The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
FEDERAL RESERVE BANK O F ST. L O U IS NOVEMBER 1969 Progress in Controlling Inflation Monetary and Fiscal Influences on Economic Activity — The Historical Evidence............................................................. 5 The Effects of Inflation (1960-68) .................. 25 Vol. 51, No. 11 Progress in Controlling Inflation S t OPPING the acceleration of price increases is currently the most important objective of national economic policy. The other major objective, “full em ployment” defined as an unemployment rate below 4 per cent, has been achieved since late 1965. The method that has been chosen to achieve stable prices is to curtail growth in total spending (GNP). His torically, a slowing of total spending has been ac companied by a slowing in the growth of real product, followed by a slowing in the rate of increase in prices and by subsequent recovery in the growth rate of real product. Consequently, the question arises as to what is implied for the course of real product under the current anti-inflation policy. The Current Anti-Inflation Program Economic developments in 1969 are proving that historical patterns of output and prices are reliable guides for any assessment of the success of the present anti-inflation program. Total spending for goods and services rose at a 7.8 per cent annual rate from mid1968 to the third quarter of this year, less than the 9.6 per cent increase in the previous year but still far in excess of the growth rate of the economy’s pro ductive potential. Prices have advanced at a 4.8 per Demand and Production R atio S cale R a tio S cale Quarterly Totals a* Annual Rates S e a so n a lly A djusted Billions o f Doll Billions o f P ol la rs cent rate since mid-1968, a much faster rate than in the previous year. Real output, on the other hand, increased at only a 2.8 per cent rate over the past five quarters, down from the 5.3 per cent increase in the year ending in mid-1968. Final sales have slowed somewhat since mid-1968, but their growth continues at a relatively rapid rate. These sales grew at an 8.8 per cent rate from mid1967 to mid-1968 and since mid-1968 have increased at a 7.8 per cent annual rate. Fiscal Conditions Federal fiscal policy has continued restrictive in its impact on economic activity since mid-1968, when the Revenue and Expenditure Control Act of 1968 was passed. On a high-employment basis, the national in come accounts budget was in surplus at a $7.8 billion annual rate in the third quarter, down slightly from the previous two quarters, but in marked contrast to the $13.7 billion annual rate of deficit in the first half of 1968. This shift of the budget from a large deficit to substantial surplus was accomplished by a sharp reduction of expenditure growth and enactment of the 10 per cent tax surcharge. 1961 1962 1963 1964 1965 1966 1967 1968 1969 Q GNP in current d o llars . Source: U.S. D epartm ent o f Com m erci [2 GNP in 1958 dollars. Percentages are a nnual rates o f change betw een periods indicated. They are presented to a id in com paring most recent developments with past "trends." la te s t d a ta p lo tte d: 3rd q uarter Page 2 Federal expenditures have risen at a 5.9 per cent annual rate since mid-1968, compared with a 11.7 per cent increase in the previous year. Defense spending has slowed markedly, rising at only a 2.8 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 per cent rate since mid-1968, compared with a 13.9 •per cent average rate of increase during the period from 1965 to 1968. Budget prospects for calendar 1970 indicate that the overall fiscal restraint w ill continue at least through the first half. Current estimates include re striction of the rate of growth in Federal expenditures at a 3 per cent rate until at least mid-1970. The tax program includes a reduction of the tax surcharge to 5 per cent on January 1, 1970 and repeal of the investment tax credit retroactive to April 20, 1969. If these expenditure and taxing expectations are real ized approximately, the degree of fiscal restraint will continue through the first half of 1970, with the highemployment budget at an estimated $9 billion rate of surplus. If, on the other hand, the surtax is allowed to expire at the end of this year, this budget would be at an estimated $5 billion rate of surplus in the first half of 1970. W hile the high-employment budget suggests con tinuation of the current posture of fiscal restraint in early 1970, the degree of restraint may be greater than this budget suggests. Consumers and businesses, to some degree, react to changes in tax rates, which they consider to be temporary, by adjusting their rate of saving rather than their spending. As a result, changes in tax rates are significant in the short run because they signal a change in the allocation of re sources between the Government and the private sec tor. According to this interpretation, the expected course of Federal spending, rather than the net budget position, may in fact provide a more reliable indica tion of the degree of fiscal restraint in the near future. Recent Monetary Actions Monetary actions became moderately less expan sive early in 1969, and since early summer a consid erable degree of restraint has been exercised. The money stock has been about unchanged in the last four months, compared with a 4.4 per cent annual rate of growth in