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. L O U IS JANUARY 1972 C O N TEN TS ST. LOUIS • LOUISVILLE : • EIGHTH DISTRICT , • M E M P H IS LITTLE ROCK Vol. 5 4 , No. 1 Government Debt, Money, and Economic Activity............................................................ 2 Two Critiques of Monetarism A Critical Look at Monetarist Economics 10 Comments on a Monetarist Approach to Demand Management ........................... 26 Government Debt, Money, and Economic Activity1 by KENNETH STEWART I HE AMERICAN economy in the last six years has experienced a high rate of inflation. The recent recession, which led to an increase in the rate of unemployment, was not accompanied by a rapid re duction in the rate of inflation. As a result, the effec tiveness of traditional stabilization measures was ques tioned, and the New Economic Program, which includes administrative controls on prices and wages, was initiated as a solution to these problems. An examination of economic evidence over the past twenty years suggests that the course of monetary expansion can explain both the emergence of inflation in the mid-1960s and the occurrence of a high un employment rate at the turn of this decade. The pat tern of monetary growth has been, in turn, greatly influenced by growth in Federal Government debt. This article relates trend rates of growth of money and changes in rates of monetary growth during the past two decades to changes in output, employment, and prices. It further analyzes the growth of Govern ment debt and its relationship to the expansion of the money stock. Money and Economic Activity According to the view presented in this article, the economy is considered to be basically stable and in the long run to move along a trend path of output determined by growth in its productive potential. Some variation in output and employment around 'This article expands some of the views initially presented in a speech by Darryl R. Francis at the Annual Intermountain Banking Seminar, Utah State University, Logan, Utah, No vember 18, 1971, as well as in papers presented by Leonall C. Andersen at the Nineteenth Annual Conference on the Economic Outlook, The University of Michigan, Ann Arbor, Michigan, November 18, 1971, and by Jerry L. Jordan at the National Bureau of Economic Research Conference on Secu lar Inflation, Chicago, Illinois, November 5 and 6, 1971. Page 2 the trend path occurs due to disturbances from labor strikes, crop failures, changes in tax rates and other factors, but these disturbances have seldom been the dominant force in causing recessions or inflations. Evidence indicates that marked and sustained changes in the rate of monetary expansion have been a major factor underlying virtually all cyclical fluctua tions and inflations. Changes in the rate of growth of the money stock have been shown to have predictable effects on total spending in the same direction.2 Changes in total spending have been associated first with changes in output and later with changes in prices.3 Consequently, the trend rate of growth of the money stock, defined in this article as demand deposits and currency held by the nonbank public, is viewed as having a major influence in determining the trend rate of growth of prices, whereas accelerations and decelerations in the growth rate of money lead mainly to short-run fluctuations in output and employment.4 These short- and long-run effects of money stock growth on prices, output, and employment are demon 2In the equation formulated by Leonall C. Andersen and Jerry L. Jordan, a marked and sustained change in monetary growth has its major effect on nominal GNP within five quarters. See Leonall C. Andersen and Jerry L. Jordan, “Monetary and Fiscal Actions: A Test of their Relative Im portance in Economic Stabilization,” this Review (November 1968), pp. 11-24. 3Equations formulated by Leonall C. Andersen and Keith M. Carlson indicate that monetary actions generally affect total spending with a two-to-three quarter lag. A change in the rate of growth of total spending was accompanied by a simultaneous change in the rate of growth of output. Prices changed more slowly following a change in total spending. See Leonall C. Andersen and Keith M. Carlson, “A Monetarist Model for Economic Stabilization,” this Review (April 1970), pp. 7-25. 4In “Money Supply and Time Deposits, 1914-69,” this Review (March 1970), pp. 6-10, changes in money growth rates and cyclical movements in economic activity were compared. JANUARY 1972 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 strated on Chart I.5 The trend rate of growth of the money stock, as shown in the top tier, increased from a 1.7 percent annual rate through most of the 1950s and early 1960s, to 3.7 percent in the first half of the 1960s, and to 5.8 percent in the second half of the 1960s and early 1970s. The trend rate of growth of prices, as shown by the General Price Index panel, rose in a similar pattern from the 1950s through the 1960s, reflecting, after about a three year lag, changes in the trend growth of the money stock. Relationships between output and employment and the growth of the money stock relative to its under lying trend rates can be observed in the top and bot tom tiers of Chart I. During the two decades covered, six periods of money stock growth occurred at rates significantly greater than the underlying trend.6 Each of these periods was accompanied (with a lag of one or two quarters) by an upward movement in real output toward or above potential real output as esti mated by the President’s Council of Economic Advisers. During this same twenty year interval the economy experienced four recessions (as defined by the Naional Bureau of Economic Research) and two periods of brief economic slowdown. Each of the four reces sions (shaded areas in Charts I and II) was preceded by a marked slowdown or an absolute decline in the rate of growth of the money stock. The recessions occurred in the periods 1953-54, 1957-58, 1960-61, and 1969-70. When the rate of growth of the money stock slowed in 1962 and 1966, the growth rate of real out put slowed, and a rise in the rate of unemployment followed. The 1962-63 and 1966-67 periods of slow down were not of significant magnitude and duration to be labeled recessions. Chart I does not offer conclusive evidence that monetary growth affects economic activity. However, the relationships shown on the chart are consistent with the view that the trend growth of money is a major influence on long-run price movements, and that accelerations and decelerations of monetary growth about the trend have predictable effects on output and employment in the short run.7 Price movements, on 5For econometric evidence supporting the interpretation of these charts, see Andersen and Carlson, “A Monetarist Model.” 6As used in this context, a period is a time interval of at least six months duration. These periods of accelerating money growth began in late 1951, 1954, 1958, 1961, 1965, and 1968. 7For an elaboration of a theoretical foundation underlying these relationships, see Karl Brunner, “A Survey of Selected Issues in Monetary Theory,” Schweizerische Zeitschrift fiir Volkswirtschaft und Statistik (No. 1, 1971), pp. 1-146. the other hand, have been little affected by short-run variations in monetary growth. The experience of the last two decades also sug gests that monetary growth has little lasting influence on the rate of unemployment and the growth rate of real output.8 Despite variations in the rate of mone tary growth about its trend as well as changes in the trend in the 1950s and 1960s, growth of real out put tended to move towards or along its potential growth path.9 The unemployment rate averaged 4.9 percent from 1952 to 1962 and averaged 4.6 percent since then. The lasting effect of monetary actions is on the trend of prices, whereas output and employ ment growth depend on real factors —labor force trends and productivity. Determinants of the Money Stock In view of these observed relationships between money and economic activity, it is important to con sider the factors which affect movements in the money stock. The money stock ( M) , defined in this article as demand deposits and currency held by the nonbank public, can be expressed as a function of the monetary base (B ) and a money multiplier (m ) such that: M = mB. Using this relationship, factors which cause the money stock to change can be summarized by changes in the monetary base and the multiplier. The multiplier over the past twenty years has been fairly stable.10 It has fluctuated over a narrow range and has been shown to be predictable.11 Conse quently, the trend rate of growth of the money stock has been dominated by the trend rate of growth of the monetary base. The close association between sFor an explanation of this observation see Milton Friedman, “The Role of Monetary Policy,” The American Economic Review (March 1968), pp. 1-17, and in The Optimum Quantity of Money and Other Essays (Chicago: Aldine Publishing Company, 1969), pp. 95-110. 9After the 1960-61 recession, the movement back toward po tential real output was relatively slow. This period followed two recessions only two years apart which provided a basis for the growing belief in the early 1960s that the economy was becoming subject to relatively short business cycles. Such a belief was probably a contributing factor to the slow recov ery to full employment in the early 1960s. In addition, the economy received a minor additional shock shortly after the 1960-61 recession when money declined relative to the trend in 1962. 10The money multiplier summarizes the decisions of commercial banks to hold excess reserves, of the Government to hold demand deposits, and of the public to hold currency, demand deposits, and time deposits. A discussion of factors affecting the money multiplier is presented by Jerry L. Jordan, “Ele ments of Money Stock Determination,” this Review (October 1969), pp. 10-19. n See Albert E. Burger, Lionel Kalish III, and Christopher T. Babb, “Money Stock Control and Its Implications for Monetary Policy,” this Review (October 1971), pp. 6-22. 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 IS JANUARY 1972 C hart I Influ en ce of M o n e y on Prices, O u tp u t a n d U n e m p lo y m e n t Latest d a ta p lo tte d : u n e m p lo y m e n t ra te -IV /1 9 7 1 ; a ll o th e r d o ta - lil/1 971 Page 4 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 JANUARY 1972 Chart I Influence of Federal G overnm ent Debt on M o n etary Expansion 195? 1953 1954 1955 1956 1957 1958 1959 I960 1961 196? 1963 1964 1965 1966 1967 1968 1969 1970 1971 197? >NS OF DOLLARS Federal Governm ent Debt 300 T ed eral Governm ent Debt is th e sum of de bt held by Federal Reserve Banks - and d e b t held by private investors. The origin al da ta may be found in the ta ble entitled "O w n ersh ip of Public D eb t" in th e Federal Reserve Bulletin. : .■ Seasonally A d juste d________________________ 4th q tr 70 ' RATIO SCALE ^ S IlllO N S OF DOLLARS Federal Government Debt Held by the Federal Reserve System Seasonally Ad juste d ! _ |___ *217, S§3rd qtr 6) 1 11 . . i i i i « i r Percent of Federal Governm ent Debt Held by the Federal Reserve System Seasonally Ad juste d RATIO SCALE BILLIONS OF DOLLARS 4 t h qtr M o n e ta ry Base S e aso na lly A djusted rd qtr 61 I. . .t_L RATIO SCALE >NS OF DOLLARS M o n e y Stock S easonally A djusted 3rd qtr 62 I 11 Latest d a ta plo tte d: F ed erol G o vernm en t d e b t, m oney sto ck-lll/19 71 ; m on eta ry base-IV/1971 Page 5 F E D E R A L R E S E R V E B A N K O F ST. L O U IS JANUARY 1972 T a b le I Sources o f the B ase M o n th ly A ve ra g e s o f D a ily Figures (D o lla r A m o u n ts in M illio n s , N o t S e a s o n a lly A d ju s te d ) Percent o f Base Dec. 1971 Sept. 1949 Dec. 1971 $ 1 7 ,4 4 1 $ 6 9 ,2 6 2 4 0 .9 % 7 6 .4 % 164 108 .4 .1 352 3 ,9 1 5 .8 4 .3 2 4 ,6 3 7 1 0 ,1 3 2 5 7 .8 1 1.2 1 0.8 8 .4 Sept. 1949 H o ld in g s o f G o ve rn m e n t Securities D iscounts a n d A dvances F loat F ederal Reserve C re d it G o ld Stock S p e cia l Tre a sury D ra w in g Rights C urre n cy 7 ,6 1 2 4 ,5 9 2 O u ts ta n d in g T re a sury Cash H o ld in g s .4 400 - 1 ,3 1 0 — 454 649 — 1 ,9 2 6 - 3.1 — .5 — 1.5 — 2.1 the monetary base (Table I ) . Since 1949, the amount of gold held has declined almost continuously. The de cline in gold stock has contributed a negative influence to growth of the base, while increases in Federal Re serve holdings of U.S. Government se curities, the dominant component of Federal Reserve credit, has contributed a positive influence. Other sources, though their net influence has been positive, have contributed relatively little to movements in the base during the past twenty years. From 1952 to the middle of 1961, increases in securities held by the Fed O th e r D eposits a n d O th e r eral Reserve System almost offset de — 2.6 2 ,3 2 4 4 .0 1 ,6 9 3 F ederal Reserve Accounts creases in the gold stock. The mone 1 0 2 .2 % 9 5 .7 % $ 4 3 ,5 3 4 $ 8 6 ,7 2 5 Source Base tary base grew slowly in this period. 2.2 4 .3 3 ,9 3 0 — 938 Reserve A d ju s tm e n t M a g n itu d e Beginning in the 1960s, increases in 1 0 0 .0 % 1 0 0 .0 % $ 9 0 ,6 5 5 $ 4 2 ,5 9 6 M o n e ta ry Base Federal Reserve holdings of Govern ment securities more than offset reduc Totals may not add due to rounding. tions in the gold stock, and the monetary base grew more rapidly. A two-tiered gold these two rates is indicated by the trend lines in the system, established in March 1968, separated the gold bottom two panels of Chart II. market into private and official sectors, each with its The monetary base represents the net monetary own price. Since April 1968, the gold stock has re liabilities of the Government (U.S. Treasury and Fed mained roughly constant and has contributed little to eral Reserve System) held by the public (commercial growth of the monetary base. Gold now represents banks and nonbank public). The monetary base has only 11.2 percent of the base. been referred to as “high powered” money because it Holdings of Government securities by the Federal can be used as reserves of commercial banks to ex Reserve represent the System’s acquisitions of Federal pand demand deposits by more than the amount of Government debt through its open market operations. reserves.12 These security holdings presently comprise 76.4 per cent of the monetary base, and since the early 1960s Given that changes in the monetary base are the changes in security holdings have been the dominant major determinant of changes in the rate of monetary influence on growth of the base. Through purchases expansion, it is important to ascertain the factors and sales of securities, called open market operations, which have led to changes in the base. Table I presents the sources of the monetary base. Growth of the Federal Reserve can control the growth of the monetary base by offsetting or complementing any the monetary base during the past twenty years has movements in other sources. been determined primarily by two sources — Federal Reserve Credit and the gold stock. An increase in the dollar amount of either of these sources, other things Influence of the Federal Government Debt equal, increases the monetary base by an equal on Monetary Expansion amount. Growth of Government securities held by the Fed In September 1949, when the gold stock source of eral Reserve System depends on the growth of Gov the base was at its peak, it comprised 57.8 percent of ernment debt and the percent of this debt the System decides to purchase. This section traces the growth of 12A discussion of the monetary base is presented by Leonall Government debt over the last twenty years, the ac C. Andersen and Jerry L. Jordan, “The Monetary Base — quisition of debt by the Federal Reserve System and Explanation and Analytical Use,” this Review (August the reasons for debt acquisition by the System. 