiguous for this case, however. Robert Rasche constructed a sophisticated version of the IS-LM apparatus which was based primarily on the textbook presentation of Robert Crouch.-5 25Robert H. Rasche, “A Comparative Static Analysis of Some Monetarist Propositions,” this Review (December 1973), pp. 15-23; and Robert L. Crouch, M acroeconomics (New York: Harcourt Rrace Jovanovich, Inc., 1972). F E D E R A L R E S E R V E BANK O F ST. L OUI S The model included wealth in the consumption and money demand functions, a Government budget con straint, and a labor sector, as well as an endogenous price level. According to Rasche’s analysis, an in crease in real Government purchases, financed either by taxes or debt issuance, increases aggregate de mand, and, consequently, the commodity price level. Although there may also be a rise in consumption owing to a presumed positive effect of debt issuance on wealth, there is an offsetting increase in the demand for money associated with such wealth gains (see Figure 5). The rise in the price level reduces private consumption as well as the real supply of money. Together with a decline in the amount of private investment owing to an increase in interest rates, these factors tend to crowd out an amount of real private expenditures equivalent to the increase in Government purchases. Crowding out occurs in this model in real terms, but with a higher price level, crowding out is not likely to occur in nominal terms. DECEMBER 1 9 7 5 Fi gur e 5 Extended IS-LM Case These results lead Rasche to conclude that nominal crowding out requires “extreme” assumptions about the interest elasticity and the wealth elasticity of the demand for real cash balances. It should be pointed out, however, that Rasche, in his manipulation of the model, did not allow for a Keynes expectation effect, an ultrarational direct substitution effect, or a Knight effect, all of which may leave the aggregate demand curve unmoved in response to an initial increase in Government spending. T he Friedm an C ase: Initial vs. Subsequent Effects — Milton Friedman’s role in the crowding-out con troversy was established in a series of articles pub lished in the Journal o f Political Econom y over the period 1970 to 1972.28 Friedman did not rely solely on the IS-LM model as a framework for his analysis, but most of his ideas can be summarized in such a context. Friedman denied emphatically that the mone tarist propositions rested on the shape of the LM locus. Instead, Friedman stressed the continuing ef fects of deficit finance, and a fundamental distinction between stocks and flows. Friedman dealt with a large number of complex issues in his reply to the critics, and it is difficult to determine to what extent he supported the notion of fiscal crowding out. His chief point seems to have -''Friedman, “Comments on the Critics”; “A Theoretical Framework for Monetary Analysis,” Journal o f Political Economy (March/April 1970), pp. 193-238; and “A Mone tary Theory of Nominal Income,” Journal o f Political Econom y (March/April 1971), pp. 323-37. been that the power of monetary actions far surpasses that of fiscal actions, which is similar to but not quite the same as declaring a belief in crowding out. Never theless, he concluded that the expansionary effect of an increase in Government spending by borrowing is likely to be minor. To illustrate the Friedman case, consider Figure 6. The IS curve is drawn quite flat, reflecting Friedman’s statement that “ ‘saving’ and ‘investment’ have to be interpreted much more broadly than neo-Keynesians tend to interpret it, . . .”27 Though Friedman does not emphasize it, this interpretation puts him close to the Knight case, because the implication of more 27Friedman, “Comments on the Critics,” pp. 915. Page 9 F E D E R A L R E S E R V E BAN K O F ST. LOUIS DECEMBER 1 9 7 5 inclusive investment tends to flatten the IS curve and dampen the power of fiscal actions.28 In addition, Friedman indicates that the wealth effects of in creased bond holdings on spending will be minimal, because increases in debt would tend to be offset by an increase in expected tax liabilities. Perhaps an even more important reason to doubt the long-run expansive capacity of increased Govern ment spending is its effect on the future production of goods and services. Friedman notes that debtsupported Government spending leads to a “reduction in the physical volume of assets created because of lowered private productive investment.”29 In other words, potential output in the future will be low ered relative to what it would otherwise be with the transfer of resources from private investment (which generates the future capital stock) to Government spending (which absorbs the capital stock). Apart from these objections to the idea of stimula tive Government actions, an initial shift of the IS curve (see Figure 6) may still be consistent with crowding out over the longer term. For a given LM curve, the relatively flat IS curve, which Friedman apparently envisions, yields a shift of aggregate demand which is very small. In addition, Friedman notes that “the evidences of Government debt are largely in place of evidences of private debt — people hold Treasury bills instead of bills issued by, for example, U.S. Steel.”30 If this statement is given the ultrarational interpreta tion discussed earlier, private expenditure is cut back, thereby offsetting the initial increase in Government spending. Whether such an effect is a partial or com plete offset is not made clear, but if it exists, the IS and aggregate demand curves move back toward their original positions. These are the initial effects of a debt-financed in crease in Government spending, but Friedman goes on to emphasize that subsequent effects will con tinue as long as a deficit exists. In later periods, the IS curve will continue shifting back to the left be cause private expenditures continue to be cut back 28T. Norman Van Cott and Gary Santoni, “Friedman versus Tobin: A Comment,” Journal o f Political Economy (July/ August 1974), pp. 883-85. In this article the authors show that the effect of broadening the interpretation of saving and investment is to make the IS schedule flatter. They demonstrate this by adding the interest rate as an argument in the consumption function, and then showing that the extent to which the IS curve is shifted is unaffected by fiscal actions; only the slope is changed. 29Friedman, “Comments on the Critics,” p. 917. 