the first half of the year and a 7 per cent increase in 1968. The demand deposit component of money declined at a 1.2 per cent rate from June to October, after decelerating from a 7 per cent increase in 1968 to a 3.7 per cent rate of growth in the first half of 1969. The monetary base, uses of which are total bank reserves and currency held by the public, has in creased only slightly since early June. The money stock and the monetary base usually grow at similar rates, as in 1968 when they both grew 7 per cent, and NOVEMBER 1 9 6 9 M on ey Stock R a tio S ca le B illio n s o f D o lla rs R a tio S ca le B illio n s o f D o lla rs Monthly Average, of Doily Figures 210 200 190 180 170 160 150 140 130 1961 1962 1963 1964 1965 1966 1967 1968 1969 Percentages are annual rates of change between periods indicated. They ore presented to aid in comparing most recent developments with past "trends." Latest data plotted: October preliminary in the last three months when they have both been about unchanged. However, small shifts frequently occur in the multiplier relationship between the mon etary base and money.1 Since early June, the con tinuing outflow of time deposits from commercial banks has been a major factor causing a net increase in the multiplier. Federal Reserve credit increased at only a 0.8 per cent annual rate from June to October, after increasing at a 6 per cent rate in the first part of the year and 10 per cent in 1968. Member bank reserves declined at about a 9 per cent annual rate from June to October, after being about un changed during the first part of the year, and growing 8 per cent in 1968. Interest rates have risen recently, though most rates are still below early October peaks. Over the past year interest rates have increased markedly, with yields on prime four- to six-month commercial paper rising about 3 percentage points. In contrast, Regulation Q ceilings have remained unchanged, and the discount rate has been increased only three-fourths of a percentage point during the past year. The dis count rate is currently about one percentage point below the three-month Treasury bill rate, compared with a spread of about one-fourth of a percentage point a year ago. Volume of commercial paper outstanding has con tinued to increase rapidly, up $7.8 billion since last ’ See Jerry L. Jordan, “Elements of Money Stock Determina tion,” this Review, October 1969, pp. 10-19. Page 3 F E D E R A L R E S E R V E B A N K O F ST. L O U I S December, compared with a $3.4 billion increase in the corresponding period of 1968. In contrast, the outstanding volume of large negotiable certificates of deposit at large commercial banks declined by $12.1 billion from last December to October 1969. Low Regulation Q ceilings relative to market rates have prevented banks from competing effectively for funds, thereby directing funds through other channels such as the commercial paper market. W hether the rerout ing of funds from commercial banks through other channels is an indication of monetary restraint is doubtful. It is clear, however, that the rerouting of funds promotes inefficiences in the channeling of funds from savers to investors. Economic Conditions Total spending, that is, nominal GNP, continued to grow from the second to the third quarter at about the same rate as in the three previous quarters, despite the restrictive position of monetary and fiscal policy. The composition of the increase in total spending in the third quarter, however, indicates a slowing of total spending might be expected in the immediate future. Though total spending continued to advance at a rapid rate, private final sales rose much more slowly in the third quarter. Total spending was buoyed by increases in the rate of inventory accumulation and the Government pay increase. If these increases are not indicative of a trend, the slowing of private final sales may portend further slowing in the pace of economic activity. Measures of income, employment and production indicate that there was some slowing in the pace of economic activity during the third quarter. W hile this observation is based on a very short period, it is con sistent with the restrictive nature of monetary and fiscal actions. Personal income slowed to a 4.8 per cent rate of increase in October, after rising at a 8.8 per cent rate in the January to September period. Payroll employment rose at a 2.4 per cent rate in the July to October period, compared with a 3 per cent rate in the previous six months. Industrial produc tion, slowed by reductions in primary metals output, declined at a 2.9 per cent rate from July to October, compared with a 6.1 per cent rate of increase from December to July. Conforming to the usual patterns of lags of price effects behind total spending, price trends have con tinued upward. The general level of prices rose at a 5.4 per cent annual rate from the second to the third quarter, compared with a 4.6 per cent increase in the Page 4 NOVEMBER 1 9 6 9 previous year. Wholesale prices of industrial com modities have risen at a 3.7 per cent rate over the past year. Consumer prices have risen at a 6 per cent rate since the first of the year, compared with a 5 per cent rate in the previous year. Outlook for Prices and Output The current stabilization program is directed to ward slowing total spending as the necessary means of limiting the advance of prices. Recent trends in income, employment