1968), pp. 7-11. Tre a sury D eposits a t Federal Reserve — Page 6 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 Growth in Federal Government Debt Outstanding Growth of Government debt is shown in the top tier of Chart II.13 Government debt outstanding oscil lated around a one percent annual trend rate of growth from the first quarter of 1952 to the third quarter of 1961. Unified budget deficits of $3.4 billion and $7.1 in fiscal years 1961 and 1962, respectively, initiated an increase in the trend rate in the early 1960s. From the third quarter of 1961 to the fourth quarter of 1966, Government debt rose by $20.2 billion, or at an annual trend rate of 1.6 percent. Large unified budget deficits of $8.7 billion and $25.2 billion were incurred in fiscal years 1967 and 1968, respectively. These deficits further increased the trend growth rate of Government debt. From the fourth quarter of 1966 to the fourth quarter of 1970 Government debt grew by $27.8 billion, or at a 2.6 percent annual rate. Government debt grew in the early 1960s mainly because of deficits incurred in fiscal years 1961 through 1965. During this period outlays for domestic civilian programs increased at about an 8 percent annual rate and tax receipts rose at a 5 percent rate. The slower growth in tax receipts reflected tax cuts in 1962, 1964, and 1965. In the second half of the 1960s, de fense expenditures rose sharply, while at the same time nondefense expenditures accelerated further. These rapid expenditure increases were not accom panied by increased tax rates, except in fiscal 1969, and as a result, large deficits were incurred in fiscal years 1967, 1968, 1970, and 1971. Federal Reserve Acquisition of Debt and Growth of the Monetary Base Federal Government debt held by the Federal Re serve System changed little in the 1950s, but then grew rapidly in the 1960s. Changes in the monetary base during the 1960s roughly paralleled that of the System’s holding of debt. Debt acquisition by the Federal Reserve System and the percent of debt held by the System are shown on Chart II (second and third panels from the top). Between the first quarter of 1952 and the third quarter of 1961, the proportion of Government debt held by the Federal Reserve System remained roughly constant at around 11 percent. As Government debt increased, securities held by the Federal Reserve Sys tem increased proportionally, and as the debt de 13Federal Government debt is defined in this article as the sum of debt held by Federal Reserve Banks and debt held by private investors. The original data may be found in the table entitled “Ownership of Public Debt” in the Federal Reserve Bulletin. JANUARY 1972 creased, securities held decreased proportionally. Variations in Government debt outstanding in the 1950s, especially late in the decade, tended to ac celerate and decelerate growth in the monetary base. Variations in the base, in turn, were a major cause of fluctuations in the money stock. When the trend rate of growth of Government debt increased in the first half of the 1960s, the percent of the debt held by the Federal Reserve also increased, as the rate of acquisition of debt by the Federal Re serve was more rapid than the expansion of the Gov ernment debt itself. Increased purchases of Govern ment securities by the Federal Reserve directly in creased the monetary base, increasing its trend rate of growth, which in turn increased growth of the money stock and economic activity. As resource utilization approached its upper limit, as defined by potential output, the rate of inflation increased. From the third quarter of 1961 to the fourth quar ter of 1966, the Federal Reserve purchased $15.9 bil lion of Government securities adding to its portfolio at a 9.1 percent average annual rate. The effect of debt acquisition on growth of the monetary base was partially offset by a $4.3 billion decline in the gold stock, and the monetary base grew by $13.7 billion. This increase accelerated growth of the base to a 4.4 percent annual rate, and growth of the money stock began to accelerate in the third quarter of 1962. Real output grew with little effect on prices until 1965 when a high level of resource utilization was reached and price increases began to accelerate. The Federal Reserve continued to rapidly increase its security holdings in the second half of the 1960s, when growth of the debt accelerated further. As a result, growth of the monetary base, money stock, and prices accelerated. From the fourth quarter of 1966 to the fourth quarter of 1970, Government debt held by the Federal Reserve grew by $17.2 billion, or at an 8.7 percent annual rate. As a consequence, the portion of debt held by the Federal Reserve increased from 16.8 percent in late 1966 to 21.1 percent in late 1970. The base increased by $16 billion, accelerating to a 5.4 percent annual rate of growth. Money stock growth accelerated to a 5.8 percent annual rate during this period and the rate of increase in prices climbed to a 5.1 percent rate beginning in the second quarter of 1969. Major Factors Influencing Acquisition of Debt by the Federal Reserve The Federal Reserve System purchases Govern ment securities for several purposes. However, conPage 7 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 JANUARY 1972 Interest Rates 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 Shaded are as represent pe rio ds of business recessions as d e fin e d b y (he N o tio n a l Bureau o f Economic Research la te s t d a ta p lo tte d : D ecember cem over market interest rate movements has been a major factor influencing Federal Reserve acquisition of Government debt over the last two decades.14 Debt issues by the Federal Government put up ward pressure on interest rates. When the Federal Reserve System buys Federal Government debt in the open market, both the supply of credit and the money stock are increased. Greater availability of funds in the credit markets initially puts downward pressure on interest rates. System actions are thereby capable of preventing interest rates from rising during times of Treasury borrowing. Purchases of securities also increase the monetary base which produces an expansion in the money stock. If growth of the money stock is greater than increases in the demand for money balances, then the difference will tend to be reflected in an increase in aggregate demand. An increase in aggregate demand stimulates economic activity and tends to increase the demand for credit placing upward pressure on market 14Michael W. Keran and Christopher T. Babb, using regres sion analysis, found that changes in Federal Reserve hold ings of Government securities and changes in the monetary base were influenced, in descending order of importance, by market interest rates, changes in the amount of United States Government debt outstanding, and economic stabili zation objectives. See Michael W. Keran and Christopher T. Babb, “An Explanation of Federal Reserve Actions (19 33 -68),” this Review (July 1969), pp. 7-20. Page 8 interest rates. If prolonged price increases accompany an acceleration in total spending, expectations of future price increases develop. Borrowers are then willing to pay and lenders demand an inflation pre mium which raises market interest rates. Thus, sus tained increases in the money stock usually exert upward pressure on interest rates. The initial, short-run impact of its security pur chases on interest rates generally has received the greatest attention in the day-to-day operations of the Federal Reserve System. Large debt acquisition by the System has resulted from attempts to maintain existing money market conditions during times of Treasury borrowing. The positive longer-run impact of monetary expansion on interest rates has been a factor leading to an accelerating trend rate of growth of the money stock in the 1960s. Variability around trend movements of the mone tary base may be attributed in considerable measure to alternating concern between reducing inflation and facilitating a relatively rapid economic expansion to lower the rate of unemployment. When the rate of inflation intensified monetary authorities sought higher interest rates; consequently, the rate of growth of the base ( and money stock) slowed markedly for a period. Shortly thereafter economic activity slowed and unem 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 ployment rose. Monetary authorities then shifted ob jectives and attempted to lower market interest rates to stimulate economic activity; consequentiy, the base increased more rapidly. This rapid monetary expan sion, after a lag, placed further upward pressure on prices, setting the basis for a future round of monetary restraint.15 Conclusions This article emphasizes a number of propositions which may be summarized as follows: 1. The trend rate of growth of the money stock plays a major role in determining the trend rate of growth of prices. Marked and sustained changes in the growth rate of the money stock are followed by short-run variations in output and employment. 15Examples of such short-run destabilizing monetary actions have been noted in this Bank’s Review. See Reprints 17, 22, 28, 39, 57, and 68, for annual reviews of monetary actions for the years 1965 through 1970, respectively. A study of the released “Minutes of the Federal Open Market Committee” for the years prior to 1965 indicates that monetary develop ments were similar in earlier years. JANUARY 1972 2. Growth of the money stock is dominated by growth of the monetary base. 3. Even though monetary authorities can independ ently control movements in the monetary base, growth of the base has been greatly influenced by growth of Government debt and concern about movements in market interest rates. A steady, moderate rate of monetary expansion can help foster noninflationary growth and promote sta bility. Such a course of monetary expansion may be difficult to achieve at the present time, unless impedi ments to such expansion are reduced. The Federal Government deficit during fiscal year 1972 is expected to be extremely large, representing a substantial de mand for credit, which in turn, would be expected to exert upward pressure on market interest rates. Public sentiment against high or rising interest rates is deeply imbedded in traditional American thought. A step towards lessening the influence of these im pediments would be for market interest rates to receive less emphasis in the determination of monetary actions. Page 9 A Critical Look at Monetarist Economics by RONALD L. TEIGEN Ronald L. Teigen is an Associate Professor of Economics at the University of Michigan. He received a PhD degree in Economics from Massachusetts Institute of Technology. Professor Teigen is the author of several articles in monetary economics, and is co-author with Warren L. Smith of R e a d i n g s i n M o n e y , N a t i o n a l I n c o m e , a n d S t a b i l i z a t i o n P o l i c y . This paper was pre sented at the Annual Conference of College and University Professors of the Federal Reserve Bank of St. Louis on November 12,1971. U N TIL JU ST a few years ago, the viewpoint which lately has come to be known as “monetarist” was not taken very seriously by anyone except a few dedicated disciples. Its central postulate — that changes in the level of aggregate money income were due essentially to prior money stock changes —was viewed as a totally inadequate oversimplification, es pecially since the proponents of this approach failed to provide an adequately detailed explanation of the theoretical structure upon which this tenet was based.1 The empirical evidence presented in support of this “quantity theory” viewpoint was subjected to criticism so severe that the evidence has never been taken very seriously.2 'In particular, Milton Friedman’s well-known article, “The Quantity Theory of Money — A Restatement,” in M. Friedman, ed., Studies in the Quantity Theory of Money (Chicago: University of Chicago Press, 1956), pp. 3-21, which has been cited as the basis for much monetarist work, has been shown by Don Patinkin to be a sophisticated version of Keynes’ liquidity preference theory rather than the up-to-date state ment of an alleged Chicago oral tradition that monetarists take it to be. See Don Patinkin, “The Chicago Tradition, the Quantity Theory, and Friedman,” Journal of Money, Credit and Banking (February 1969), pp. 46-70. 2I am referring chiefly to the controversy triggered by the work of Milton Friedman and his associates in the late 1950s and early 1960s, especially Friedman’s evidence on lags observed between changes in the rate of change of the money stock and changes in GNP, as presented in his paper, “The Supply of Money and Changes in Prices and Output,” Joint Economic Committee, U.S. Congress, 1958, and else where, and in the Milton Friedman and David Meiselman paper on, “The Relative Stability of Monetary Velocity and the Investment Multiplier in the United States, 1897-1958,” in Commission on Money and Credit, Stabilization Policies (Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1963). The regression results reported in the latter paper were severely criticized by Donald Hester in the November 1964 Review of Economics and Statistics and by Albert Ando-Franco Modig liani and Michael DePrano-Thomas Mayer in the September 1965 American Economic Review. The lead-lag observations discussed in the former paper were criticized by John M. Culbertson in the December 1960 Journal of Political Econ omy, and by James Tobin in the May 1970 Quarterly Journal of Economics. Digitized for Page 10 FRASER However, recent years have witnessed something of a turnaround. The conventional wisdom as em bodied in modem Keynesian theory has been cast into doubt, while monetarist thinking has increased greatly in popularity, to the point where its propo nents, and even some of its critics, speak of a “mone tarist revolution”.3 The reasons for this rather sudden change are no doubt related in part to the apparent inconsistency of the Keynesian analysis (or at least an elementary version of it) with economic events in the United States during the late 1960s,4 in some 3See Karl Brunner, “The ‘Monetarist Revolution’ in Monetary Theory,” Weltwirtschaftliches Archiv (No. 1, 1970), pp. 1-30, and Harry G. Johnson, “The Keynesian Revolution and the Monetarist Counter-Revolution,” American Economic Review, Papers and Proceedings (May 1971), pp. 1-14. 4The apparent failure of the income tax surcharge of June 1968 to reduce aggregate demand rapidly has been inter preted by some to be evidence of the failure of the “new” economics. However, it is not at all clear that the surtax was ineffective. In a recently-published study by Arthur Okun, evidence is provided that, at least in some categories of spending (nondurable goods and services in particular), the surcharge seems to have reduced demand substantially. But in other categories ( especially demand for new automo biles) no reduction is apparent. See Arthur M. Okun, “The Personal Tax Surcharge and Consumer Demand, 1968-70,” Brookings Papers on Economic Activity (No. 1, 1971), pp. 167-204. More generally, the notion that demand should have been observed to fall after the surtax was imposed is based on simplistic and partial analysis. When the surtax is analyzed within the context of a complete model (in which govern ment spending is taken into account), and one which in corporates the sophisticated theories of consumption be havior recently developed — the “permanent income” hypo thesis of Milton Friedman or the “life-cycle” hypothesis of Albert Ando and Franco Modigliani —there appear a number of considerations which suggest that no substantial diminu tion of total demand could be anticipated. This point of view is argued persuasively by Robert Eisner in his paper, “Fiscal and Monetary Policy Reconsidered,” American Economic Review (December 1969), pp. 897-905. Eisner reasons that rising Government expenditure had been expanding demand rapidly at the time when the surtax was enacted; furthermore, under the Friedman and Ando-Modigliani theories, which postulate that it is some long-run measure of income or wealth rather than current-period income which determines JANUARY 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 degree to monetarist criticism of Keynesian analysis (mostly directed at a very elementary version of it), and in part to other causes, including substantial de velopment by the monetarists of their own theoretical position, as well as the appearance of new and more convincing empirical findings.5 While the increase in popularity of monetarism has been rapid, and the rate of growth of the monetarist literature impressive, a critical literature has also ap peared, charging that monetarist theory has turned out largely to consist of old concepts clothed in new names, and that the empirical evidence purportedly supporting the monetarist position is biased and un dependable.6 The purpose of the present paper is to attempt to summarize in a general way the main fea tures of the present monetarist theoretical stance, and to examine the monetarist view of modem Keynes ianism. Since much of the debate bears directly on the stabilization policy process and the relative use fulness of different instruments of policy, particular attention will be given to the nature of the transmis sion mechanism under the two approaches. The em pirical evidence will not be discussed in a systematic way in this paper, although reference will be made to it, where appropriate, in the discussion of the theories. In conducting this comparison, I shall attempt to identify issues between the two camps which are real, and those which seem to be false. The Structure of Monetarist Thought Although the roots of modem monetarist thought extend far back in time (the writings of classical economists are often cited, Irving Fisher being partic ularly popular), it is only lately that detailed exposi tions of this theory have begun to appear. In this paper, no systematic discussion of the entire literature a household’s living standard, a temporary tax change (such as the 1968 surcharge) would be expected to have only minor effects on spending because it does not change longrun expected income significantly. See Milton Friedman, A Theory of the Consumption Function (Princeton, N.J.: Prince ton University Press, 1957), and Albert Ando and Franco Modigliani, “The ‘Life-Cycle’ Hypothesis of Saving: Aggre gate Implications and Tests,” American Economic Review (March 1963), pp. 55-84. 5Harry Johnson, “The Keynesian Revolution and the Mone tarist Counter-Revolution, ’ suggests that the successful mone tarist upsurge may also be due to the factors related to the conversion of the “Keynesian revolution” of the 1930s into the economic orthodoxy of the 1960s. 6Ibid., for a general discussion of monetarist theory and its relationship to Keynesian orthodoxy. There have been pub lished a large number of papers critical of the recent mone tarist empirical studies; references to some are given in footnote 2, and a summary of the criticism of more recent monetarist empirical work is contained in Ronald L. Teigen, “The Keynesian-Monetarist Debate in the U.S.; A Summary and Evaluation,” Statsokonomisk Tidsskrift (January 1970), pp. 1-27. 