3°Ibid. Page 10 as Government debt is substituted for private debt. Eventually, the stock of private wealth will be re duced relative to what it otherwise would be because of reduced investment, thereby reinforcing the left ward movement of the IS curve.31 Because Friedman is not clear with regard to the role of commodity prices in his analysis, it is difficult to assess his view of real versus nominal crowding out. It is perhaps best simply to conclude that the im pact of an increase in debt-financed Government spending is veiy small, and that there is little differ ence between the effects of debt- versus tax-financed expenditure. A relatively flat IS curve yields these 31For a recent paper that works out a numerical example of the first round and subsequent effects of a fiscal action in an IS-LM framework, see Laurence H. Meyer, “The Bal ance Sheet Identity, The Government Financing Constraint, and the Crowding-Out Effect,” Journal o f Monetary E co nomics (January 1975), pp. 65-78. F E D E R A L R E S E R V E BAN K O F ST. L O U IS results, and any ultrarational effects would reinforce them. CROWDING OUT AND STABILITY CONSIDERATIONS The Friedman emphasis on the longer-run effects of monetary and fiscal actions prompted two major papers (one by Alan Blinder and Robert Solow and the other by James Tobin and Willem Buiter) that attempted to demonstrate that the crowding-out ef fect of fiscal actions is not consistent with the assump tion of stability of the economic system, as repre sented by the IS-LM model.32 Both of these papers are discussed in this section along with a third — by Karl Brunner and Allan Meltzer — which actually antedates the other two.33 All three models essentially employ comparative static tools to examine a dynamic phenomenon. The Long-Run Balanced Budget Models Blinder and Solow — Recently, Blinder and Solow developed a rigorous theoretical attack on the crowding-out thesis.34 They envisioned three possible levels of crowding out: 1) The Government undertakes activities which would otherwise be provided, on a one-for-one basis, by the private sector. They point out that this sort of crowding out (to the extent it exists) would occur regardless of how the Government spending was financed; 2) Debt issues floated by the Government to fi nance its spending drive up interest rates and crowd out private borrowing; 3) Increases in wealth, derived from the issuance of Government bonds, increase money demand, that is, shift the LM curve leftward sufficiently to negate the rightward shifts of the IS curve. Blinder-Solow constructed an extended version of the IS-LM framework which incorporated consump :1-Blinder and Solow, “Does Fiscal Policy Matter?” and Tobin and Buiter, “Long Run Effects.” :!3Brunner and Meltzer, “Money, Debt, and Economic Activity.” 34For papers criticizing the Blinder-Solow analysis, see Albert Ando, “Some Aspects of Stabilization Policies, The Mone tarist Controversy, and the MPS Model,” International E co nomic Review (October 1974), pp. 541-71; Paul E. Smith, “The Government Budget Constraint, Crowding Out, and Stability of Equilibrium,” unpublished (May 1975); and James R. Barth, James T. Bennett, and Richard H. Sines, “Fiscal Policy and Macroeconomic Activity,” ( Paper pre sented at the Meetings of the Southern Economic Associa tion, New Orleans, Louisiana, November 14, 1975). DECEMBER 1 9 7 5 tion and money demand as functions of wealth, and a Government budget constraint providing for Gov ernment debt interest payments. They adhered to the usual IS-LM customs of treating the price level as fixed and of ignoring the existence of a banking system. Blinder-Solow then attempted to discern the likeli hood of crowding-out phenomena occurring by in vestigating the stability properties of the model. They derived the following theoretical conclusions: 1) if Government spending financed by bond issu ance is contractionary, as (according to Blinder-Solow) monetarists claim, the IS-LM model is unstable; 2) if Government spending financed by bond issu ance is expansive, as neo-Keynesians claim, but less expansive than Government spending financed by money creation, the model is unstable; 3) if Government spending financed by bond issu ance is m ore expansive than Government spending financed by money creation, the model is stable. The unusual result that theoretical stability condi tions imply that bond-financed Government spending is more stimulative than money-financed Government spending comes about because of the inclusion of interest payments on outstanding debt in the Gov ernment budget constraint. For the model to be stable, the budget must be in balance in the long run to ensure unchanging stocks of money and debt. In or der for the budget gap to close after the initial shock of fiscal stimulus, income must rise by a larger amount in the bond-financed case than in the money-financed case. This result follows because higher tax receipts must be induced to offset the increased interest pay ments on the Government debt. Tobin and Buiter — Recently Tobin and Buiter also formulated an IS-LM model for the purpose of ex amining the crowding-out thesis. Although some of the equations differ from those employed by BlinderSolow, the basic assumptions, such as a constant price level, and the methodology, which is marked by the stability requirement of a balanced budget process, are virtually the same.35 Like Blinder-Solow, TobinBuiter utilized more than one variation of the basic IS-LM model, and like Blinder-Solow, they arrived at the conclusion that the stability considerations inher ent in the balanced budget requirement generate a 35Although the bulk of their analysis assumes a constant price level, as does an earlier model on which their paper was based, Tobin-Buiter present one version of the model which employs a variable price level. Page 11 FEDERAL- R E S E R V E BAN K O F ST. LO UIS positive Government spending multiplier. TobinBuiter emphasized that the analysis is conducted for periods in which the economy is less than fully em ployed. Furthermore, that crowding out occurs at full employment is, for them, not a foregone con clusion, in view of a positive fiscal multiplier in their full-employment model. Brunner and M eltzer — Another model has recently been developed which is adaptable to analysis of the crowding-out question. Brunner and Meltzer con structed a model of the economy which differs sig nificantly in orientation from the standard IS-LM model. The Brunner-Meltzer model contains markets for real assets, financial assets, and current output, and permits wealth owners to choose among money, bonds, real capital and current expenditures. In contrast with the Blinder-Solow and basic Tobin-Buiter mod els, the Brunner-Meltzer model permits the price level to be determined endogenously and includes a banking sector. The analysis also features, as do the other models, stability considerations and a Govern ment sector which issues interest-bearing debt. Apparently, these common elements of the models are the elements which lead to the unusual results already noted in the Blinder-Solow model, and which also emerge in the Brunner-Meltzer model. In par ticular, Brunner-Meltzer find that Government spend ing financed by debt issuance is more stimulative than Government spending accompanied by expansionary monetary actions. Such a result is again dictated by the requirement of a balanced budget for long-run equilibrium. Once disturbed by, say, an increase in Government spending, the budget is required to re turn to balance, and the presence of interest pay ments in the budget constraint means that a larger increase in income is required for bond-financing than for money financing. Brunner-Meltzer recognized this obvious discrep ancy between their model results and the historical evidence, particularly as interpreted by monetarists. They note, that their model results imply “that infla tion or deflation can occur without any change in B [the monetary base, which is the prime determinant of the money supply].”36 Brunner and Meltzer take a markedly different view of the causes of inflation outside their model construct and in the context of observable phenomena: “Our analysis of inflation, pre sented at the Universities-National Bureau Confer ence on Secular Inflation, analyzes the issue in more 36Brunner and Meltzer, “Money, Debt, and Economic Activ ity,” p. 