and production suggest that this program of restrictive monetary and fiscal actions has begun to take effect, though as yet there has been no discernible effect on prices. Past experience, however, indicates that restraint in total spending has its effect first on output growth, and only later on prices. The rate of increase in output must be ex pected to slow, as an inevitable by-product of an anti-inflationary program. It is of interest, therefore, to trace out the possible course of real output during this transition period. Total spending has decelerated only slightly since mid-1968. The restrictive economic policies should soon begin to affect demand growth significantly. The rather steady rate of demand growth has been accompanied by a slowing in the growth of real out put and an acceleration in the rate of price increase. To aid in the exploration of the outlook for the near-term, alternative growth rates of money and Federal expenditures can be postulated, and the most probable future movements in total demand, output and prices, projected on a basis of historical ex periences, can be examined. The monetary and fiscal restraint that has already occurred provides the basis for substantial further slowing in output growth, and for gradual slowing in the rate of price advance. The choice of alternative rates of monetary expansion at this point (fourth quarter 1969) w ill be instrumental in determining the extent and duration of the slowing of total output. Prices can be brought down more rapidly by restricting growth in money severely, but with larger cost in terms of restriction of real output. Monetary and fiscal restraints are probably being manifested in reduced total dollar demand and real output growth. The task of policymakers is to avoid a prolongation of a degree of restraint which would lead to excessive restriction of output growth, but at the same time to avoid a premature reversal of the policy of restraint such that significant progress would not be made in effectively reducing inflationary pressures. Monetary and Fiscal Influences on Economic Activity —The Historical Evidence* by MICHAEL W . KERAN In November 1968 this R e v i e w included an article which tested the relative importance of monetary and fiscal influences on economic activity for the postwar period 1953-68. The conclusions of that article were that monetary influences had a stronger, more predictable, and faster impact on economic activity than fiscal influences. The intent of this article is to consider the same issue in a longer, historic context (1919-69). Have mone tary influences dominated economic activity in periods when financial and institutional factors were substan tially different, as in the 1920’s, and when the general trend of economic activity was largely depressed, as in the 1930’s? The results presented in this article indicate that monetary influences have dominated fiscal influ ences on economic activity in all subperiods considered, with the single exception of the years covering W orld W ar II. This article also presents evidence that the movements in the money stock have been dom inated by the behavior of the monetary authorities and not by the behavior of the public. SURJECT of continuing interest in professional and recently in popular economic writing is the rela tive role of monetary and fiscal influences in deter mining economic activity.1 This debate has been re newed by Leonall Andersen and Jerry Jordan (AJ) in an article published in this Review.2 That article presented evidence which indicated that mon etary influences had a larger, more predictable, and faster effect on economic activity than fiscal influ ences in the period from 1953 to 1968. “The content and presentation in this article have been substantially improved by the suggestions of the author’s colleagues in the Research Department of the Federal Reserve Bank of St. Louis: Homer Jones, Leonall Andersen, Christopher Babb, Denis Karnosky, and W illiam Yohe. In addition, he received valuable comments and criticisms from Oswald Brownlee, Karl Brunner, Philip Cagan, Albert Cox, Milton Friedman, Harry Johnson, John Kalchbrenner, Thomas Mayer, David Meiselman, and Allan Meltzer. l This issue was first raised in a somewhat different context by Milton Friedman and David Meiselman in “The Relative Stability of Monetary Velocity and the Investment Multi plier in the U.S.” Stabilization Policies, The Commission on Money and Credit, Prentice-Hall, 1963. 2Leonall C. Andersen and Jerry Jordan: “Monetary and Fis cal Actions: A Test of Their Relative Importance in Eco nomic Stabilization,” this Review, November 1968. These results have stimulated considerable interest and discussion.3 The ensuing debate has mainly con fined itself, however, to the time period used in the original AJ article (1953-68). Since other economic experiences might suggest a different assessment of monetary and fiscal influences, it seems useful to ex pand the testing periods to include a longer period in United States economic history. It is reasonable to assume that tests obtained from a wider range of experience would go a long way 3Richard G. Davis, “How Much Does Money Matter?”, Monthly Review, Federal Reserve Bank of New York, June 1969; Edward M. Gramlich, “The Role of Money in Economic Activity: Complicated or Simple?,” Business Economics, Sep tember 1969; “The Usefulness of Monetary