1972 will be undertaken. Instead, important summary statements which recently have become available in articles by Andersen, Brunner, Fand, Friedman, and others will be taken to be representative of presentday monetarist thought.7 Models, Assertions and Themes As a useful starting point in establishing a general framework for the discussion to follow, we may refer to recent articles by Brunner and Friedman contain ing inclusive statements of the monetarist position.8 Friedman provides an explicit statement of the staticequilibrium structure which he views as being con sistent with both the monetarist and Keynesian schools of thought. The theme he stresses — that it is the particular features of or assumptions about particular characteristics of the general analytic structure, rather than the fundamental nature of the structure itself, which differentiate monetarists and Keynesians — also appears in the writings of Brunner and others. In summary form, the model set out by Friedman is as follows: (1 ) | = C ( J , r) + I(r) (2 ) Mo = p*L( — r ) P (3 ) Y = py where Y is money income, p is the general price level, r is the rate of interest, M0 is the nominal exogenouslyset money stock,9 y is real income or output, and C, 7Some of the important articles include Leonall C. Andersen, “A Monetarist View of Demand Management: The United States Experience,” this Review (September 1971), pp. 1-11; Leonall C. Andersen and Keith M. Carlson, “A Monetarist Model for Economic Stabilization,” this Review (April 1970), pp. 7-25; Leonall 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; Karl Brunner, “The Role of Money and Monetary Policy,” this Review (July 1968), pp. 9-24; idem, “The ‘Monetarist Revolution’ in Monetary Theory;” idem, “A Sur vey of Selected Issues in Monetary Theory,” Schweizerische Zeitschrift fiir Volkswirtschaft und Statistik (No. 1, 1971), pp. 1-146; idem, “The Monetarist View of Keynesian Ideas,” Lloyds Bank Review (October 1971), pp. 35-49; David I. Fand, “Keynesian Monetary Theories, Stabilization Policy and the Recent Inflation,” Journal of Money, Credit, and Banking (August 1969), pp. 556-87; idem, “Monetarism and Fiscalism,’ Banca Nazionale del Lavoro Quarterly Review (Sep tember 1970), pp. 275-89; idem, “A Monetarist Model of the Monetary Process,” Journal of Finance (May 1970), pp. 275-89; Milton Friedman, “A Theoretical Framework for Monetary Analysis,” Journal of Political Economy (M arch/ April 1970), pp. 193-238; idem, “A Monetarist Theory of National Income,” Journal of Political Economy (M arch/ April 1971), pp. 323-37. 8Friedman, “A Theoretical Framework,” and Brunner, “The ‘Monetarist Revolution’.” !)In one version of Friedman’s statement, the money supply is made a function of the interest rate rather than being as sumed to be exogenous. However, this makes no essential difference to the present discussion, as Friedman points out. Page 11 F E D E R A L R E S E R V E B A N K O F ST. L O U IS I, and L stand for the consumption, investment, and demand-for-money functions, respectively. Equation (1) is of course the familiar IS curve, from which can be obtained all combinations of real income and the interest rate which will make the flow of planned spending equal to available output, and hence will result in equilibrium in the market for goods and services. Equation (2 ) is the LM curve, which yields all combinations of real income, the in terest rate, and the price level which will equate the demand for real balances with the real value of the nominal money stock. Equation (3 ) is a definition relating nominal income and real income or output through the price level. There are of course other markets which could be considered, but which are not explicitly accounted for in equations (1) or (2); in particular, the bond and labor markets are not made explicit. Friedman argues that the assumptions made by the two camps in order to accommodate these markets and simultaneously close the system of equa tions constitute a fundamental point of difference be tween monetarists and Keynesians. As written in equa tions ( l ) - ( 3 ) , the model posited by Friedman con tains four endogenous variables —Y, p, r, and y — and therefore is underdetermined. Monetarism is said by Friedman to include with the above equations a vast number of additional relationships; specifically, a whole Walrasian system of demand equations, supply equations, equilibrium conditions, etc., which in and of themselves determine y, the level of real output. The inclusion of a Walrasian system of course implies that the equilibrium position of the model is one of full employment. (There is no such implication for the short-run dynamics of the system, however.) With real output predetermined from the standpoint of equations ( l ) - ( 3 ) , equation (1) can be solved for the equilibrium value of the interest rate, and (2) yields the equilibrium price level. Elementary manip ulation of this system gives the result that only the price level (and the money wage rate, which is not made explicit in equations (1) - ( 3 ) ) will change in response to a money stock change; the equilibrium value of the interest rate is not shifted, and therefore is said to be determined only by “real” variables.10 In other words, this version of the model displays the well known “classical dichotomy.” According to Friedman, the Keysian approach util izes a much different and less satisfactory procedure luThis statement is not accurate if the system contains a gov ernment sector which issues money-fixed claims against it self, and if real wealth is an argument in the expenditure functions, and/or if the government establishes a tax-expenditure system based on nominal variables. Page 12 JANUARY 1972 by assuming that the price level, rather than real in come, is determined outside of the postulated struc ture (Friedman refers to “. . . a deus ex machina with no underpinning in economic theory.”).1 By taking 1 the price level to be exogenous with respect to this structure, the number of variables again is reduced to three (Y, y, and r in this case). However, the system no longer is dichotomized, and all of the variables now are determined jointly rather than recursively. In particular, the static equilibrium levels of both real income and the interest rate can now be changed by both money stock and expenditure changes.1It would be a mistake to conclude from the forego ing discussion that monetarists view themselves as differing from Keynesians only in terms of the assump tions utilized to provide a unique equilibrium solution to die static IS-LM model. There are several other typically monetarist assumptions about the static and dynamic dimensions of this system. Recently, Karl Brunner has introduced four propositions which he asserts are “defining characteristics of the monetarist position.” These are: (1 ) the transmission mechanism for monetary impulses involves a very general kind of portfolio adjustment process ultimately affecting the relationship between the market price of physical as sets and their production cost, rather than only the relationship between borrowing costs and the internal rates of return on potential acquisitions of new physi cal capital, as is asserted to be the mechanism char acteristic of modem Keynesian analysis; (2 ) most of 11Friedman, “A Theoretical Framework,” p. 222. 12ln a more recent article, Friedman has proposed another means of closing this system of equations, which he labels a “third way” to distinguish it from the two procedures out lined in the body of the present paper. He views this ap proach as intermediate in respect to its theoretical position vis-a-vis the others. However, since it reduces to a relation ship between income and the past history of the money stock, as Friedman demonstrates, it seems clearly to fit in with the monetarist point of view. In this approach, it is assumed that the current market rate of interest and the ex pected market rate are kept equal by the actions of asset holders. The expected market rate, in turn, is set by the ex pected real rate plus the expected rate of price change (which by definition is the difference between the expected rate of change of nominal income and of real output). By assuming the expected real rate of interest, the expected rate of growth of real output, and the expected rate of growth of nominal income all to be determined outside the system, the market rate of interest is made into a variable determined outside the system also. Assuming further that the income elasticity of demand for money is unity, Fried man establishes a direct link between nominal income and the money stock (because under his assumptions, velocity becomes a predetermined variable); this, in turn, enables the “real” sector to be solved. One of the features of this procedure is that it provides an alternative to the assump tion of full employment. However, it entails some disadvan tages of its own, which are noted in the section of the present paper entitled “Stabilization Policy.” See Friedman, “A Monetary Theory of Nominal Income.” F E D E R A L R E S E R V E B A N K O F ST. L O U IS the destabilizing shocks experienced by the system arise from decisions of the government with respect to tax, expenditure, and monetary policy, rather than from the instability of private investment or of some other aspect of private-sector behavior, as the Key nesian view is said to assume. A related belief is that the demand-for-money function is very stable, while the policy-determined supply of money balances is unstable; (3 ) monetary impulses are the dominant factor in explaining changes in the pace of economic activity, in contrast to the Keynesian position which assertedly takes real impulses as primary; (4 ) in analyzing die determinants of change in the level of aggregate activity, detailed knowledge of “allocative detail” about the working of financial markets and institutions is of secondary importance and can be disregarded. This implies that the relationship be tween policy instruments and economic activity can be captured in a very small-scale model —perhaps even in one equation —while the Keynesian position is that knowledge of allocative detail (e.g., substitu tion relationships between various financial assets) is necessary for the proper understanding of policy processes, implying a need for complex structural models.13 JANUARY 1972 The quantity theorist accepts the empirical hypoth esis that the demand for money is highly stable — more stable than functions such as the consumption function that are offered as alternative key relations. . . . . [T ]h e stability he expects is in the functional relation between the quantity of money demanded and the variables that determine i t . . . [and] he must sharply limit, and be prepared to specify explicitly, the variables that it is empirically important to in clude in the function. F o r to expand the number of variables regarded as significant is to empty the hy pothesis of its empirical content; there is indeed little if any difference between asserting that the demand for money is highly unstable and asserting that it is a perfectly stable function of an indefinitely large number of variables.15 (2) A particular aspect of the demand for money emphasized by monetarists is that, in their analysis, the stable demand for money is concerned with real, not nominal, balances, while the authorities control the nominal supply, which tends to be quite variable relative to demand.16 This state of affairs is usually contrasted with the Keynesian case, in which the de mand for money is said to be a demand for nominal balances, either because it is (incorrectly) specified that way,17 or because, as in Friedman’s discussion summarized above, the price level is fixed so that real and nominal balances are the same. Monetarists use The statements by Brunner and Friedman are at this distinction as part of a rationalization for their tempts to sketch the fundamental structure of mone contention that their analysis implies a much broader tarism. As such, they do not emphasize or even iden concept of the transmission mechanism for monetary tify explicitly some of the specific characteristic impulses than does the Keynesian model, being based themes which permeate monetarist writing, including on a very general portfolio adjustment process work their own. Several such themes can be identified. ing through changes in a broad spectrum of asset (1) Great importance is attached to the demand- yields and price level changes, in contrast to the nar for-money function, and it is in fact the central be row cost of credit channel which is implied by the havioral relationship in the monetarist model.14 Par Keynesian demand-for-money function. This point is ticular stress is laid on its stability, by which is meant developed further in the section entitled “The Trans not only that the variance of its error term is small, mission Mechanism for Monetary Impulses” below. but much more importantiy, tiiat it contains very few arguments. Friedman has written that: (3 ) Further, monetarists believe the interest elastic 13These “defining characteristics” are discussed at some length in Brunner, “The ‘Monetarist Revolution’,” Section II. 14Thus, for example, David Fand states, “The quantity theory, in its post-Keynesian reformulation, is a theory of the de mand for money and a theory of money income,” “Keyne sian Monetary Theories,” p. 561. Also, he writes, “. . . the modem quantity theory uses the money demand function to predict the level of money income and prices if output is given, or changes in money income if output varies with changes in [the money stock],” “Monetarism and Fiscalism,” p. 228. Friedman has written, “The Quantity theorist not only regards the demand function for money as stable; he also regards it as playing a vital role in determining variables that he regards as of great importance for the analysis of the economy as a whole, such as the level of money income or of prices. It is this that leads him to put greater emphasis on the demand for money than on, let us say, the demand for pins, even though the latter might be as stable as the former,” “The Quantity Theory of Money — A Restatement,” p. 16. ity of demand for money balances to be quite low. 15Friedman, “The Quantity Theory of Money — A Restatement,” p. 16. 16On this point Fand writes, “The sharp distinction drawn be tween the supply determined nominal money stock and the demand determined real stock — a key feature of monetar ism —endows the authorities with elfective control over the nominal money stock, while severely limiting the extent, and the circumstances, in which they may hope to influence the real value of this stock. If the former assumption ex tends their control over nominal variables, the latter as sumption severely limits their influence and control on en dogenous variables such as the real money stock.” See “Monetarism and Fiscalism,” pp. 280-81. 17This view is taken by David I. Fand in, “Some Issues in Monetary Economics,” Banca Nazionale del Lavoro Quar terly Review (September 1969), pp. 228-9 and footnote 24, p. 229. Page 13 F E D E R A L R E S E R V E B A N K O F ST. L O U IS Until recently, it was generally thought that they viewed this elasticity to be zero so that the demand for money was linked directly to income as implied by the naive quantity theory. However, such a view has been rejected outright by Friedman and others;18 if it ever was held, the accumulation of empirical evidence to the contrary has ma'de it untenable now.19 Presently, monetarists take the reputedly different views held by themselves and Keynesians on the size of this elasticity as a basis for contrasting inferences about the expected behavior of velocity in response to a monetary shift. A substantial interest elasticity of demand for money, said to be the Keynesian position, is viewed as implying unstable velocity; Keynesians are viewed by monetarists as not being able to “de pend” on the stability of velocity, for as the money stock rises and falls, offsetting velocity changes insu late the rest of the system to a great extent. On the other hand, while not believing velocity to be per fectly constant, monetarists take the position that .. although marginal and average velocity differ, the velocity function is sufficiently stable to provide a re lation between changes in money and changes in money income.”20 In other words, some, but not much, short-run variation in velocity may be ex pected.21 To some monetarists, the essential differ ence between the two positions is summed up in the demand for money-velocity nexus. Fand writes: The post-Keynesian quantity and income theories thus differ sharply in their analysis of the money demand function. In the modem quantity theory it serves as a velocity function relating either money and money income or marginal changes in money and money income . . .; in the income theory, it serves as a liquidity preference theory of interest rates, or of changes in interest rates (if the price level is given and determined independently of the monetary secto r).22 Although it has become fairly common practice to dis cuss the behavior of velocity in terms of the proper ties of the demand-for-money function, it is improper to do so because observed velocity depends on all of the behavior —real and monetary —in the macroeco 18Milton Friedman, “Interest Rates and the Demand for Money,” Journal of Law and Economics (October 1966), pp. 71-86. 19Some of this evidence is summarized in David Laidler, The Demand for Money: Theories and Evidence (Scranton, Pa.: International Textbook Company, 1969). 20Fand, “Keynesian Monetary Theories,” pp. 563-4. 21Monetarists do not necessarily expect velocity to change inversely with changes in the money stock. Friedman re cently has written that “. . . the effect on [velocity] is empirically not to absorb the change in M, as Keynesian analysis implies, but often to reinforce it. . . “A Theoreti cal Framework,” p. 217. 22See Fand, “Some Issues,” p. 228. Page 14 JANUARY 1972 nomic system. This point will be discussed in greater detail below. (4 ) The final monetarist theme which I shall men tion is concerned with the nature of the response of interest rates to a monetary shift. Monetarists distin guish three components in the observed movement of interest rates: a “liquidity” effect, which is the im mediate response before income or other variables have changed, and thus is expected to be in the oppo site direction of the monetary shift; an “income” ef fect, which is the induced reaction of interest rates to the change in income brought about by the monetary impulse, and hence is expected to be in the same di rection as the money stock change; and a “price expectations” effect, which comes about because monetary changes cause lenders and borrowers to anticipate a changing price level and lead lenders to protect themselves against the expected depreciation in the value of their funds by charging higher rates. This last effect would cause market interest rates to change in the same direction as the monetary change.23 In looking back over this summary of monetarist thought, it becomes quite apparent that there is a good deal of truth to Friedman’s contention that the differences between Keynesians and monetarists are essentially empirical rather than theoretical, having to do with the assumptions made about specific aspects of the commonly-accepted structure, the rela tive stability and importance in the analysis of differ ent functional relationships, the sizes of various elasti cities, etc.24 There appears to be little disagreement between the two camps over the specification of Friedman’s basic model.25 And of Brunner’s four 23For a discussion of these distinctions, see e.g. William Gib son, “Interest Rates and Monetary Policy,” Journal of Political Economy (May/Iune 1970), pp. 431-55. 