973 (bracketed words supplied). Page 12 DECEMBER 1 9 7 5 detail and explains why most inflations or deflations have resulted from changes in money.”37 One must bear in mind that the results of the Brunner-Meltzer model are predicated on: (1) the absence of money illusion (in the usual sense), but the existence of a possible wealth illusion by way of incomplete discounting of future tax liabilities; (2 ) the requirement of a balanced budget; (3 ) a fixed capital stock (Blinder-Solow, in contrast, present a variation of their model in which the capital stock is permitted to grow); (4 ) no labor sector ( to facilitate changes in output in lieu of the absence of a changing capital stock); and (5 ) the presumption that asset prices respond more strongly to an increase in Govern ment debt than to an increase in the monetary base.38 Shortcomings of the Balanced-Budget Models The recent attack on the crowding-out thesis by way of stability analysis introduces a new element into the controversy. There are several reasons to to question the implications of these models of the economy which indicate that crowding out is not con sistent with model stability. Treatm ent o f Price L ev el C hanges — The BlinderSolow model and the basic Tobin-Buiter model, which are somewhat sophisticated versions of the standard IS-LM apparatus, permit no role for price level changes.3” Considering world-wide economic devel opments over the past decade, one must question the relevance of so-called “structural” models which omit the existence of inflationary pressures and inflationary expectations. Moreover, an important channel through which crowding out might occur is closed off when price level changes are forbidden to emerge. Blinder-Solow recognized this deficiency of their model to some extent, as indicated by their acknowl edgement that the fiscal policy multiplier would be lowered in several ways by the inclusion of an endogenously-determined price level: (1 ) higher prices lower the real value of the money stock and shift the 37Ibid. :!8The last-mentioned item is particularly critical for the Brunner-Meltzer results. Whereas asset prices can be ex pected to respond in a positive manner to increases in the monetary base, there is ambiguity in the response of asset prices to the issuance of Government debt. A positive wealth effect ( given incomplete discounting of future tax liabilities) must outweigh a negative substitution effect (caused by Government debt competing in asset markets with private debt) for the Brunner-Meltzer results to hold. 39The Brunner-Meltzer model permits price level flexibility, but excludes a labor sector, which presumably plays an important part in realistic attempts to capture the economic structure. DECEMBER 1 9 7 5 F E D E R A L R E S E R V E BANK O F ST. L OUI S LM curve to the left; (2) higher prices reduce real wealth, and thus consumption, shifting the IS curve to the left; (3 ) progressive taxes combined with in flation increase the real yield of the tax system, which also tends to shift the IS curve leftward; (4 ) a rising price level depresses exports and induces imports in an open economy, which again pushes the IS curve to the left.40 Blinder-Solow maintained that although the fiscal multiplier will be less than before with the inclusion of price level changes, the sign of the multiplier will remain positive. Because it is their view that the crowding-out hypothesis requires the fiscal multiplier to be negative, the authors considered only the sign of the coefficient to be at issue. This, however, is a gross exaggeration. To our knowledge, there have been no claims that the crowding-out hypothesis requires that a dollar of Government spending, unsupported by monetary expansion, must reduce private spend ing by m ore than a dollar, which is the implication of a negative fiscal policy multiplier.41 Crowding out of the private sector occurs not only when $1 of Gov ernment spending reduces private spending by $1 (a multiplier of zero), but when $1 of Government spending reduces private spending by 50 cents (a multiplier of 0.50). Crowding out, then, is a matter of degree rather than of absolute magnitudes. A negative multiplier is not a necessary condition for crowding out. And the omission of changing price levels in various IS-LM models contributes to the likelihood that crowding out tendencies will not emerge. B alanced Budget Equilibrium — The three models under consideration show that in order for the budget to be balanced, and for the model to be in long-run equilibrium, the fiscal policy multiplier must be posi tive. A full equilibrium requires that the levels of stocks and flows be unchanging. But the question re mains, how does such a formal analysis contribute to an explanation of the empirical results that imply crowding out occurs? Tobin-Buiter made two significant points in this connection. First, they questioned the ability of eco nomic analysis — presumably, as incorporated in ab stract models — to track changing economic variables to some logical end. “The trouble with such discus 40Blinder-Solow added this final price effect in “Analytical Foundations of Fiscal Policy,” in Alan S. Blinder, Robert M. Solow, et al., The Econom ics o f Public Finance (W ash ington, D.C.: The Brookings Institution, 1974), p. 47. 41It should be pointed out that various econometric models indeed have uncovered negative fiscal multipliers (see p.14 of this article). sions, including this one, is that a long run constructed to track the ultimate consequences of anything is a never-never land. For that abstraction we apologize in advance.”42 If one is really interested in tracking changes in economic variables over time, the better approach would be to construct dynamic models rather than comparative static models. Second, Tobin-Buiter questioned the stability re quirements (including a balanced budget) associated with the IS-LM investigations into the crowding-out controversy. Their concluding remarks were: Finally, we observe again that it is disturbing that the qualitative properties of models — the signs of important system-wide multipliers, the stability of equilibria — can turn on relatively small changes of specification or on small differences in values of coefficients. W e do not feel entitled to use the ‘correspondence principle’ assumption of stability to derive restrictions on structural equations and pa rameters. T h ere is n o d iv in e g u a ra n tee th a t th e ec o n o m ic system is sta b le.*3 The economic system may be stable in the sense that the U.S. economy has not exploded, but it is a long jump from that sort of stability to one which requires stock-flow equilibrium including a balanced budget. Indeed, the budget of the U.S. Government has been in deficit in eleven of the past fifteen years. The stock-flow equilibrium models discussed here, then, are basically empty of empirical content. Al though there may have been periods in which some of the relevant flows were approximately in balance, one would be hard pressed to uncover data points corresponding to periods of unchanging stocks. With out the necessary data and a translation of the ab stract models in a form which is testable, it is impos sible to confirm or refute the hypotheses associated with these stock-flow equilibrium models. Fiscal vs. M onetary Stimulus — The underlying as sumptions and stability requirements of the models in question combine to produce a most curious result: Government spending financed by debt issuance is more expansionary than Government spending ac companied by money creation. The expansionary ef fect is summarized in terms of real output in the Blinder-Solow model and prices in the Brunner-Meltzer model. These theoretical implications run contrary to vir tually every investigation conducted into the impacts of fiscal and monetary policy actions on economic 42Tobin and Buiter, “Long Run Effects,” p. 1. 