and Fiscal Policy as Discretionary Stabilization Tools,” (presented at the Ameri can Bankers Association, Conference of University Professors, Milwaukee, September 19 6 9 ); Frank de Leeuw and John Kalchbrenner, ‘ Monetary and Fiscal Actions: A Test of Their Relative Importance in Economic Stabilization — Comment,” this Review, April 1969; Paul S. Anderson, “Monetary Velocity in Empirical Analysis,” Controlling Monetary Aggregates, pre pared by the Federal Reserve Bank of Boston, September 1969; M. J. Artis and A. R. Nobay, “Two Aspects of the Monetary Debate,” National Institute Economic Review, Vol. XLIX (August 19 6 9 ), pp. 33-51; and W ilfred Lewis, Jr., “ ‘Money is Everything’ Economics — A Tempest in a Teapot,” National Conference Board Record, Vol. VI, No. 4 (April, 19 6 9 ), pp. 32-35. Page 5 F E D E R A L R E S E R V E B A N K O F ST. LO U I S toward answering some of the questions raised about the A J article. If the dominance of monetary influ ences prevailed in earlier periods, then confidence in the reliability and stability of the original results and their continued applicability is enhanced. On the other hand, if the dominance of monetary influ ences is shown to be confined to only the most re cent time period, then it could be asserted that special circumstances are at work in the present period which could not be relied upon to continue. The intent of this article is to test the relative impact of monetary and fiscal influences on economic activity in the United States on a quarterly basis from 1919 to 1969 and for selected subperiods. This article is organized in the following way. First, a brief and highly simplified review is given of some of the theoretical and statistical issues which have been raised in connection with the type of tests used by AJ. This review will allow us to see what can and cannot be deduced from any results. Second, the test results for the 50-year period from 1919 to 1969, w ith 200 quarterly observations, will be presented, together with a historical review and comparison. Finally, the statistical reliability of the results will be considered. Theoretical and Statistical Issues There are two primary ways to study the relative importance of monetary and fiscal influences on eco nomic activity. First, their effects can be inferred within the context of a fully specified and statis tically estimated structural model of the economy, as in the FRB-MIT model.4 The monetary and fiscal variables are introduced in the structural model at those points where their functional roles are indicated by economic theory. The measured impact on eco nomic activity of the monetary and fiscal variables is dependent upon the explicit transmission mechan ism which is postulated and built into the structural model. Second, monetary and fiscal influences can be measured by direct estimation of a single regression equation. In this case, some measure of economic activity is regressed directly against the monetary and fiscal variables without specification of a trans mission mechanism. 4See Frank de Leeuw and Edward M. Gramlich, “The Chan nels of Monetary Policy,” Federal Reserve Bulletin, June 1969. A structural model is one in which the major behavioral assumptions of a theory are explicitly included in the statistical estimates. It is fully specified if there are as many equations as there are endogenous variables. Page 6 N OV E M B E R 1 9 6 9 The Large Structural Model Approach There are advantages and disadvantages associated with each of these approaches. An important ad vantage of the large structural model is that it allows one to distinguish between direct and indirect mone tary and fiscal influences, and to see how subsectors of the economy are affected. In formal terms a struc tural model is essentially a hypothesis of the model builders about the interrelations in the economy. The statistically estimated equations represent compon ents of that hypothesis. If it turns out that the model builders’ view of the economic mechanism is reason ably correct, then the “structural richness” of the large models permits a wider range of questions to be answered. The major disadvantage of structural models is that the model builder may have omitted an important channel of transmission and, consequently, incorrectly estimated the magnitude of the monetary or fiscal influences. Indeed, even if the model builder has a good idea of the transmission channels, it may be technically impossible to estimate them because the channels have not or cannot be quantified. For exam ple, assuming that the cost of borrowing is an im portant link in the monetary transmission mechanism, it is quite possible that this is not accurately measured by market interest rates. Both changes in credit ra tioning and compensating balance requirements, for which there are no available quantified measures, could affect the cost of borrowing yet not be re flected in changes in market interest rates. The Single Equation Approach An advantage of the single equation approach is that if the monetary and fiscal variables are correctly specified, and if they are not themselves determined by economic activity, they w ill capture the direct and indirect impact of monetary and fiscal influences on economic activity, irrespective of the transmission channels. The single equation approach avoids the problem of specifying and measuring specific links between