24This position is expressed in several of Friedman’s writings; for example, see Milton Friedman and David Meiselman, “The Relative Stability,” p. 168, and Milton Friedman, “Post-War Trends in Monetary Theory and Policy,” Na tional Banking Review (September 1964), reprinted in M. Friedman, The Optimum Quantity of Money and Other Essays (London: Macmillan and Co., Ltd., 1969), p. 73. 25Not all monetarists view this particular model as an appro priate description on which to build an analysis, however. Brunner recently wrote, “It is useful to emphasize . . . that the logic of the monetarist analysis based on the relative price theory approach requires that attention be directed to the interaction between output market, credit market and Walrasian money market. This requirement cannot be satis fied by the general framework used by Friedman. This framework is the standard IS-LM analysis offered in an es sentially Keynesian spirit. And this very choice of basic framework actually creates the analytical problems clearly recognized by Friedman in his subsequent discussion. . . . Our analysis . . . established however that the standard IS-LM diagram is not a very useful device for the analysis of monetary processes.” Karl Brunner, “A Survey of Selected Issues in Monetary Theory,” p. 82. F E D E R A L R E S E R V E B A N K O F ST. L O U IS points, at least two are essentially empirical (points numbered (2 ) and (3 ) above), while one of the re maining two (point (1 ) above) makes a distinction between monetarist and Keynesian views of the trans mission mechanism which I believe is false with re spect to current post-Keynesian income-expenditure analysis. Only his last point — that it is appropriate to study the relationship between policy instruments and economic activity without depending on knowledge of “allocative detail” — appears to be one about which there are genuine differences at the theoretical (or perhaps more properly, the methodological) level. Finally, among the four monetarist themes mentioned above, the third one is clearly empirical in nature, and monetarists and Keynesians both in fact hold that this elasticity is nonzero but small in absolute value. In the next section, it is demonstrated that modern Keynes ians take the price level to be endogenous, which sug gests that the monetarist-Keynesian distinctions sum marized above as the second theme are not valid. I shall try to show below that monetarist emphasis on the importance of the demand-for-money relationship (the first theme) is unwarranted, at least in so far as this relationship is viewed as the basis for predicting velocity. I shall also show that the two components of interest rate change in response to a monetary impulse identified in theme four as monetarist are either clearly present in or at least consistent with Keynesian analy sis and assumption. Monetarism, Keynesianism and the Price Level As already noted above, monetarists see one of the essential differences between the two sides to be the question of the determinants of the price level in comparative static equilibrium analysis. Keynesians are said to take prices to be fixed so that monetary shifts are reflected in output changes, while quantity theorists believe that monetary changes affect only the price level in this sort of analysis, with real output being determined by a separate subsector of the system. There is no doubt whatsoever that many practi tioners of the Keynesian viewpoint have assumed that prices could conveniently be taken as given for some problems — especially those associated with substan tial unemployment — and that it has often been con venient for simplicity of exposition in undergraduate classroom exercises or for other purposes to make the assumption of rigid prices. It is quite dubious, how ever, that this assumption, or the liquidity trap as sumption which also has been an important element JANUARY 1972 in the monetarist view of Keynesianism, reflects the thinking of most Keynesian economists today.26 Rather, the standard static “complete Keynesian sys tem” is widely recognized to be one in which the general price level is one of the variables determined by the interaction of the system, and hence is free to move, but to be one in which there are imperfections in the labor market —most typically, a money wage rate which is inflexible downwards. In other words, rather than assuming that prices are fixed as a means of making the simple static model determinate, mod em Keynesians introduce an aggregated labor market and production function into the analysis.27 This could be viewed as the Keynesian equivalent of the “Wal rasian system of equations” asserted by Friedman to be the hallmark of the adherents to the modern quan tity theory approach. It is of course much less satis factory in that all labor market activity and all kinds of production are aggregated into perhaps as few as two equations (i.e., a reduced-form labor market equation and an aggregate production function) rather than having each market and each activity represented by specific equations. It is more satisfac tory on two counts: first, the equations at least are explicitly specified, and second, these equations do not yield the full employment outcome, as is typically the case when depending on a Walrasian system.28 26The liquidity trap is rejected by most economists today be cause little support for it has been found in the many empirical studies of the demand for money which have recently been made. For a summary of some of this evi dence, see Ronald L. Teigen, “The Demand for and Sup ply of Money,” W. L. Smith and R. L. Teigen, eds., Readings in Money, National Income, and Stabilization Policy, rev. ed. (Homewood, 111.: Richard D. Irwin, Inc., 1970), Table 2, p. 98, or “The Importance of Money,” Bank of England Quarterly Bulletin (June 1970), pp. 159-198. 27As evidence for the assertion that modern post-Keynesian static analysis in its most general form typically assumes the price level to be an endogenous variable, and that the system of equations usually is made determinate by introducing a supply subsector consisting of a labor market and aggregate produc tion function, the following standard works are cited: Gardner Ackley, Macroeconomic Theory (New York: Mac millan, 1961), Chap. IX; R.G.D. Allen, Macro-Economic Theory (London: Macmillan, 1967), Chap. 7, esp. sections 7.6-7.8; Martin J. Bailey, National Income and the Price Level, 2nd. ed. (New York: McGraw-Hill, 1971), Chap. 3, esp. sec tion 2; Robert S. Holbrook, “The Interest Rate, the Price Level, and Aggregate Output,” in W .L. Smith and R.L. Teigen, eds., Readings in Money, National Income, and Stabilization Policy, rev. ed.; Franco Modigliani, “The Mone tary Mechanism and its Interaction with Real Phenomena,” Review of Economics and Statistics (February 1963 Sup plement); and Warren L. Smith, “A Graphical Exposition of the Complete Keynesian System,” Southern Economic Journal (October 1956), reprinted in W. Smith and R. Teigen, eds., Readings in Money, National Income, and Stabilization Policy, rev. ed., as well as in several other standard collections of readings in macroeconomics. 28This discussion is not meant to imply that the simple static Keynesian system contains an adequate description of the processes which determine the price level. It states simply that the price level is an endogenous variable in the moael. Page 15 JANUARY 1972 F E D E R A L R E S E R V E B A N K O F ST. L O U IS The essential difference in this regard between Keynesians and monetarists therefore would appear to be that the former view all prices (including wages) as flexible, while the latter consider all prices except the money wage rate to be flexible (money wages are viewed as inflexible, at least in a down ward direction, due to such structural phenomena as minimum wage laws, union contracts, and the like). This distinction has significant implications for the analysis. In the first place, the Keynesian treatment now cannot be said to be fundamentally less satisfactory than the monetarist one in terms of methodology, ex cept perhaps on grounds having to do with problems of aggregation ( Friedman, it will be recalled, used the pejorative term “deus ex machina” to describe what he understood to be the Keynesian approach). Rather, the difference now lies in the analytic usefulness of the assumptions themselves. Is it more appropriate to assume that wages and prices are flexible, or that money wages are sticky while prices can adjust? The answer to this question depends on the nature of the problem being studied in any particular case, and this suggests that an important difference between the two schools of thought may be that Keynesians are more concerned with short-run analysis (for in stance, that related to countercyclical stabilization) while monetarist assumptions are more consistent with long-run analysis. Second, dropping the rigid-price assumption tends to reduce the basis for the heavy emphasis placed by monetarists on the demand-for-money function and its properties. One place where such emphasis is evident is in the discussion of velocity. We turn next to an inquiry into the factors affecting velocity, with partic ular emphasis on the relationship of velocity to the demand-for-money function. general demand-for-money function ( — = L(y, r)) would be characteristic of both. This point can be demonstrated quite easily. First we note that the definition of velocity implies the following relationship: (4 ) E = E y«M o + E yM 0 -1 , p *m 0 where E stands for elasticities calculated on the basis of the interaction of the entire structure, so that (for instance) E ,. m0 represents the elasticity of real out put with respect to changes in the nominal money stock when the response of the entire economic system to the money stock change is taken into account. To distinguish such “systemic” elasticities from “partial” elasticities —those calculated along one function only — the symbol r] will be used to represent partial elasticities. Thus, for instance, r]L r will stand for the > interest elasticity of the demand for real balances, holding income and other variables constant. Under the monetarist assumption of flexible wages and prices, real output is determined uniquely by Friedman’s “Walrasian system” and, as he points out, is to be considered as predetermined from the stand point of equations (1) - (3). This means that a mone tary shift cannot change real output (i.e., the multi plier = 0 ), so that E ,.M , which is defined to be w j dM o 0 ~dM ~’ a^ 's zero- By differentiating equations (1) S0 -(3) with respect to M0 while holding y constant, it is easy to show that the elasticity E P , which is equal .M t o Mo dp j ias a v a Jlle 0 f unity. Inserting these results p dMo The Demand-for-Money Function and Velocity into (4 ) gives the quantity theory result that E V = 0, .M the “stable velocity” result referred to previously. It is important to note that no particular assumptions unique to the monetarist position were made about the demand for money per se; the assumption which yielded this result was that the demand for labor and the supply of labor both were functions of the real wage rate, and that the market was always cleared. Monetarists, as we have already noted, tend to think of the demand-for-money function as a “stable velocity function” while holding that Keynesians view velocity as unstable, justifying this position by appeal to contrasting assumptions about the price level and the interest elasticity of demand for money (see e.g. the quotes from Fand and others above). The fact of the matter is that the behavior of velocity under the two approaches in response to a monetary shift de pends basically on the assumptions made about the labor market, not about the demand for money or about prices, since, as we have seen, both approaches take prices as flexible and, if that is the case, the same On the other hand, let us consider the Keynesian case, which we now define as one in which money wages are sticky (i.e., there exists money illusion in the supply of labor), but in which the price level is an endogenous variable. To analyze this case, we must add three equations to the basic model: an aggregate production function (equation (5 ) below); a labor market summary equation which states that the sup ply of labor services per unit time (N ) is infinitely elastic over a wide range of employment at whatever money wage rate prevails, and that the demand for labor ( N D) is determined by the real wage (w ) ( equation ( 6 ) ) ; and a definition which states that the Page 16 F E D E R A L R E S E R V E B A N K O F ST. L O U IS JANUARY real wage is the ratio of the money wage rate (W ) and the price level ( equation ( 7) ) . The bar over the money wage rate indicates that it is being held constand here.-” This gives: (5 ) y = y(N ) (6 ) N = ND(w ) By differentiating the system defined by equations ( l ) - ( 3 ) and (5) - (7) totally with respect to M0, expressions for the systemic elasticities E ,.m o and E p.m0 can be found. They are as follows (see the appendix for their derivation): (8 ) E y .M = ----------------------- 1---------------------T1 T1 S*y L *r ------------------+ n —n l n — -------------------- w n n D _ rn n y*N n u N • w n S»y w n L »r ----------- + -T l St T1 — 1 To summarize, the main point of this exercise was to show that, using a common model with no special assumptions about the properties of the demand for money, it has been possible to derive “monetarist” and “Keynesian” results for the response of velocity to a monetary shift. It is improper to speak of the demand for money as a “velocity function”, especially in the monetarist case where it is assumed that money wages are flexible so that the system equilibriates at full employment. In that case, the velocity elasticity will be zero no matter what the sizes of the demand-formoney elasticities. Eliminating the rigid-price assumption as a basic point of difference between the two schools reduces the basis for monetarist emphasis on the demand for money for other reasons besides its implications for velocity. It also is important for monetarist views on differences in the nature of the transmission mech anism for monetary policy. It is to this subject that we turn next. L .y Here S stands for the savings function; otherwise all of the notation has already been defined. The usual slope assumptions are made, and on the basis of these assumptions, both of these systemic elasticities will be positive.30 Whether velocity will rise, fall, or remain constant in the face of a monetary shift depends on the sizes of all of the partial elasticities and their relationships to one another as given by these expres sions. The demand-for-money elasticities play a role, but are by no means the only relevant elasticities. In general, we would not expect the elasticity of velocity with respect to nominal money balances to be minus unity in value, as the “liquidity trap” assumption im plies. It will approach that value if ^ L r or ^ s are very large, or if the term ( r|^ —r|^ ) is very close to 29This is the simplest method of introducing a Keynesian-type assumption into the analysis; it is by no means the only possible way of doing so. The nature of and reasons for the existence of money illusion in the labor market is the subject of a considerable amount of literature. See, for example, Axel Leijonhufvud, On Keynesian Economics and the Economics of Keynes (London: Oxford University Press, 1968). 30It is assumed that T )g>r is either positive or, if negative, that it is smaller in size than the absolute value of n , . A 11»r listing of all the slope assumptions is given in the appendix. :;|Since the numerator of the expression for rig>y is one minus the MFC, r|g,y is not expected to be large. As noted in footnote 26, belief in a very large interest elasticity of demand for money (tlL.r ) is not a characteristic Keynesian stance. Reference to the summaries of available empirical evidence mentioned in that footnote will show that this elasticity actually appears to be rather small (almost cer tainly less than unity in absolute value, and in many studies smaller in absolute value than 0.2). 