43Ibid., p. 42 (italics supplied). Page 13 F E D E R A L R E S E R V E BAN K O F ST. LO UI S activity. None of the architects of these models at tempted to reconcile the model implications with the mass of empirical studies contradicting them. Brunner and Meltzer acknowledged this discrep ancy. However, they offered no explanation for the fact that even though their model implies that bondfinanced Government spending is more inflationary than money-financed spending, their own empirical studies indicate just the opposite.44 One is led to conclude that manipulation of these theoretical mod els constitutes an interesting academic exercise, but contributes little of practical significance to the crowd ing out controversy. With empirical considerations coming to the fore, the discussion now turns to the econometric literature to determine what evidence that approach has brought to bear on the issue of crowding out. ECONOMETRIC MODELS AND CROWDING OUT In a recent study of a number of econometric models, Gary Fromm and Lawrence Klein published simulation results showing the implied Government expenditure and tax mulitpliers for these models.45 The results showed long-run Government spending multipliers ranging from about 1 to 5 when measured in terms of impact on current dollar GNP.40 How ever, the majority of the large models surveyed re vealed that crowding out did occur in real terms over time. Some indicated $1 of Government spending for goods and services crowded out even more than $1 of private spending. For example, the Wharton Mark III Model yielded a multiplier of minus 3 after forty quarters, and the Bureau of Economic Analysis (U.S. Department of Commerce) Model gave a real Government spending l4Karl Brunner, Michele Fratianni, Jerry L. Jordan, Allan H. Meltzer, and Manfred J. Neumann, “Fiscal and Monetary Policies in Moderate Inflation: Case Studies of Three Coun tries,” Journal o f Money, C redit and Banking (February 1973), pp. 313-53. 45Gary Fromm and Lawrence R. Klein, “A Comparison of Eleven Econometric Models of the United States,” T he American Econom ic R eview (May 1973), pp. 385-93. These models, unlike the IS-LM abstractions discussed earlier, were not forced to a full stock-flow equilibrium. " ’Blinder-Solow cited these results as attesting to the absence of crowding out in large income-expenditure models. Ac knowledging the nonexistence of Government budget con straints in the models, they added that despite this de ficiency, “All we can do now is render a verdict on the basis of the evidence already in.” They ignored the real crowding-out results implied by the econometric models, which is surprising, in that their own model emphasized the crowding-out issue in real terms. See Blinder and Solow, “Analytical Foundations,” p. 78. Page 14 DECEMBER 1 9 7 5 multiplier over the same time period of minus 23. These results go well beyond monetarists’ contentions that complete crowding out gives a multiplier of ap proximately zero, though these results are less than clear on the issue of nominal crowding out. The Fromm-Klein survey of the empirical results suggested that crowding out typically occurred be cause of a rising price level, capacity constraints, and rising nominal interest rates. These results are con sistent with those implied by the extended IS-LM case described above, and do not necessarily cor roborate crowding out of the nonshifting aggregate demand variety, that is, those cases which imply that crowding out occurs because fiscal actions are off set by other components of aggregate demand. However, Fromm-Klein recognized that the model simulations produced evidence not in accord with the usual standard Keynesian presumption of positive Government spending multipliers: Conventional textbook expositions generally depict real expenditure multipliers approaching positive asymptotes. In fact, most of the models here show such multipliers reaching a peak in two or three years and then declining thereafter in fluctuating paths. At the end of five to ten years, some of the models show that continued sustained fiscal stimulus has ever-increasing p e r v e r s e im pacts.47 Klein suggested elsewhere that perhaps these new estimates of the fiscal multiplier are not as damaging to the Keynesian position as they initially appear.48 After all, it takes a considerable length of time in some of the models for the Government spending multiplier to approach zero or turn negative, and policymakers historically have shown little concern for the long run. We would only add that this argu ment reflects the progression of the debate on crowd ing out from “Does it exist?” to “What is the time period?” As far as small models are concerned, the monetarist model of the Federal Beserve Bank of St. Louis set off much of the current controversy. Fiscal crowding out emerges in the reduced form equations published in the St. Louis R eview only after a period of time, even though it is a much shorter period of time than that of the large income-expenditure models, and it occurs in nominal terms rather than in just real terms. Government spending, as measured by high-employment expenditures, exercises a relatively strong in 47Fromm and Klein, “A Comparison,” p. 393 (italics supplied). 48See Lawrence R. Klein, “Commentary on ‘The State of the Monetarist Debate,’ ” this R eview (September 1973), pp. 9-12. F E D E R A L R E S E R V E BAN K O F ST. L OUI S fluence on GNP (assuming a constant change in the money supply) in the current quarter and the next quarter, but is approximately offset within a year’s time. These results, which are confirmed by regression analysis employing data through mid-1975, should not be interpreted to suggest that “Government spending doesn’t matter”; it matters very much over a certain period. Moreover, if Government spending were to accelerate or decelerate rapidly rather than be held to a steady rate of change, the impact on GNP would be considerable. The chief reason that these reduced form results are of interest is that they do not follow from a structural model that constrains the channels of trans mission from fiscal actions to economic activity. Gov ernment expenditures cover a wide range of activities, some of which substitute for private consumption and investment, and others which serve as substitutes or complements to private factors of production.49 With such diverse effects, any model which restricts the transmission of fiscal actions to income and/or interest rate channels, runs the risk of missing the full effects of Government interaction with the private sector.50 The St. Louis results certainly do not do justice to the measurement of the effects of the complexities of the Government spending process, but they serve the function of questioning the results from models which restrict the operation of fiscal actions via fixed channels. SUMMARY AND CONCLUSIONS This article has surveyed the recent literature on the subject of the crowding-out effect of fiscal actions. Crowding out was defined as a steady state Govern ment spending multiplier of near zero, a definition which was extended to differentiate the terms “nom inal” and “real” crowding out. 49We, like most other analysts, have had little to say about the effect of fiscal actions on aggregate supply. For an attempt to enrich standard macroeconomic analysis with such considerations, see Kenneth J. Arrow and Mordecai Kurz, Public Investment, the Rate o f Return, and Optimal Fiscal Policy (Baltimore: The Johns Hopkins Press, 1970); and Lowell E. Galloway and Paul E. Smith, “The Govern ment Budget Constraint and Aggregate Supply,” (Paper presented at the Meetings of the Southern Economic Asso ciation, New Orleans, Louisiana, November 14, 1975). 