monetary and fiscal influences and economic activity, and w ill generally be consistent with a wide range of theories (hypotheses) about the structural interrelations in the economy. One major disadvantage of the single equation approach used here is that it can deal with only a single question, the relative impact of monetary and fiscal influences on economic activity. It does not distinguish between the direct and indirect impact of the monetary and fiscal influences on economic 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 activity or how subsectors of the economy are affected.5 In addition, both the structural model ap proach and the single equation approach face the same problem of selecting measures of monetary and fiscal influences which are exogenous in a statistical sense. In order to derive results which are comparable with A J’s work, the single equation approach (the so-called St. Louis equation) is used here. However, before presenting the test results, it would be useful to consider what can and cannot be implied by using this approach. First, as was previously noted, the single equation approach restricts us to considering just one question — the relative impact of monetary and fiscal influences on economic activity. W e cannot say what the channels of the influence are. Second, the single equation approach used here does not allow us to discriminate between economic theories. Take the generalized statement of the single equation which is used in this article: AY = oto + oci AM + 0C2 AF where AY = changes in economic activity, AM = changes in monetary influences, A F = changes in fiscal influences. The parameters, a j and a 2, indicate the magni tude of the impact of monetary and fiscal influences, respectively, on economic activity, and a 0 is a proxy for the net trend of all other influences on economic activity. Assume for the moment that the statistical results of a test using this format substantially favor monetary influences ( A M) over fiscal influences ( A F ) in determining economic activity (AY). These results do not provide clear-cut evidence to help an swer the question of whether the Keynesian IncomeExpenditure Theory or the Modern Quantity Theory is a better representation of the economic world. Both theories provide an operational rule for mone tary influences, and thus the dominance of the mone tary variable does not discriminate between them.6 A test of competing economic theories can be con5One way to handle this disadvantage is to regress the mone tary and fiscal variables against the components of GNP to see which broad sectors of the economy are affected. See Leonall C. Andersen, “Money and Economic Forecasting,” Business Economics, September 1969, for the results of such a test. 6There are a number of empirically estimated Keynesian eco nomic models which have a “weak ’ monetary sector. Evidence that monetary influences are important would tend to cast doubt on the usefulness of those models. However, this is more a criticism of the particular model and not the underlying Keynesian theory. W ithin the context of standard Keynesian theory, there are circumstances where strong monetary and weak fiscal influences could exist. NOVEMBER 1 9 6 9 ducted only if the alternative behavioral assumptions are made explicit.7 Third, the single equation approach does not necessarily tell us anything about monetary and fiscal policy decisions of the authorities. If the independ ent variables have been chosen properly, they will indicate monetary and fiscal influences on the econ omy. One can assert that such influences are simul taneously a measure of the policy intentions of the authorities only if additional external evidence is pro vided, which indicates that the policymakers have acted either consciously or otherwise to systematically control the monetary and fiscal variables used in the equation.8 The third point can be clarified with an example: Assuming there are two countries, A and B. Statistical tests indicate that the monetary variable dominates the fiscal variable in influencing economic activity in each country. However, it is also known that Country A does not have a central bank, while Country B does. Obviously, we can only talk about discretionary monetary policy in Country B, but we can talk about monetary influence in both countries. In Country A, the monetary variable is dominated by factors other than by the actions of a central bank —perhaps by the domestic gold supply. In Country B (with a central bank), the monetary variable could be dom inated by the central bank; however, our statistical results do not provide any evidence with respect to that issue. Such evidence can be derived only by an explicit investigation of the behavior of the central bank in Country B. Thus, discretionary monetary policy and monetary influences are not necessarily measured by the same variable." 7A test of competing economic theories conceptually could be conducted either with a single reduced-form equation or with a more fully specified structural model. W hen Friedman and Meiselman, “The Relative Stability . . .” attempted such a test using the single equation reduced-form approach, a consid erable controversy occurred within the economics profession. To the best of the author’s knowledge, no one has attempted to compare competing theories by a test of alternative struc tural models. 