1972 The Transmission Mechanism for Monetary Impulses One of the most characteristic themes of mone tarism is the heavy emphasis which is placed on dif ferences between the quantities of money demanded and supplied as the prime factor motivating spending and, hence, changes in income and prices. Friedman and others have explained again and again how the authorities can change the nominal money stock, but how it is money holders who determine the velocity with which that stock is used, and ultimately who determine the stock of real balances through the ef fects of spending decisions on the price level. As Friedman puts it, “The key insight of the quantitytheory approach is that such a discrepancy [between the demand for and supply of money] will be mani fested primarily in attempted spending, thence in the rate of change in nominal income.”32 In other words, when households and firms are holding more cash balances than are desired at current levels of income and interest rates, they convert these excess balances into other assets, both financial and physical; the market value of physical assets ultimately changes, making the production of new assets more attractive. The change in the general price level which occurs as a result of this process, and the change in output, both work toward a re-equating of the real value of the nominal money stock and the demand for real bal ances. Thus the monetarists clearly embrace a very 32Friedman, “A Theoretical Framework,” p. 225. Page 17 F E D E R A L R E S E R V E B A N K O F ST. L O U IS general kind of portfolio adjustment view of the trans mission mechanism in which the relevant portfolio contains financial and physical assets of all kinds.33 It will be recalled that this is the first of Brunner’s four “defining characteristics.” At the same time, monetarists have been taking Keynesian analysis to task for focusing almost exclusively on interest rates representing the “cost of finance” as the channel through which monetary impulses are felt. The fol lowing quotation makes these distinctions very clear: The Income-Expenditure theory of the Fiscalists adopts a particular transmission mechanism to an alyze the effects of a change in the money stock (or its growth rate) on the real economy. It assumes that money changes will affect output or prices only through its effect on a set of conventional yields — on the market interest rate of a small group of finan cial assets, such as government or corporate bonds. A given change in the money stock will have a calcul able effect on these interest rates . . . given by the liquidity preference analysis, and the interest rate changes are then used to derive the change in invest ment spending, the induced effects on income and consumption, etc. Monetarists, following the Quantity theory, do not accept this transmission mechanism and this liquidity preference theory of interest rates for several reasons: First, they suggest that an increase in money may directly affect expenditures, prices, and a wide vari ety of implicit yields on physical assets, and need not be restricted to a small set of conventional yields on financial assets. Second, they view the demand for money as determining the desired quantity of real balances, and not the level of interest rates. Third, and most fundamentally, they reject the notion that the authorities can change the stock of real balances — an endogenous variable — and thereby bring about a permanent change in interest rates. . . . Monetarists reject the liquidity preference interest rate theory because it applies only as long as we can equate an increase in nominal money with a permanent increase in real balances. This suggests that the liquidity preference theory may be useful as a theory of the short run interest rate changes — the liquidity effect — associated with the impact effects of nominal money changes.34 Statements like this, and the quotation from Fried man in footnote 14 indicate that monetarists believe their view of the transmission mechanism to differ from the position they impute to the Keynesian camp most essentially in differences in assumptions about characteristics of the demand-for-money function. The interpretation of the interest rate term in this function plays a role; so does the question of price flexibility. 33A description of the classes of assets involved and the na ture of their yields is given in Milton Friedman, “The Quantity Theory of Money — A Restatement.” 34Fand, “A Monetarist Model,” pp. 280-81. Digitized for Page 18 FRASER JANUARY 1972 As the preceding discussion and quotation indicate, monetarists think of their own view as an extremely general one. The interest rate term in their model really stands for a vector of yields on many assets, some of them financial yields determined in the money and capital markets, and some of them implicit yields on real assets. A monetary impulse sooner or later affects all of these yields, and hence adjusts the demand for real balances directly as well as indirectly through the effects of yield changes on income. At the same time, changes in the price level which result will adjust the real value of the nominal money sup ply. Therefore the adjustment process is seen as being summarized in the characteristics of the demand-forreal balances function and its relationship to the nominal money supply. Keynesians are said to include only a few market-determined yields on financial as sets in their liquidity-preference function; further more, the price level is exogenously determined. Therefore the process of adjustment to a monetary impulse is supposedly seen by them in much nar rower terms —the entire process takes place through adjustment of the demand for money, and basically is said to focus on the cost of credit as reflected in mar ket interest rates. Furthermore, the belief in a sub stantial interest elasticity of demand for money, often attributed to Keynesians, means that a monetary im pulse will have a relatively small effect even on these rates. These distinctions must be regarded as artificial. First, there is nothing inherent in the Keynesian sys tem which is inconsistent with the introduction of a general portfolio adjustment transmission mechanism; and, indeed, there has been a substantial development in this direction in Keynesian thinking and practice during the last several years. On the theoretical side, the work of Tobin and others may be cited, while at the operational level, the developers of the Federal Reserve Board-MIT econometric model of the U. S. economy have attempted to incorporate such a mech anism into their model.35 While all of the problems in volved in this attempt have not yet been solved, work is continuing and improvements will be made. Second, as we have already shown, Keynesians take the price level to be endogenous, and thus recognize 35For a non-monetarist example of the development of portfolio theory, see James Tobin, “An Essay on Principles of Debt Management,” in Commission on Money and Credit, Fiscal and Debt Management Policies (Englewood Cliffs, N.J.: Prentice-Hall Inc., 1963), pp. 143-218, esp. Part II. Features of the Federal Reserve Board-MIT model are discussed in Frank de Leeuw and Edward M. Gramlich, “The Channels of Monetary Policy,” Federal Reserve Bulletin (June 1969), pp. 472-91. F E D E R A L R E S E R V E B A N K O F ST. L O U IS the same process of adjustment of the nominal money supply through price level changes as the monetarists.38 There remain certain problems with monetarist thought on two subjects related to the transmission mechanism. One is a misunderstanding, in my opinion, of the relationship between money and interest rates implied by Keynesian theory. The other has to do with the monetarist position on the money stock as a force driving income through the portfolio process mentioned above. Liquidity preference theory, money, and the rate of interest — Monetarists view themselves as holding a “monetary theory of the price level” under which monetary shifts are reflected (in the longer run, at least) primarily in price level changes. They take the stance that Keynesians hold a “monetary theory of the interest rate.” Under this phrase, at least two posi tions are subsumed. Some monetarists seem to think that Keynesians see the money supply together with the demand-for-money function (specified in nominal terms) as determining the level of interest rates. Others recognize that the interest rate in Keynesian analysis is determined jointly as one of the outcomes of an interacting system of relationships rather than just by one behavioral relationship (i.e., by some ver sion of an IS-LM system like Friedman’s summary model). Whichever view is held, however, it is as serted that Keynesian analysis leads to the conclusion that monetary shifts result in interest rate changes in the opposite direction, while monetarist analysis sug gests that movements of M and r in the same direction will be observed.37 Neither version of the “monetary theory of the in terest rate” is an accurate representation of Keynes 36Semantic as well as real issues are involved in discussions of this subject. For example, Brunner labels anyone who sub scribes to a portfolio adjustment view of the monetary trans mission mechanism a “weak monetarist”. See Karl Brunner, “The Role of Monetary Policy,” this Review (July 1968), pp. 9-24. 37As an example of the first of these positions, the following quotation from a recent article by Fand is offered: “In the Keynesian theory the exogenously given quantity of money, together with the liquidity preference function, determines the interest rate.” Fand, “Keynesian Monetary Theories,” p. 564. The second is illustrated by a quotation from Zwick: “The alternative concepts of Keynes and Fisher con cerning the adjustment of the economy to monetary changes are mirrored in their different notions concerning interest rate determination and the response of interest rates to monetary changes. The IS-LM framework suggests that, so long as the IS and LM schedules represent independent relations, a monetary expansion causes interest rates to fall because of the outward shift of the LM schedule. In tiie Fisherian model, a monetary increase raises the level of expenditures; the upward response of loan demand due to the increased expenditures causes interest rates to rise.” Burton Zwick, “The Adjustment of the Economy to Mone tary Changes,” Journal of Political Economy (January/ February 1971), p. 78. JANUARY 1972 ian thought, for both imply that an expansionary monetary impulse (for example) can only result in a lower interest rate in the new equilibrium. In other words, it appears that of the two monetary effects on interest rates often mentioned by monetarists which are relevant for static analysis —the liquidity effect and the income effect — Keynesians are supposed to recognize only the liquidity effect, or more generally, are supposed to be basing their analysis on assump tions which can only result in an inverse relationship between monetary impulses and interest rate changes. This is certainly not the case. When the entire struc ture is taken into account, rather than only the liquid ity preference function, the level of interest rates in the new equilibrium relative to the initial position is determined by a number of elasticities, most impor tantly those which are the determinants of the slope of the IS curve. If its slope is positive —which is the case if all of the propensities to spend with respect to total income sum to more than unity — then both income and interest rates will be higher in the new equilibrium than in the old.38 This is such a wellknown case as to require no further comment. Of course, equilibrium positions are not observed in the real world; instead, the economy is always in transition, moving toward resting points, which them selves are repeatedly being disturbed. It may be in ferred from some monetarist writings that it is the observed tendency of interest rates and money to move in the same direction which is thought to be inconsistent with Keynesianism, rather than the possi bility that money and interest rates can move together in terms of comparative equilibrium points. In other words, the discussion may refer to the dynamics of the system, rather than the comparative statics. In this area, the monetarists have done us all a service by stressing the possible importance of price-expectation effects on interest rates, a phenomenon which typi cally has not been incorporated into dynamic Keynes ian models. I will argue that observed parallel move ments between money and interest rates are quite con sistent with the basic IS-LM structure (no matter which way the IS curve slopes), given the reasonable and widely-accepted premise that the monetary sector adjusts much more rapidly than the real sector to ex38An upward-sloping IS curve cannot be obtained from Fried man’s summary model, because only consumption spending is related to income in that model, and the notion that the MPC is less than unity is a fundamental postulate of macro economic analysis. However, the level of income might well appear in other expenditure functions, such as the invest ment relationship (where the rationalization would be that investment depends on profits, which in turn are a function of the level of income). Page 19 F E D E R A L . R E S E R V E B A N K O F ST. L O U IS JANUARY 1972 temal shocks. Under this premise, observed values of income and the rate of interest may be supposed, at least approximately, to be such that the LM equation is always satisfied during the process of adjustment from one equilibrium to another, while the IS equa tion is not. I will argue further that price-expectation effects are readily accommodated by this analysis. The implications of these differing speeds of ad justment are illustrated on the accompanying figure, which happens to be drawn with a downward-sloping IS curve. Assume the system to be initially in equili brium at point F, so that the equilibrium values of the interest rate and income levels are r and y. Now let there occur an expansion of the money supply, so that the LM curve shifts outward to a new position, LM'. According to the assumption made above concerning the relative speeds of adjustment of the monetary and real sectors, this shift will result first in a fall in the interest rate from its initial equilibrium level to a new level, r'. It should be noted that this is the “liquidity effect” which is recognized by monetarists as being present both in then- own and in Keynesian thinking. It represents a movement along the liquidity prefer ence function in response to a change in the money supply, holding income constant. Next, income will begin to respond, and income and the rate of interest both will rise along the segment GH of LM ' to point H, the final equilibrium position. This movement, of course, reflects the “income effect.” If rising income is accompanied by rising prices, there will also be an induced shift of the LM curve during the transition. For example, it might move to a position like LM " as shown. Alternatively, it could move to a position to the right of LM'. Such LM shifts reflect the operation of two forces. First, rising prices reduce the real value of the new nominal money stock and “tighten the money market” after the initial expansionary pulse. This has the effect of moving the LM curve leftward. Second, rising prices may engender expectations of future price increases. If, as has been suggested, the demand for money depends on nominal interest rates while real expenditures are determined by real rates, then the “price expectations effect” mentioned previously would cause a rightward LM shift, resulting in a lesser leftward overall shift in the LM curve than that brought about due only to the drop in the real value of the nominal money stock, or perhaps even a net rightward movement (in this discussion, the vertical axis is interpreted as measuring the real rate of inter est). If these effects are present, the adjustment path followed from point G might be the dotted one instead of the solidly-drawn one, and the system would end Page 20 up at a point like J instead of H, so that the new equilibrium income level would be y", and the equilib rium real interest rate r". Incidentally, if price-ex pectation effects are present, a value of r" for the real rate is quite consistent with a market rate above r. We may conclude from this discussion that there is no reason to be surprised by the fact that during much of the time following an increase in the money supply, interest rates are observed to rise. A standard assump tion about relative speeds of adjustment, much used by Keynesians, directly reflects both the “liquidity effect” and the “monetary effect” often discussed by monetarists, and is perfectly consistent with the pres ence of price expectation effects. Second, it is appro priate to point out that this entire discussion has been carried out in the context of a pure multiplier model. If accelerator effects are present, they may accentuate the pure multiplier effects of a monetary shift on in terest rates, at least during parts of the adjustment period. Finally, there is the likelihood that in many cases in which interest rates and the money stock move together, the monetary authorities are reacting to shifts in spending. For instance, if total spending rises, interest rates will go up and the monetary au thorities will often try to moderate the interest rate increase by expansionary open market operations, re sulting in a rise in the money stock. The monetarist view of money as a force driving income — It is self-evident that monetarists typically have assigned great importance to changes in the money stock as the prime moving force behind in come changes. For instance, one of Brunner’s “de 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 fining characteristics of monetarism” is that . . the monetarist analysis assigns the monetary forces a dominant position among all the impulses working on the economic process.”'iu And, of course, Friedman’s investigations into the lead-lag relationship between changes in the rate of change of the money stock and changes in income are too well-known to require fur ther comment.40 At the same time, monetarist writ ings often seem to suggest that Keynesians view monetary policy as ineffective. Keynesians view monetary policy as effective and useful, and to suggest the opposite is to raise false issues. But this does not mean that they necessarily consider changes in the money stock to have particu lar causal significance. Monetary policy is carried out through the traditional instruments — open market op erations, discount rate changes, and variations in re serve requirements — and not by direct manipulation of the money stock. It is true that in simplified ver sions of the Keynesian model, monetary policy is represented by the money stock, which is assumed to be controlled by the authorities and which replaces the instruments named above. It is also possible that the authorities could control the nominal money stock to almost any desired degree of precision. But in the real world, or in the more sophisticated models of it, the nominal money stock is not exogenous, nor has it been controlled as an objective of policy by the central bank in the United States; it, or its compo nents, are determined jointly by the central bank, the commercial banks, and the public, and it is basically a passive outcome of the interaction of the economic system, not a driving force. The doubt that Keynesians feel concerning mone tarist assertions about the potency of money stock changes reflects the fact that monetarist descriptions of the adjustment process themselves seem to give no particular reason for regarding money stock changes as causal. These descriptions typically run as follows, using an open market purchase of Treasury bills as an example:41 at the outset, there is an exchange of assets between the central bank and a Government securities dealer, with the central bank giving the dealer its cheek drawn on itself in exchange for bills. S0Brunner, “The ‘Monetarist Revolution’,” p. 7. 