50See R. L. Basmann, “Remarks Concerning the Application of Exact Finite Sample Distribution Functions for GCL Estimators in Econometric Statistical Inference,” Journal o f the American Statistical Association (December 1963), p. 944, where he says: . . . the entire burden of statistical inference in econo metric simultaneous equations models falls on the un DECEMBER 1 9 7 5 This survey indicates that the controversy has taken place on two fronts — theoretical and empirical. First, the theoretical literature has developed primarily with reference to the IS-LM model or modifications thereof. Several cases were examined which serve as candi dates providing theoretical support for the crowdingout hypothesis. In addition, the role of stability con ditions in the crowding-out controversy was examined. In general, the conclusion was that stability considera tions are of limited relevance with respect to the ac ceptance or rejection of the crowding-out hypothesis. The empirical literature, on the other hand, has taken the form of simulations of Government actions and has yielded results that show signs of being con sistent with the crowding-out hypothesis. This crowd ing out tends to be very slow in developing, however, and occurs in real rather than nominal terms. The St. Louis results still stand out relative to the large econometric models in that crowding out occurs more quickly and also in nominal terms. As a result of this survey, it is clear that the crowding-out controversy continues to exist. Appar ently these issues will not approach resolution until additional structural models are developed and tested. The Keynesians have developed many models, but these models have not been tested as interdependent units.51 Monetarists, on the other hand, have not of fered structural models to go along with their reduced form results.52 Such a turn toward hypothesis testing could lead toward a resolution of the issues in the crowding-out controversy. Although the controversy has been explored in this article primarily on a theo retical level, the implications of these issues for practi cal matters of stabilization policy are of great significance. constrained estimates and test statistics associated with the reduced-fonn, at least, if empirical confirmation of the underlying economic postulates is the goal aimed at. Whenever the unconstrained reduced-form statistics are judged to be in good agreement with the propo sitions (theorems) deduced from the underlying eco nomic postulates, then do the structural estimates emerge as sound and convenient summaries of that part of the sample statistical information which is relevant to the numerical values of structural pa rameters, but generally not otherwise. 51See Keith M. Carlson, “Monetary and Fiscal Actions in Macroeconomic Models,” this Review (January 1974), pp. 8-18. A suggested testing of models as interdependent units requires that the model be specified in structural form, but the testing of the model should focus on the reduced form. For further discussion of this approach, see James L. Murphy, Introductory Econom etrics (Homewood, Illinois: Richard D. Irwin, Inc., 1973). 52For recent efforts in this direction, however, see Leonall C. Andersen, “A Monetary Model of Nominal Income Determination,” this Review (June 1975), pp. 9-19. Page 15 F E D E R A L R E S E R V E BAN K O F ST. L OUI S DECEMBER 1 9 7 5 APPENDIX For purposes of definition consider the accompanying Figure, panel (A ), which is a representation of the m arket for total output of goods and services. The intersection of aggregate supply (ASo) and demand (ADo) determine the equilibrium level of output, Xo, and the price, Po, at w hich it will b e sold. L abel this intersection as point A and interpret it as an initial equilibrium. Now, introduce an expansionary fiscal action like increased Government demand for goods and services financed by sales of Government debt to the public. Assume that the net effect of increased Government demand and the issuance of debt is an increased de mand for goods and services, as indicated by the shift of the demand curve to A D i. Further, suppose that the expanded Government sector adversely affects efficiency and productive capacity, resulting in a shift of the supply curve to ASi. If the new equilibrium occurs anywhere on the vertical line through point A, say at point B, we say that re a l crowding out has occurred. T h at is, increased real Government spending has been completely offset by a decline in real private spending. Consider now Panel ( B ) in the Figure. T h e curved line drawn through point A is a rectangular hyperbola indicating that P times X , which is defined as the nominal value of total output (th at is, G N P ), is con stant and equal to Po Xo. In other words, there is an infinite number of combinations of P and X , besides Po and Xo, which would give the same dollar value of total output as at point A. Suppose that in response to an expansionary fiscal action, aggregate demand and aggregate supply shift in various directions (d e pending on the assumptions m ade) and the new equilibrium settles on the curved line, say at point B or C. Under these conditions, n om in al crowding out is said to occur. T hat is, an increase in Government spending has been offset by a decline in the dollar amount of spending by the private sector. Page 16 This distinction betw een nominal and real crowd ing out is important because clearly one does not im ply the other. This is shown in Panel (C ) which combines the definitions of real and nominal crowd ing out from Panel (A ) and ( B ) . T h e solid lines are not demand and supply curves, b u t are the loci of points defining real and nominal crowding out. Note that the lines are now drawn as the midpoint of a shaded band. This is done to reflect the crowdingout hypothesis; th at is, an increase in Government demand, not supported by monetary expansion, re sults in a steady state income multiplier of a p p ro x i m ately zero. T h e middle of these bands represents those points at which $1 of Government spending crowds out exactly $1 of private spending. T h e shad ing to the right of either line describes th at area in which partial crowding out (a multiplier betw een 0 and - ) - l ) occurs; the shading to the left of either line describes that area in w hich over crowding out (a multiplier betw een 0 and —1) occurs. O f course, it is possible that a dollar of Government spending might crowd out more than two dollars of private spending, resulting in a multiplier of less than —1 and an equilib rium point to the left of either of the bands. Various combinations of real and nominal