40Milton Friedman, “The Supply of Money and Changes in Prices and Output,” in The Relationship of Prices to Eco nomic Stability and Growth, Compendium of Papers Sub mitted by Panelists Appearing Before the Joint Economic Committee, 85th Congress, 2nd sess., 1958, pp. 241-56. +lSee, for instance, Milton Friedman and David Meiselman, “The Relative Stability,” Sec. VII, and Milton Friedman and Anna J. Schwartz, “Money and Business Cycles,” Re view of Economics and Statistics (February 1963 Supple ment), esp. pp. 60-61. JANUARY 1972 This exchange results in the following: (1 ) a reduc tion in the yield on bills, with consequent disequili brium among holders of securities; (2 ) an increase of bank reserves of an equivalent amount (disregarding drains into currency holdings, etc.); (3 ) an initial increase in the money supply of the same amount as the transaction; and (4 ) a decrease in bill holdings by the private sector, with a concomitant increase in the central bank’s portfolio. In a process described in some detail by Friedman and Schwartz, the next step will involve action to readjust portfolios in response to yield and wealth changes; meanwhile, banks will be interested in expanding loans on the basis of their newly-acquired reserves (and incidentally in creating new deposits). Eventually the adjustment affects the yield on equities and therefore the market value of the existing stock of physical capital. The existing capital stock will rise in value, stimulating the produc tion of new capital and thus causing income to rise. There may also be other effects, such as direct effects on spending of changes in wealth. The question would seem to be whether it is the initial increase in the money stock, the full increase (including the new deposits generated as a conse quence of loan decisions), the increase in bank re serves, the reduction in private bill holdings, the fall in yields, the increase in the central bank’s portfolio, or some other factor which is responsible for the in come change. Rather than arbitrarily selecting some one factor from this list, it would seem preferable to take die more general view that the initiating force was the disturbance of a portfolio equilibrium, effected in this case through open market operations. (Such a disturbance, with similar effects, could arise for other reasons: e.g., if there were a change in wealthholders’ preferences for holding a particular security category at existing yields.) The change in the money stock is properly viewed as one of the several results (along with changes in income, interest rates, prices, etc.) of this disturbance. Such a position of course implies that monetary policy is effective, but does not assign the starring role in the drama to changes in the money stock. Stabilization Policy Modem Keynesian static analysis, based on the complete Keynesian system with flexible prices and inflexible money wages, yields the result that both monetary and fiscal policy are able to effect changes in income, interest rates, prices, employment, and other variables. Monetarist analysis, however, takes the position that only monetary policy has significant effects on the pace of economic activity, at least in 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 the short run. This suggests that the two schools of thought disagree not in their views about monetary policy, but rather on the effectiveness of fiscal policy. Until recently, monetarists were interpreted as bas ing their belief that fiscal policy is ineffective directly on the presumed existence of a stable demand-formoney function with zero interest elasticity, together with the assumption of an exogenously-set money stock. Such a demand-for-money function links money and income directly together, so that income cannot change unless the money stock changes. Shifts in gov ernment spending financed by bond issue, for in stance, were said to result in interest rate changes of sufficient magnitude to reduce private spending to the degree required to keep total demand at a con stant level. However, given the many research studies which show otherwise, it has become impossible to maintain that the interest elasticity of the demand for money is zero. This development has had a considerable effect on the tone of monetarist discussions. Thus Fand, in discussing stabilization policy, refers to . . the exceptional case of a completely (interest) in elastic demand for money.42 Furthermore, a relevant recent finding is that the supply of money is interestelastic, and that this is sufficient to loosen the tight link between the money stock and income even if the interest elasticity of demand is zero. Therefore monetarists have had to rationalize their dismissal of fiscal policy in other ways. Some have tried to find other means of solidifying the moneyincome link and of segregating the monetary sector from the remainder of the system by neutralizing the connection provided by the interest rate. One way of doing so is by considering the interest rate to be de termined exogenously. This, in effect, is the procedure followed by Friedman in his paper entitled, “A Mone tary Theory of National Income.”43 If interest rates do not respond to changes in real and financial vari ables, the rigid money-ineome connection is preserved. This may be considered the most extreme approach, because under it fiscal policy does not even affect the rate of interest and the division of output among the various sectors. Another way is to make the standard quantitytheory assumption of flexible wages and prices, and hence full employment, while accepting the fact that the demand for and supply of money balances are interest-elastic. In such a world, fiscal policy cannot 4-Fand, “Monetarism and Fiscalism,” p. 289 (italics added). 43See the discussion of this approach in footnote 12. http://fraser.stlouisfed.org/ Page 22 Federal Reserve Bank of St. Louis JANUARY 1972 affect the levels of real variables like output or em ployment, which are entirely determined by the labor market and the production technology of the system — but then, neither can monetary policy. Assumptions are not a matter of logic, assuming that they are internally consistent. In weighing these various approaches to the analysis of stabilization pol icy, the most important questions probably should be: Which of the alternative approaches is the most realistic and the most relevant for the real-world question of fiscal policy’s effectiveness? Is it the case of flexible wages and prices, so that full employment is the rule and not the exception, and neither mone tary policy nor fiscal policy can affect the level of real activity? Is it the case involving exogenouslydetermined interest rates, so that fiscal policy cannot even affect the division of output, let alone the level of activity? Or is it the case of flexible prices but a sticky wage level, in which case monetary and fiscal policy both are capable of affecting the level of real activity? Brunner has taken a somewhat different approach to the analysis of fiscal policy than have most other monetarists. He asserts that fiscal policy is ineffective or perverse because the effects on asset values due to interest-rate changes of the cumulation or decumula tion of claims against the Government held by the public, resulting from a fiscal policy deficit or surplus, outweigh the direct effects on the flow of output and income of new spending and taxing and of the changes in the stock of financial claims held by the private sector which result.44 This position implies the view that the disturbance of portfolio equilibrium from any source (not only money stock changes) has powerful repercussions, and thus paradoxically tends to downgrade the importance of changes in the money stock. As far as is known, this position is not supported directly by empirical evidence. Summary In this paper, I have attempted to sketch the main outlines of monetarist thought and to examine some aspects of the monetarist view of Keynesian analysis. In doing so, I have paid particular attention to the roles of the instruments of stabilization policy under the two views. My examination of the monetarist-Keynesian debate has indicated that the version of Keynesianism which the monetarists use to establish a contrast for their own point of view is out of date and inadequate — a "Karl Brunner, “The ‘Monetarist Revolution’.” F E D E R A L R E S E R V E B A N K O F ST. L O U IS “vulgar” version of post-Keynesian thinking, to use Professor Johnson’s term. When it is recognized that Keynesianism implies sticky wages and money illu sion in the labor market rather than rigid prices, and that portfolio adjustment as the basis for the trans mission of monetary impulses is not only consistent with the Keynesian approach but indeed is being built into Keynesian models, it is seen that there is very little if anything in monetarist theory which is new and different. Rather, the two approaches di verge in ways which basically are methodological and operational. The monetarists are willing to commit themselves to the use of very simple, very small ( even one-equation) models for policy analysis; Keynesians typically are not. On this point, the monetarist stance seems to be a matter of faith rather than logic; the common theoretical basis on which both positions rest certainly implies the use of a structural approach.45 There certainly are substantial differences in the kinds of operational assumptions that are made about par ticular dimensions of the theoretical structure, and these have implications of various kinds for policy. The typical Keynesian assumption of money wage inflexibility is consistent with a shorter-run analysis; it leads to the conclusion that both monetary policy and fiscal policy can affect the level of activity. The typical monetarist assumption of wage and price flexibility (i.e., of full employment) is more relevant for the analysis of secular changes. This assumption essentially bypasses the whole question of short-run policy effects. For the long run, paradoxically, it suggests that fiscal policy is more important and interesting than monetary policy, for fiscal policy at least changes the rate of interest (un less the rate of interest is exogenously determined), and therefore the division of output, and presumably affects growth; whereas monetary policy affects only prices, money wages, and the like.46 There appear to 4BKarl Brunner has written, “The monetarist disregards . . . the allocative detail of credit markets when examining pat terns of allocation behavior. . . . Such detail is simply as serted . . . to be irrelevant for aggregative explanation.” Ibid., p. 15. 46The reservations expressed in footnote 10 apply to this state ment also. JANUARY 1972 be some analytic confusions in many monetarist dis cussions. I have tried to show above that it is incor rect to view the demand-for-money function as a velocity relationship from either point of view. In the monetarist case, this is especially true because the stability of velocity in the face of monetary changes depends on assumptions about the labor market and is unrelated to the characteristics of the demandfor-money relationship. It also appears that mone tarist fascination with the money stock is unwar ranted by monetarist logic, which seems to me to place great emphasis on portfolio disequilibrium as a potent driving force in the economy. It does not fol low from this view, as a matter of logic, that observed changes in the money stock have any particular sig nificance as a causative force. On the positive side, monetarists have contributed to the development of macroeconomic thought by stressing that the links relied upon for years by most Keynesians to connect the real and monetary sectors overlook entirely the important substitution and wealth effects which are the concomitants of portfolio adjustment. They also have called our attention to the distinction, apparently first made by Irving Fisher many years ago, between market and real interest rates, and therefore to the potentially important role of price expectations in dynamic macroeconomics. These phenomena are extraordinarily difficult to cap ture in empirical models, but work is proceeding along these lines. It is to be hoped that during the next few years, they will be made standard features of Keynesian (that is, structural) theoretical and empirical models, and that dependable evidence will be gathered so that the real questions which divide us — chiefly, in my opinion, the question raised by Brunner and others concerning the need for largescale structural models for aggregative analysis —can be answered satisfactorily. This article and the accompanying one by Robert H. Rasche are available as Reprint No. 74 Appendix begins on following page. Page 23 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 JANUARY 1972 APPENDIX Following are the derivations which underlie equations ( 4 ) , ( 8 ) , and ( 9 ) in the text. They are based on equa tions ( l ) - ( 3 ) and ( 5 ) - ( 7 ) , which are reproduced here for convenience: (1 ) y = C (y,r) + I(r) (2 ) ^ = L (y ,r) B. The Monetarist Case Monetarists assume that wages and prices are flexible so that real output, y, may be considered exogenous for the purpose of static analysis, and only equations ( 1 ) and ( 2 ) are relevant. Differentiating ( 1 ) , which is the IS curve, yields: Mm 0 (3 ) Y = py (5 ) y = y(N ) W (B .2) (C r + Ir) The following slope assumptions are used throughout: O < C y < 1; C,. < O or, if positive, C r < |I,.|; I r < O; L y > O; L r < O; y N > O; N “ < O. dMo dr dy + Lr ( B .3 ) Ly T J-*r J dMo dMo dMo Equation (4 ) in the text is an expression for the elasticity of velocity with respect to a monetary shift, and is reproduced for convenience: = E y*M + E - 1 dY Mo dM„ dY dM0 , d> ' dM(, = E + y dp dM0 + E 1, which is equation (4 ). This result is derived only from definitions. Next we investigate the values of E y. M o and E p.m0 , and there fore of E v.m , which are implied by monetarist and Keynesian assumptions respectively. http://fraser.stlouisfed.org/ Page 24 Federal Reserve Bank of St. Louis dr dM0 = 0, (B .3) reduces to: = E -m g = 1 - Substituting these findings into (A .3 ), we find that E t . m0 = 0 using static analysis under monetarist assumptions. C. The Keynesian Case Keynesians take money wages to be inflexible while prices are an endogenous variable. This means that real income or output may no longer be considered exogenous; instead, it becomes endogenous, and equations ( 5 ) - ( 7 ) are added to the IS-LM system as represented by ( 1 ) and (2 ) in order to close the set of equations. and Ep.ii , we must again differentiate the system totally with respect to M0, now treating y as a variable. In addition Substituting (A .2) into (A .l) and multiplying the re sulting equation by 1 ° yields (A .3) E 7 ^ = To derive expressions for the elasticities E y.M o From ( 3 ) , we have (A.2) (B '4 ) Mo dp . Since we have p2 dM0 i. Thus we have: dV dM0 = 0. P*Mn It is derived by differentiating the expression for velocity (V = ) with respect to the money stock, and convert ing the result into elasticity form. (A .l) clMo Differentiating the LM curve (2 ) yields: A. The Elasticity of Velocity y*M n dy dM0 = 0, which implies that dy found that, in this case, ,,-r (4 ) E dr dM0 However, if y is exogenous to this system, = 0 so ,i , . ClMo that we get: (6 ) N = ND(w ) (7 ) w = + (Cr + Ir) to equations ( B .l ) and (B .3 ), this differentiation yields W dp w p2 dMo , td < > BE “ which is derived by differentiating equations ( 5 ) - ( 7 ) and substituting where possible. It will be convenient to make some further substitu tions. First, since the MPC with respect to income is one minus the MPS with respect to income, and since the JANUARY 1972 F E D E R A L R E S E R V E B A N K O F ST. L O U IS MPC with respect to the interest rate is the negative of the MPS with respect to the interest rate, we make the substitutions (1-C y) = Sy and C r = —Sr, where S stands for the saving function (the model implies S = S ( y ,r ) ) . Second, ( C .l ) can be used to eliminate the term involving ~~ in (B .3 ). Making these substitutions and collecting terms yields the following pair of equations in the two variables and dy (C .2 ) Sy dM0 (C.3) (Ly = (C .6 ) E (Ir — Sr) Mo W yNN £ ) dr dMc n dy + L, dr dM„ ' dM0 gives; n •w £ _ z (b ) to find the systemic elasticity E y.M , both sides of (C .4 ) must be multiplied by — . Carrying out these ^ operations and cancelling terms where possible, we get ^ q < 0 is that S -r l<n (In l ) +n S*y L-r L*y (In - n Ir S-r Thus the larger in value are r)giy, |). and | L>r | , the r| more likely it is that E v. m < 0- Finally, for nonzero but finite values of r)y< and t) ndw , E y.M and E p.M N o owill tend toward zero (and E V M toward — 1) if ri S .y or . 1 O n S‘y L*r S*r pends on a special case of equation (C .5 ) in which the last denominator term approaches zero. W hether E V. M is posi ( tive or negative in this case depends on whether E y M is > ^ greater or smaller than unity. The condition for E V ,M Ir It —n zero and the response of velocity to a monetary shift de In-n "x H z.x’ —n I* r brought about by a monetary shift, so that E = 0 and E p.M = 1, resulting in stable velocity. Second, if o either > 1 or r| D are extremely large, E M approaches (a ) each of the propensities (or partial derivatives) shown in the denominator may be converted into a partial elasticity by using the relationship between any two variables x and z given by the definition of a partial and thus elasticity; i.e., if z = f ( x ) , then r|z.x fv n n n S*y L t in th e system . F irst, if eith er r|y v or r| ndw > are zero, output will not change in response to a real wage change To convert this into elasticity form, two steps are needed: n n y*N ND*w From (C .5 ) and (C .6 ), it can be seen that the behavior of velocity now depends on all of the partial elasticities ________ p__________ dy __ Mo SvLr dM0 + Ly __ Ir—Sr ' ' WyNNw (C .5 ) E y. Mo = n p*Mo = 0 Solving these equations for (C .4 ) To find an expression for the systemic elasticity E p.M , equation (C .4 ) is substituted into ( C .l ) and a systemic expression f o r ^ j ^ is derived. When this expression is multiplied by , the partial derivatives are converted to elasticities, and the necessary algebra is carried out, the following expression results: S*r r|L*r '[ are very large, or if (ri i* r —n o * r.) is very close to a zero in value. A large value for f] L>y would also give this i + n L,)' n n y*N ND*w result. Page 25 Comments on a Monetarist Approach to Demand Management by ROBERT H. RASCHE Robert H. Rasche is Assistant Professor of Economics at the Wharton School o f Finance and Commerce, University of Pennsylvania. He received a PhD degree in Economics from the Uni versity of Michigan. Professor Rasche has been closely involved in the development of the FRBMIT econometric model of the United States economy. The author is on leave during the 197172 academic year and is a visiting scholar at the Federal Reserve Bank of St. Louis. This paper was presented at the Annual Conference of College and University Professors o f the Federal Reserve Bank of St. Louis on November 12, 1971. CASUAL reading of the popular discussion of stabilization policy over the past four or five years would suggest that the definition of a monetarist was firmly established. In the monetarist camp are Milton Friedman, Karl Brunner, Allan Meltzer, and the model of the St. Louis Federal Reserve Bank. Among the nonmonetarists are Walter Heller, Gardner Ackley, Arthur Okun, James Tobin, and the large econome tric models such as the Wharton model and the FRBM IT model. Sometimes the distinction between the two groups has been summarized in the allegation that a monetarist is one who not only believes that money matters, but also believes that money is the only thing which matters.1 A close reading of the writings of those associated with both points of view, suggests that distinctions are not completely clear at the level of monetary theory. Leonall C. Andersen has characterized the mone tarist position on stabilization policy as holding that “the major impact of monetary actions is . . . on longrun movements in nominal economic variables such as nominal GNP, the general price level, and market in terest rates. Long-run movements in real economic variables such as output and employment are con sidered to be little influenced, if at all, by monetary actions.”- On the other hand he admits a clear role for fiscal policy, if not the conventional stabilization policy role: “their [fiscal actions] main impact is on long-run movements of real output. . . . In the short run, fiscal actions . . . exert some but little lasting in fluence on nominal GNP expansion and, therefore, have little effect on short-run movements of output and employment.”3 1Walter W. Heller, “Is Monetary Policy Being Oversold?”, in Milton Friedman and Walter W. Heller, Monetary vs. Fiscal Policy (New York: W.W. Norton & Co., Inc., 1969), p. 16. 2Leonall C. Andersen, “A Monetarist View of Demand Man agement: The United States Experience,” this Review (Sep tember 1971), p. 4. 3Ibid., p. 4. Page 26 The question is what theoretical framework can produce these types of conclusions, and can it be tested? Again quoting the Andersen paper, “monetary actions . . . are considered a disturbance which influ ences the acquisition of financial and real assets. Rates of return on real and financial assets and market prices adjust to create a new equilibrium position of the economy; therefore these changes are considered the main channels of monetary influence on aggregate demand.”4 Thus the monetarist conception of what has been called the transmission mechanism is one of monetary disturbances which change interest rates and the rela tive prices of real and financial assets. Such changes induce a reallocation of asset portfolios which can in clude changes in the demand for real assets. Finally, the portfolio adjustments and relative price changes can change the demand for consumables. In an earlier article in this Review, Karl Brunner characterizes a similar position as the “weak monetarist thesis.”5 This construct of the world is apparently one which is widely accepted among monetary economists today and thus does not discriminate among the monetarist and nonmonetarist positions. Certainly a whole suc cession of writings by James Tobin suggests an ex planation quite consistent with this view of the trans mission mechanism of monetary policy.6 In fact, An dersen admits that he would view his mechanism as “close to the Tobin view, except that it takes into consideration many more rates of return and market prices of goods and services.”7 An examination of the *Ibid., p. 3 (italics are added). 8Karl Brunner, “The Role of Money and Monetary Policy,” this Review (July 1968), pp. 18-19. 6James Tobin, “An Essay on Principles of Debt Management,” in Commission on Money and Credit, Fiscal and Debt Management Policies (Englewood Cliff, N.J.: Prentice-Hall Inc., 1963) pp. 143-218; and, “A General Equilibrium Ap proach to Monetary Theory,” Journal of Money, Credit and Banking (February 1969) pp. 15-29. 7Andersen, “A Monetarist View,” p. 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 writings of other nonmonetarist economists will show similar consistencies with this view of the transmission mechanism. Therefore “weak monetarism,” as a the oretical position, does not appear to be a monopoly of the monetarists. Given this apparent agreement on the theoretical basis of the mechanism through which monetary policy actions affect the economy, one can question whether the “monetarist counterrevolution” is more than an attempt at product differentiation, such as economists usually associate with monopolistic competition. A pragmatic view of the discussion suggests that at least four substantive issues are involved: (1 ) the usefulness of the IS-LM aggregate demand framework for policy formulation; (2) the dynamic adjustment of the econ omy to a new equilibrium after a policy shock; (3) the mode of conduct of monetary policy; and (4 ) an econometric issue of large versus small models. Limitations of Policy Prescriptions from the IS-LM Framework A major source of monetarist criticism has been the use of the IS-LM framework for aggregative policy analysis. In this Review, Ronald Teigen has attempted to defend the IS-LM framework from one monetarist accusation that this framework holds that an increase in the stock of money lowers the interest rate and raises output.8 He demonstrates that with certain assumptions about the relative speeds of adjustment of various markets, it is possible to show that interest rates over time will first fall and then rise again as the system returns to a new equilibrium. At the same time Teigen admits that this framework has ignored price expectations, and in addition, that it is not easy to incorporate price expectations, a dy namic phenomenon, into the static framework. This appears to sidestep the crux of the monetarist com plaints. Not only does the conventional IS-LM analysis ignore price expectations, but it usually ignores effects from changes in the level of prices. The omission of such price level effects is possible only when the macroeconomic model “ specified totally in terms of is real flow variables. In sophisticated analysis, such as that of Martin Bailey, price level effects of various kinds are introduced, and it can be shown that the position of either the IS or the LM curve (in the interest rate - real income plane) is dependent upon the current price level.9 8Ronald L. Teigen, “A Critical Look at Monetarist Economics,” this issue of the Review, pp. 19-20. 9Martin Bailey, National Income and the Price Level (New York: McGraw-Hill, 1962). JANUARY 1972 Most macro-economists will acknowledge the valid ity of the price level effect on the LM curve arising from the specification of the demand for money as a demand for real balances. Similarly, a specification of the consumption function in terms of income and wealth as implied by a permanent income or “life cycle” hypothesis will generate a family of IS curves, one for each level of real wealth.10 Once both are con structed as functions of the price level, any policy action which generates a change in the price level will not only have a direct impact on the IS (fiscal policy) or the LM (monetary policy) curve, but also will cause additional shifts in both curves through the changes in the price level. Under these circumstances, simple multiplier calculations do not adequately rep resent the reaction of the economy to the policy action. Accurate policy conclusions cannot be derived with out estimates of the parameters of the system. The situation is further complicated when it is as sumed that monetary policy can affect the market value of assets, such as corporate equities, through induced changes in the rate of return on these assets. Then the specification of a consumption-wealth rela tionship implies that any change in interest rates will induce a shift in the IS curve. Teigen has already indicated that it is difficult, if not impossible, to incorporate dynamic phenomenon such as price expectations into the static framework of this construct.11 Yet, as the monetarists have rightly pointed out, adjustments in specified behavioral rela tionships have to be made for such expectations under conditions of anticipated inflation (and particularly when the rate of inflation is anticipated to be changing). It appears that the monetarists have justifiable com plaints with the policy analysis derived from this framework, which is common to popular macroeco nomic textbooks and past annual reports of the Council of Economic Advisers, even if one is pre pared to accept the proposition that there are no deep theoretical differences in the transmission mechanism for monetary and fiscal policy. In many respects the monetarist attack on the IS-LM framework is fighting a “straw man.” Many of the limitations of this highly aggregative framework, such as those indicated above, have been attacked in 10It is not necessary to assume this kind of consumption func tion to generate such a family of curves. Price elasticities of imports and exports with respect to domestic and foreign prices, or income tax functions which have nominal income elasticity greater than one, will generate the same effect. 11Teigen, “A Critical Look,” p. 19. Page 27 F E D E R A L R E S E R V E B A N K O F ST. L O U IS larger econometric models of income determination. This is not to say that these models have captured the various effects with a high degree of precision. In particular, the econometric problem of estimating dis tributed lags has proven particularly difficult, thus the timing of responses to policy actions as implied by large statistical models is a major area of uncertainty at the present time. The Process of Adjustment to Policy Actions A close reading of the monetarist evaluation of the relative strengths of monetary and fiscal policy leaves the impression that they are not quite talking about the same thing as the nonmonetarists. This can be illustrated by the first quote from Andersen, above, which sets out the transmission of monetary policy on the economy as the adjustment of rates of return and prices to a new equilibrium ( in the absence of further exogenous shocks). A similar characterization has been made in a recent analysis by Friedman in which he distinguishes “Keynesian” analysis as a framework in which prices are assumed exogenously constant.12 Most empirical models of income determination to day regard the price level as endogenously deter mined, with the independent variables in the price level equation specified as money wage rates and productivity separately or in a composite form as unit labor costs.13 In addition, money wage rates are usu ally assumed to be endogenous variables in such models. The behavior of wage rates in such models is usually specified to follow a modified version of the “Phillips Curve.”14 This specification is a disequilib rium mechanism which holds that, in the presence of an excess demand or supply in the labor market ( usu ally measured by the unemployment rate), money wage rates will change. However, it says nothing about the nature of the equilibrium toward which the market is presumably adjusting. This is a modification of the early post-Keynesian models in which the money wage rate was taken theoretically, if not em pirically, as fixed in the short run. 12Milton Friedman, “A Theoretical Framework for Monetary Analysis,” Journal of Political Economy (March/April 1970), pp. 193-238. 13See Lawrence R. Klein and Michael K. Evans, The Whar ton Economic Forecasting Model, 2nd ed., (Philadelphia: University of Pennsylvania, 1968); Robert Rasche and Harold T. Shapiro, “The F.R.B.-M.I.T. Econometric Model: Its Special Features,” American Economic Review, Papers and Proceedings (May 1968), pp. 123-149; and Otto Eck stein and Gary Fromm, “The Price Equation,” American Economic Review (December 1968) pp. 1159-1183. 14A. W. Phillips, “The Relationship Between Unemployment and the Rate of Change of Money Wage Rates in the United Kingdom, 1861-1957,” Economica (November 1958), pp. 283-299. Page 28 JANUARY 1972 It seems appropriate to conclude that the mone tarists and their opponents are discussing policy effects on the economy over two different time spans. The nonmonetarists have implicitly concerned themselves with models in which the labor market, at least, re mained in disequilibrium. The monetarists on the other hand, in discussing policy impacts when the “new equilibrium position of the economy” has been achieved, implicitly appear to be concerned with the period of time in which all markets, including the labor market, have adjusted themselves to the policy shock. The recent reinterpretation of the General Theory by Leijonhufvud offers an explanation of this debate in terms of the dynamics of the labor market. He argues: The revolutionary impact of Keynesian Economics on contemporary thought stemmed in the main, we have argued, from Keynes’ reversal cf the conven tional ranking of price and quantity velocities. In the Keynesian models price velocities are not infinite; it is sometimes said that the implications of the model result from the assumption that money wages are “rigid”. This usage can be misleading. Income-con strained processes result not only when price-level velocity is zero, but w henever it is short of infinite.15 He further argues that this is an appropriate as sumption if there are substantial information costs as sociated with trading in the labor market, as recent theories of labor market behavior have postulated:16 In the absence of perfect knowledge on the part of transacting units or of any mechanism unrelated to the trading process itself that would supply the needed information costlessly, the presumption of infinite price velocity disappears.17 Unfortunately, little attention has been given to empirical investigation of the process by which labor markets adjust to equilibrium. This adjustment pro cess has significant implications for the reaction of an economic system to pure fiscal policy changes such as increased government expenditures on goods and services and increased real disposable income of con sumers through tax reductions unaccompanied by in creases in the money stock. Acceptance of the Keynesian concepts of a con sumption function and an interest elasticity of the 15Axel Leijonhufvud, On Keynesian Economics and The Eco nomics of Keynes (New York: Oxford, 1968), p. 67. 16Armen A. Alchian, “Information Costs, Pricing, and Re source Unemployment,” pp. 27-52, and Charles C. Holt, “Job Search, Phillips’ Wage Relations and Union Influence: Theory and Evidence,” p. 53-123, both in E.S. Phelps et. al. Microeconomic Foundations of Unemployment and Inflation (New York: WAV. Norton & Co., 1970). 1‘ Leijonhufvud, On Keynesian Economics, p. 69. 