crowd ing out are possible, given an expansionary fiscal ac tion. For example, at point A, there is partial nominal and partial real crowding out. At point B, there is partial nominal, but over real crowding out and so on for other combinations around the intersection of the two bands. A t some point outside this area, such as point E , there is partial real crowding out, b u t a com plete absence of any sort of nominal crowding out. It is clear that a com plete analysis of the fiscal process requires an assessment of both the demand and sup ply factors involved in order to describe accurately the extent to which nominal and real crowding out m ight occur. F E D E R A L R E S E R V E B A N K O F ST . L O U I S DECEMBER 1 9 7 5 Definitions of Crowding Out (A) (B) Real Crowding Out Nominal Crowding Out (C) Sum mary Page 17 The Origin and Impact of Inflation Remarks by DARRYL R. FRANCIS, President, Federal Reserve Rank of St. Louis Refore the Joint Seminar of The Canadian Council of Financial Analysts and The Toronto Council of Financial Analysts Toronto, Canada, November 18, 1975 J t IS A pleasure to be here in Toronto and to share with you my views on inflation. This is a subject whose popularity has fluctuated with cyclical fluctua tions in business activity; it is debated during up swings only to recede into oblivion during down swings. Yet, in my opinion, it is a subject which should be analyzed at all times since it is during downswings that the seeds of inflation are sown. You have suggested that I speak on the monetarist view of inflation. While the framework within which I analyze the causes and consequences of inflation is of the monetarist variety, I think I should mention that what I consider most important does not neces sarily represent the views of all monetarists. In order to put things into perspective, I should like to outline this framework of analysis. THE FRAMEWORK An increase in the total money stock, when it is not accompanied by a similar increase in output, has a predictable effect on behavior. Individuals will at tempt to divest themselves of what they consider to be their excess money balances by bidding for other, nonmoney assets. As the prices of these assets rise, output is stimulated. But such increases in output are limited by the growth of resources. Expansion of the money stock produces only a transitory increase in production, while it leads to a permanent rise in the rate of increase of prices. Evidence confirming these results is not difficult to find; rates of growth of money and rates of increase in the price level closely parallel each other when viewed as long-term trends. A great deal of evidence has been amassed showing that an increase above the trend growth of money which persists for at least two quarters will lead to a Page 18 rise in the rate of output growth which is quite short lived. However, as the rate of production returns to its trend level, the rate of inflation increases. We have observed a symmetrical situation for declines in the rate of money growth. Such declines create transitory recessions that are replaced by lower inflation rates in six to eight quarters. Despite many arguments to the contrary, it is clear that central banks can control the money supply within a very narrow range over a time period of a quarter or more. But if we accept the above relation ship between money supply and the price level, why has the money stock been allowed to grow in such a way as to produce persistent and accelerating infla tion punctuated by occasional recessions? Have cen tral banks produced this growth pattern through some nefarious design? Have they merely been incompe tent? I, for one, believe that neither is the case and that we must look to our political and social aspira tions for the root causes of the economic dilemma upon whose horns we sit so very uncomfortably. In doing this, I shall confine my observations to the American experience, simply because I am most fa miliar with the trials of the U.S. economy. I am quite sure, however, that parallels can be drawn for Canada and many other Western industrialized na tions which face the same problems of inflation and unemployment. EXPANDING GOVERNMENT SECTOR: HAS ANYTHING BEEN ACCOMPLISHED? For many years, Government spending and the size of the Government sector have expanded at an in creasing rate. Since 1950 total annual Government ex- DECEMBER 1 9 7 5 F E D E R A L R E S E R V E BAN K O F ST. L OUI S penditures have risen by about $454 billion, with $328 billion of that having been recorded in the past ten years. This growth was spurred by an underlying philosophy which contends that greater direct Govern ment activity is the best way, if not the only way, to achieve certain economic and social goals. So let us consider the claims of those who espouse this philos ophy and examine their validity. Has this spending accomplished what it set out to do? Was it indeed the “best” way? And finally, has it had other consequences, too important to be termed merely “side effects,” which have imposed high costs on us all? Economic Stability One of the oldest arguments in favor of increased Governmental incursion into economic life holds that fiscal policy is the proper, indeed the necessary, tool to stimulate the economy and combat unemployment. In addition to the automatic stabilizing effects of tax and transfer payment policy, it has been alleged that the Government should introduce significant spending efforts when the activity of the private sector is inade quate for full employment, however defined. And it is argued that this spending should engender deficits, since financing through higher taxes would reduce private purchasing power and frustrate the attempt to expand total demand. Historically, Government deficit spending has had no stimulative effects except insofar as it was accom panied by monetary expansion. Thus the stimulation desired could have been accomplished directly through monetary expansion without the Government encroachment into the private sector that is inherent in expansive fiscal policy. More important, we know that the fiscal stimulus is only transitory — that the output effects of excessive money growth are quickly dissipated and that the only lasting result is ever ag gravated inflation. Consider our actual performance. Have we reduced fluctuations in output and employ ment through the wide use of fiscal deficits and sur pluses? Obviously the answer is no. Since the inception of these policies in the early 1930s, the frequency and magnitude of economic fluctuations have not differed significantly from those prior to that period. Fuller Utilization of Labor Resources A second popular argument, and on the surface a very persuasive one, states that it is the proper func tion of Government to employ those resources, par ticularly labor, which the private sector is unwilling to employ. Presumably, the whole society benefits from such programs at no cost, since additional production is being provided by those who were previously con tributing nothing. This is a seductive argument which merits careful examination. Surely we must agree that private enterprise will always take advantage of the opportunity to employ resources which it expects to use profitably. When some resources are not so em ployed, it means only that their services are not worth the price attached to them. For the cause of this situation, we must again