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 demand for real cash balances, does not imply in creases in output and employment from the above types of fiscal policy actions if the labor market is al lowed to adjust to the equihbrium supply and de mand functions which prevailed before the policy shock.18 The restoration of such an equilibrium im plies the same real output and employment which pre vailed before the policy action, that is complete “crowding out” of the fiscal stimulus in real terms. Which elements of real private demand are displaced by the fiscal stimulus will depend on the specifica tions of the sector demand functions. If the money demand function is completely interest inelastic then, when the new equilibrium is achieved, velocity must be unchanged in the absence of an accommodating monetary expansion, and complete “crowding out” must also occur in nominal terms. With an interest elastic money demand function, higher prices can occur in the new equilibrium as a result of higher nominal interest rates. In this case complete “crowding out” will not occur in nominal terms. The results of this model are consistent with the monetarist position on the role of fiscal policy as out lined by Andersen and others. Therefore, if the ad justment process in the labor market is relatively rapid, then the weight would seem to be with the monetarist contention that fiscal policy is a relatively weak tool for short-run stabilization. On the other hand, if the adjustment is very slow, “transitory im pacts” of fiscal policy actions of the type usually de rived from income determination models exist, and may have an important role in stabilization policy.19 The Conduct of Monetary Policy A consistent characteristic of the monetarist posi tion is the insistence that monetary policy should be conducted in terms of controlling the rate of growth of monetary aggregates rather than controlling inter est rates or money market conditions. This position can be traced back at least as far as early post-war proposals of Friedman during the period when the Federal Reserve was still supporting the price of Government debt.20 In support of this position, the monetarists have developed a number of valid objections to the money market approach. First, they would hold that the view 18Gardner Ackley, Macroeconomic Theory (New York: Macmillan, 1961), pp. 382-387. 19It is noteworthy in this respect that Friedman seems to always be careful to acknowledge that there are short-run “transitory” effects of fiscal policy actions on real output and employment. -"Milton Friedman, A Program for Monetary Stability ( New York: Fordham University Press, 1959). JANUARY 1972 that monetary influences are channeled to aggregate demand only through impacts of interest rates on in vestment demand is too narrow a view of the role of monetary policy. It has been argued above that there exists fairly widespread support for this argument to day, at least as a theoretical position. Second, they would argue that observed market rates of interest are nominal rates, and that in times of changing ex pectations of future inflation rates, the relevant inter est rates for economic decisionmaking are ex-ante real rates of interest — nominal rates less the anticipated rate of future inflation. Under these circumstances it is possible that changes in nominal interest rates may give a completely wrong impression of what is really happening in terms of real rates of return. As an example of this, the St. Louis Federal Reserve Bank has published from time to time over the past several years, estimates of a real interest rate series which proposes to measure long-term yield such as the corporate bond rate adjusted for inflationary an ticipations. The contrast in the behavior of this series and the corresponding nominal series is quite striking. It is well known that the nominal series has achieved historically high levels in the past several years. On the other hand, the proxy for the real rate has re mained remarkably steady over the latter part of the decade. One can object, of course, to the techniques used to approximate this series. Nevertheless, the monetarists have made a valuable contribution in em phasizing this distinction, because the existence of “inflation premiums” in market interest rates, particu larly long-term rates, over the past several years is now widely acknowledged.21 An additional argument which has not been ad vanced by the monetarists to my knowledge, can be derived from recent work in investment theory. Con siderable empirical testing has now been done on the neoclassical theory of investment which is most closely associated with the work of Jorgenson.22 This theory indicates that interest rates are but one component of the quasi rent on new capital, which is a postu 21For example, “Financial Developments in the Third Quarter of 1971,’ Federal Reserve Bulletin (November 1971), p. 871, states: The key factor contributing to interest rate de clines, however, was the marked change in market expectations triggered by the President’s new eco nomic program. Expectations of inflation, and hence the inflationary premium on interest rates, appear to have been reduced by the temporary freeze on wages and prices and by the indication that a program of strong continuing controls would follow. 22Robert E. Hall and Dale W. Jorgenson, “Application of the Theory of Optimum Capital Accumulation, ’ pp. 9-60, and Charles W. Bischoff, “The Effect of Alternative Lag Dis tributions,” pp. 61-130, both in Gary Fromm, ed., Tax Incentives ana Capital Spending (Washington: the Brook ings Institution, 1971). Page 29 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 JANUARY 1972 lated determinant of investment activity. Various tax policy actions can also affect the rate of return on capital, and it is the net effect of the changes in interest rates and these tax policy actions which is the relevant influence on investment behavior. In particu lar, in discussions of monetary policy in late 1971, arguments that interest rates must be brought down to stimulate investment may be overly emotional. With the resumption of permanent tax credits on equipment, and the reduction of useful lives allowed for tax purposes earlier in the year through the revi sion of Treasury regulations, it is likely that the net effect on investment demand through the so called “cost of capital” channel would be expansionary, even if interest rates were to rise significantly over the first part of next year. priate at this point.25 It appears inappropriate to ar gue about the stability of the demand for money function, in the sense that the aggregate demand for real cash balances can be thought of as a stable func tion of a few parameters.26 This proposition has been implicitly accepted by all empirical research into the nature of this function.27 It is also true that most of these studies have concluded that statistically signifi cant interest elasticities of money demand do exist. On the other hand the studies which have attempted to distinguish between short-run and long-run interest elasticities have consistently found that the short-run elasticities are quite small relative to the long-run elasticities because of a relatively slow speed of ad justment back to the long-run function after a distur bance from an initial portfolio equilibrium position. If these arguments are valid in minimizing the im portance of money market conditions as the target of monetary policy, then the question which remains to be answered is why concentrate on the rate of growth of monetary aggregates. It would seem that there is nothing in the “weak monetarist thesis,” as outlined above, which is sufficient to call for the conduct of monetary policy in terms of controlling the rate of growth of monetary aggregates. The theory must be supplemented by additional hypotheses about the short-run behavior of velocity. It should be noted that these propositions say nothing about the control of the money stock through open market operations aimed at money market con ditions versus control through open market operations aimed at reserve aggregate targets. This issue involves the elasticity of the supply function for money, rather than the demand function, and goes beyond the scope of the present discussion. Brunner establishes a necessary condition for this policy orientation in what he calls the “strong mone tarist thesis,” which maintains that the variability of monetary impulses is also large relative to the speed at which the economy absorbs the impact of environ mental changes.23 It does not seem totally appropriate to interpret this statement as holding that velocity is ( or must b e ) constant in the short run, as some recent commentators seem to imply.24 All that appears nec essary is that if velocity changes, it must change in a manner which is predictable from the time path of past or predicted future behavior of the money supply. It can be demonstrated with currently popular formu lations of the money demand function that the lower the short-run interest elasticity of the demand for money, the more likely this condition will be met. We shall return to this point in the next section where some comments are presented on the St. Louis equation. The discussion up to this point has centered on monetarism as monetary theory and its prescriptions for monetary policy. It seems appropriate to conclude with some remarks on the St. Louis equation. Several remarks on the current state of empirical research on the money demand function are appro2:iBrunner, “The Role of Money and Monetary Policy,” p. 19. -'Paul A. Samuelson, “Reflections on the Merits and Demerits of Monetarism,” in James J. Diamond, ed., Issues in Fiscal and Monetary Policy: The Eclectic Economist Views The Controversy (DePaul University, 1971), pp. 7-21. Page 30 The St. Louis Equation This regression has been the subject of varied inter pretation since it first appeared. In their original arti cle Andersen and Jordan state: This article does not attempt to test rival economic theories of the mechanism by which monetary and fiscal actions influence economic activity. Neither is it intended to develop evidence bearing directly on any causal relationships implied by such theories. More elaborate procedures than those used here would be required in order to test any theories under lying the familiar statements regarding results ex pected from monetary and fiscal actions. However, empirical relationships are developed between fre quently used measures of stabilization actions and economic activity. These relationships are consistent 2r,A useful summary of research on the money demand func tion is provided in David Laidler, The Demand for Money: Theories and Evidence (Scranton, Pa.: International Text book Co., 1969). 2tiMilton Friedman, “The Quantity Theory of Money — A Restatement,” in Studies in the Quantity Theory of Money (Chicago: University of Chicago Press, 1956), pp. 3-21. 27The derivation of regression estimates of any function pre supposes stability of that function, in the Friedman sense, over the sample period. F E D E R A L R E S E R V E B A N K O F ST. L O U IS with the implications of some theories of stabiliza tion policy and are inconsistent with others. . . ,28 A later article states: This general specification represents the reduced form for that class of structures which has AM [changes in money stock] and A E [changes in F ed eral expenditures] as exogenous variables. In this form the total spending equation remains uncom mitted as to structure; it is potentially consistent with both Keynesian and quantity theory models.-9 In the latter article, it was also noted that equations had been constructed using percentage changes, rather than first differences, and that the results were basi cally unaffected by the change in specification. These equations and their established forecasting properties have remained somewhat a mystery to eco nomists associated with the nonmonetarist position and the tradition of large econometric models. It has been subjected to numerous attacks on the choice of independent variables and problems of statistical bias.30 In general it would appear that the equation has withstood these attacks surprisingly (or disturb ingly ) well. We shall offer yet another interpretation of the St. Louis equation which suggests that it is not a “reduced form” of an unspecified system, but rather only one component of the structural system. This interpreta tion of the St. Louis results is not sympathetic to the view that the St. Louis equations are a competitive econometric model of the United States economy. Judged in this perspective, it is possible to rationalize its forecasting performance. Recognizing that percentage changes are approxi mately equal to changes in logarithms for small changes, the St. Louis equation can be rewritten as: (1 ) A ln Y = a + 2 |5i A l n M - i + 2 y i A l n E - i i—0 i—0 where Y = GNP in current dollars M = money stock E = high-employment Government expendi tures. 28Leonall C. Andersen and Jerry L. Jordan, “Monetary and Fiscal Actions: A Test of Their Relative Importance in Economic Stabilization,” this Review (November 1968), p. 11. 29Leonall C. Andersen and Keith M. Carlson, “A Monetarist Model for Economic Stabilization,” this Review ( April 1970), p. 9. 30Frank de Leeuw and John Kalchbrenner, “Monetary and Fiscal Actions: A Test of their Relative Importance in Economic Stabilization — Comment,” this Review ( April 1969), pp. 6-11; and Edward M. Gramlich, “The Useful ness of Monetary and Fiscal Policy as Discretionary Stabili zation Tools,” Journal of Money, Credit, and Banking (May 1971), pp. 506-532. JANUARY 1972 This can in turn be conveniently transformed into a velocity (V ) equation: (2 ) AlnV = (A ln Y —AlnM ) = a + (|30 - 1 ) AlnM + Zpi AlnM -i + 2 yiA lnE-i i—1 i= 0 Recent empirical formulations of the demand for money function specify that in the long run velocity is a function of interest rates, and assume that private economic units adjust to their long-run equilibrium cash balances with a lag. Such models can be trans formed into a specification:31 (3 ) AlnV = yo Ain g (r) + y i AlnMt + Z yi AlnMt-i i= l where r = interest rate. If for the moment the interest rate term is ignored, this equation appears quite similar to the transformed Andersen-Jordan equation (1) above. It can be clearly seen from this equation that such specifications of the money demand function relate changes in velocity to current and lagged changes in the money stock ( all in logarithms). It further suggests that changes in inter est rates will induce additional changes in velocity. Most economists would hold that changes in interest rates and money stock are jointly determined, and consequently, forecasts from equation (1), ignoring the induced changes on interest rates from changes in the money stock, will cause forecasting errors. How ever, if the short-run interest elasticity of the money demand function is very small, then an estimated equation omitting this term would most likely produce a credible forecasting record. In addition, it is likely that the distributed lag on high-employment Govern ment expenditures used by Andersen and Jordan is correlated with interest rates in both the sample and post sample periods, and serves as an effective proxy variable for forecasting purposes.32 Summary Monetarism and the monetarist approach to de mand management has raised many issues in the past several years which have significantly influenced the attitudes of professional economists on the question of 31See Appendix for the derivation. 32Equations of the form of (3 ) have been estimated using the data of the current St. Louis forecasting equations, both with and without the high-employment expenditure varia bles. Space constraints permit only the comment that the interest rate variable, either a short-term rate such as the Treasury bill rate, or a long-term rate such as the Cor porate Aaa rate, show up as highly significant variables, though with very low short-run elasticities. Even with the interest rate variable in the specification, some of the co efficients in the distributed lag on high-employment ex penditures remain significant. Page 31 F E D E R A L R E S E R V E B A N K O F ST. L O U IS JANUARY 1972 how to pursue stabilization policy. Monetarist models have to date established a forecasting record which is credible when compared to the more entrenched in come determination approach. Assume that in the long run velocity is a function of interest rates: Recently, considerable work has been done to ela borate an extensive theoretical framework which pur portedly underlies the policy prescriptions and the “reduced form” monetarist models of aggregate eco nomic activity. The comments above suggest that much of this theoretical framework is shared with economists of nonmonetarist persuasion, but that there are a number of areas in which substantially different views of the world exist. Unfortunately, few attempts have been made by the monetarists and nonmonetarists to identify the common and contrast ing elements of their theoretical constructs. Even less work has been done to attempt to disprove the specific hypotheses of market behavior in the areas of conflict, most of which, I believe, involve the dynamics of price adjustments.33 Only as such analysis becomes available will we be able to resolve important policy issues such as the relative strengths of fiscal and mone tary actions under various conditions of the economy, and the speed at which policy actions affect aggregate demand, employment, and prices. (2) M ° = [ i ] Y = Yg(r) ^ 33For example, Milton Friedman in , “A Theoretical Frame work,” argues that a major unresolved issue in his analysis (as well as that of others) is the response of real output and prices individually to policy shocks. APPENDIX The purpose of this appendix is to demonstrate the derivation of equation (3 ) in the text from a money demand specification. (1) V = F ( r ) ^ dr > 0 or < 0 In addition private economic units are assumed to adjust to their long run equilibrium cash balance positions with a lag, which is usually specified as: (J) M_i [ / M_jJ When this is expressed in logarithms it becomes: (4) A l n M = 6 1 n M l>- 61n M_j A more general form1 of the distributed lag adjustment mechanism can be specified as: lriM = /?0 l n M ° + l n M_ j + . . . + /8n l n M _ n Substituting for M “ gives: (5) l n M = /3 0 In [g(r)]+ /30 In Y + X/3; lnM_j i= l which can be rewritten as a velocity equation: (6) l n V = ( l n Y - l n M ) = -/3c l n [g(r)]+ ( 1 - / 3 J I n Y - £ f t ln M ,* i=l Taking first differences in the logs: (7) A in V = -/80 Ain [g(r)] + (1 -/3 J A ln Y - AlnM_; i=l which can be transformed to: (8) AlnV = - j80A ln [g(r)] + (1-/S0) AlnV + (l-/30) A ln M - jfftA ln M .! i-1 (9) A lnV = i-[-/ 8 0 A ln[g(r)] + (l-/30) A ln M - S f t A l n M . j l PoL i=i iRobert M. Solow, “On a Family of Lag Distributions,” Econometrica (April 1960), pp. 393-406. This article and the accompanying one by Ronald L. Teigen are available os Reprint No. 74 Page 32 J