look to the influence of Government. Hedged in as we have become by laws requiring the payment of minimum wages and “equal pay for equal work,” we have seen more and more of the labor force become unemploy able. And when the Government puts them to work, one basic result is the same. To the extent that these people are being paid more than the market decrees, there is a real transfer of wealth to them from the rest of society. Real output may be greater, but much of the increase in their welfare comes not from their new productivity but from the rest of us. To gauge the accomplishments of these policies, whatever their redistributive effects, we need only to look at what has occurred. In the face of many jobcreation programs, we find that output growth has risen at approximately a constant trend rate since 1946, irrespective of the rate of Government spending. And in the same period, unemployment fluctuated around an average of 4.9 percent until its recent increase. Satisfaction of Social Needs An argument of more recent vintage maintains that the goods and services provided by the private sector in response to society’s demands do not respond to the so-called “true needs” of society. It follows from this that the Government should divert resources to the satisfaction of these needs. More and more programs have been enacted in areas ranging from health care to cultural pursuits. Whether they have increased our welfare is highly questionable. We have obtained these services only by sacrificing other things we would have chosen for ourselves. But in their efforts to make it appear that there is indeed such a thing as a free lunch, our elected officials have increased Gov ernment expenditures without attempting a corres ponding rise in taxes. As a result, monetary growth and inflation have provided the means of transferring control of resources from private hands into the hands of bureaucrats who, it would seem, know our needs better than we ourselves do. Page 19 F E D E R A L R E S E R V E BAN K O F ST. L OUI S More Equal Distribution of Income Finally, implicit in all the arguments of the advo cates of interference is the assumption that an ex panded Government role in economic activity will, and should, redistribute income in the direction of some notion of greater equality. Whether this redis tribution is indeed desirable is an argument which has probably existed since the first two humans met. I will not attempt to make any enlightening contributions to that debate. It is fair to ask, however, what has been achieved. In spite of the expanding role of Govern ment activity since World War II, the distribution of income has changed very little. The income group representing the lowest twenty percent received 5 percent of total income in 1947 and 5.5 percent in 1971, while the share of the highest 20 percent fell from 43 percent in 1947 to 41.6 percent in 1971. This can hardly be considered a significant accomplish ment, expecially in view of the costs incurred. NEGLECTED CONSIDERATIONS These proposals to improve our socio-economic wel fare have, through design or through ignorance, overlooked the problem of financing the additional expenditures. The basic issue in the financing of Gov ernment programs is that resources have to be trans ferred from one sector of the economy to another. This can be accomplished in only three ways. One is to tax current private consumption and investment — that is, to increase taxes. The second is to tax future private consumption by incurring a deficit and selling Govern ment securities to the private sector. This method moves resources immediately by reducing the pur chasing power of security buyers only, but ultimately spreads the burden to all taxpayers when the securi ties must be redeemed. And the third is to finance the deficit by indirectly selling securities to the central bank which buys them with newly created money. Inflationary Financing When deficits are financed by the sale of Govern ment securities, the attendant additions to the demand for credit must exert upward pressure on interest rates. Aside from directly discouraging private con sumption and investment spending, higher interest rates, like taxes, are politically undesirable. Hence, these first two methods have typically not been favored. If the central bank must submit to political pressure to contain increases in interest rates, the solution is clear. The monetary authority is compelled to buy at least a portion of the Government issues Page 20 DECEMBER 1 9 7 5 from the private sector. This action undoubtedly mitigates the initial pressure on interest rates, but at the same time it stimulates money growth and the ensuing inflation leads eventually to higher interest rates. The process I have outlined here is not hypothet ical; we have seen it in operation over the greater part of the past thirty years. Since 1950, the Federal Gov ernment’s debt has grown by $176 billion. In that same period, the Federal Reserve System’s holdings of debt have grown by $68 billion and the money stock has increased by $176 billion. Meanwhile, proponents of deficit spending as a stimulus have proudly pointed to their successes as they saw output and employment increase — however briefly — with each new deficit, and considered the attendant inflation a small price to pay for the short-run achievements. To sum up: there is no convincing evidence that increased Government spending, with its accompany ing deficits, has accomplished its stated social goals. There is no evidence whatsoever that it is the most efficient way to pursue these goals or even that any benefits have exceeded the costs involved. On the other hand, there is overwhelming evidence that it has led to our persistent inflation. I can therefore say unequivocally that not only are the causes of inflation identifiable, but they can be eliminated. That they should be eliminated becomes clear once we consider the consequences of inflation. Economic Instability One of these results is that it can inspire monetary policies which reinforce inflationary pressures. I have already discussed the fact that increased Government borrowing exerts an upward pressure on interest rates. When the central bank is then called upon to mone tize a part of the debt in order to counteract that pressure, inflation ensues. Each time this process has been pursued, interest rates have not stayed down for long. As people become aware of inflation and the expanded money supply, they expect prices to rise further. Interest rates rise as inflationary premi ums are incorporated into them. The central bank again attempts to resist the rise by increasing the money supply and the whole cycle is renewed. A closely related policy-induced effect is the reces sions brought about by recurrent efforts to reduce inflation rapidly and drastically. When the concern for inflation becomes greater than that for interest rates, there are periodic attempts to reduce the rate of price rise by sharp reductions in the rate of money F E D E R A L R E S E R V E BAN K O F ST. LO UI S growth. These reductions have been responsible for most of our recessions and increases in unemployment. Arbitrary W ealth Transfers The less visible consequences of inflation are per haps even more ominous. An inflation which is not fully anticipated brings about a redistribution of wealth from creditors to debtors. When people see this occurring, they will bend their efforts toward pro tecting themselves from these effects. Another subtle aspect of inflation is the loss which inflation imposes on all holders of money. Inflation leads all economic units, both individual consumers and firms, to try to maintain smaller money balances and, as they become a more costly productive re source, to make greater attempts to economize on their use. But these attempts require the use of substitute resources, not the least of which are the time and effort involved in devising alternatives to money trans actions. I think you can easily visualize where this leads; we are all aware of the inefficiences of bilateral barter transactions. Money is a useful good which permits increased specialization in production and any decrease in that specialization necessarily leads to a reduction in output. The recorded instances of very rapid rates of inflation in Europe and South America convincingly illustrate this fact. Increased Uncertainty A major consequence of the inflation that we have experienced is the increased uncertainty which has had an impact on every aspect of our economic life. There are really two factors at work here. First, when a society has come to expect a fluctuating inflation rate which cannot be accurately predicted, long-term financial contracts become increasingly risky to both lenders and borrowers; hence, they become increas ingly rare. I am sure you are all aware that since the early 1930s the average time to maturity of debt obli gations has decreased substantially. Greater uncer tainty — that is, greater risk — as to the financing of long-term investment leads to reluctance to undertake such investment. As a result, productive capacity is lowered and future consumption possibilities are decreased. Another source of increased uncertainty, and one whose effects become immediately apparent, is that we have been led to expect the Government periodi cally to attempt to combat inflation in ways and at times that we cannot predict. Many of these tech niques, such as wage and price controls and the re DECEMBER 1 9 7 5 actions to them, can, and already have, produced serious distortions in the economic process. An excellent example is the phenomenon observed in the American automobile industry in the past year. Faced with poor sales, manufacturers reacted not with straightforward price cuts, but with elaborate rebate programs which were more costly for both them and their customers. The only reason which I can see for this extraordinary maneuver is that they feared the imminent reimposition of price controls and wished to insure themselves the greatest possible flexibility in the face of this threat. It is the long-term, often slowly working, and hardly visible effects of inflation, which, in my opinion, repre sent the greatest danger. They lower the standard of living; they undermine the fiber of our political, eco nomic and social system; and because they are not readily apparent, inflation frequently is considered to be of secondary importance to more visible, but transi tory, economic problems. Our current situation affords us a perfect example of the problems I have outlined. Although it seems that we have reached the bottom of the recession and that recovery is surely underway, unemployment rates remain relatively high and some industries still suffer low rates of growth in demand. As recovery progresses and inventory liquidation ceases it is reasonable to expect that private borrowing will increase; this is bound to exert an upward pressure on interest rates. Now, how will the Government react to this combi nation of circumstances? Will it again consciously dis regard the dangers of inflation, addressing itself to the short-run unemployment problem with traditionally ill-conceived and ineffective spending programs? Such a course of action will engender massive Government demands on the credit market, adding to the upward push on interest rates. To combat this, money growth must accelerate, bringing with it greater inflation in a year or so and still higher interest rates. What then? Will aggravated inflation be permitted or will we subject the economy to another recession? Or shall we, alternatively, break from our traditional response, allow the economy to continue the progress it has begun, and not create new problems by attempts to accelerate that progress or to depress the interest rate. These are the alternatives which face policy makers. SUMMARY AND CONCLUSIONS In conclusion, let me restate my fundamental propo sitions. First, it is quite evident that inflation is the Page 21 F E D E R A L R E S E R V E B A N K O F ST . L O U I S result of excessive monetary growth and that demandinduced recessions are caused by sharp downward deviations from this growth path. Second, monetary growth in excess of resource growth has been the most dependable result of Government deficits and the de sire to mask the resource transfers that these deficits are assumed to entail. Third, deficits have typically arisen from attempts to change socio-economic condi tions — attempts which have, just as typically, been futile. Solutions are readily available, but they require a time horizon which extends beyond the next election and beyond the short-term outlook and narrow analyt ical base of many economists. The basic requirement is the realization that all social and economic programs entail a cost which must be paid in one form or another. If this realization becomes prevalent and if the costs become clear, there will be no need for central bank financing of huge Government deficits. Neither will there be a necessity for maintaining in terest rates at some predetermined level. In short, there will be no need to fool the electorate. This Page 22 DECEMBER 1 9 7 5 would free the monetary authorities to control the growth of the money stock, keeping it at a rate con sistent with the rate of growth of output and eliminat ing the major cause of both inflation and demandinduced recession. Meanwhile, in the current circumstances, it is per fectly feasible to permit interest rates to seek their market-determined level and to start a very gradual deceleration in the trend rate of money growth. It may take a year or two or three, but inflation can be re duced without the emergence of recession. But again, a necessary condition is the discipline imposed by public knowledge that any service provided by the Government must be paid for by the public itself and must be paid immediately. Perhaps such knowledge will reduce demands for Governmental services, or at least eliminate the politi cal pressures to pretend that these services can be provided free of charge. And in my opinion, these pre tentions are the major impulses which set in motion the causes of inflation. EIGHTH FEDERAL RESERVE DISTRICT HEAD OFFICE, BRANCH, AND OTHER CITIES OF OVER 10,000 POPULATION Legend o H e a d O f f ic e ® B ra n c h O ffic e s m o f the FRB, St. Lo uis o f the FRB, S t. Louis S ta n d a rd M e tro p o lita n S t a t is t ic a l A re a s ■ P la c e s ■ P la c e s o f 4 0 - 5 0 ,0 0 0 o ver 5 0 ,0 0 0 District States P la c e s o f 3 0 -4 0 ,0 0 0 • P la c e s o f 2 0 -3 0 ,0 0 0 o P la c e s o f 1 0-2 0,00 0 r - T - i l l l . llnd Jj^Mo.Sjv^ T e n n *_^ P o p u la tio n is b a s e d on the 1970 c e n su s. y iss M — S ta te Bo u n d a rie s — D is tric t 0 32 F e d e r a l R e s e rv e B a n k o f S t. L o u is B o u n d a ry 64 96 D E C EM B ER 1975