8Such information would come from studies of the “reaction function” of the policy-making authorities. There have been a number of such studies of the monetary authorities. For example: ( 1 ) William Dewald and Harry Johnson, “An Objective Analysis of the Objectives of American Mon etary Policy, 19 5 2 -19 6 1,” Banking and Monetary Studies, ed. Deane Carson (Homewood, Illinois: Richard D. Irwin, 19 6 3 ); ( 2 ) James W . Christian, “A Further Analysis of the Objectives of American Monetary Policy,” The Journal of Finance, volume XXIII, June 1968; (3 ) Michael W . Keran, and Christopher T. Babb, “An Explanation of Federal Re serve Actions (19 3 3 -6 8 ),” this Review, July 1969; (4 ) John Wood, “A Model of Federal Reserve Behavior,” Staff Economic Study No. 17, Board of Governors of the Federal Reserve System. nThis point is quite important and open to some misunderstand ing. To link monetary policy with the indicator of monetary Page 7 F E D E R A L R E S E R V E B A N K O F ST. LO U I S Given this array of caveats with respect to the single equation approach, it is nevertheless highly useful in indicating monetary and fiscal influences on economic activity. The key reason has already been discussed. An economy is an extremely complex array of interrelated and interdependent markets tied together by the price mechanism. Millions of indi vidual decision-making units participate in these markets. In this complex web of interrelationships, attempts at specific and detailed measurement of the channels through which monetary and fiscal in fluences operate on economic activity are quite hazardous. Given the complexities of the economy and the existing uncertainty about the transmission mechan ism, it is useful to measure the monetary and fiscal influences directly, without constraining them to op erate within our imperfect notions about how they operate. Freedom from this type of specification error is perhaps the principal virtue of the single equation approach. Problems of the Single Equation Approach The key methodological and statistical problems with the single equation approach are related to selection of appropriate indicators of monetary and fiscal influences. First, a theoretical justification for using particular variables is required. Such justifica tion naturally evolves from the various economic the ories (hypotheses) which have been developed to explain the determination of aggregate economic ac tivity. For example, bank credit or free reserves are unlikely indicators of monetary influence because there is no well-specified economic theory from which these variables are a derivable consequence. Even if statistical results indicate a close relation between bank credit and economic activity, it is difficult to interpret the results. On the other hand, the money stock is a good choice as an indicator of monetary influence because it plays an important role in both the Keynesian Income - Expenditure Theory and the Modern Quantity Theory of Money. Second, there must be evidence that the actions of monetary and fiscal authorities determine the influence, it is not necessary that the authorities consciously control the value of the monetary variable. All that is required is that in controlling some monetary variable the authorities in the process also dominate movements in the variable used to indicate monetary influences. If the author ities have not deliberately attempted to control the variable which is the best indicator of monetary influence, then their policy actions could be criticized. However, this is not neces sarily an argument against using that variable as an indicator of monetary influence. Page 8 NOVEMBER 1 9 6 9 movements in the variables selected. It is not neces sary that the policymakers have acted consciously to control the specific variables used; it is only neces sary that policy actions systematically dominate movements in the indicated variable. This leads naturally to the third and final condi tion. To be able to interpret the regression coefficients meaningfully in the single equation approach, the monetary and fiscal variables must be statistically exogenous. The economic meaning behind this condi tion is that the variables selected to represent mone tary and fiscal influences should not be contemporan eously determined by the behavior of the public, as measured by changes in economic activity. If this exogeneity assumption is not satisfied, the direction of causality is uncertain, and a close statistical asso ciation with economic activity does not provide any evidence of the magnitude of the impact from mon etary and fiscal influences. This is the so-called “reverse-causation argument” against the single equa tion approach. The next section presents the results of various statistical tests of monetary and fiscal influences on economic activity. The last section w ill consider the reverse-causation argument and whether movements in the monetary variables are dominated by the mon etary authorities or by the public. Because the theoretical justification for the monetary and fiscal variables used in this article has already been con sidered in the AJ-article, it will not be presented here. Monetary and Fiscal Influences The test procedure used in this article is to regress quarter-to-quarter changes in a measure of economic activity against quarter-to-quarter changes in the indicators of monetary and fiscal influence. Because of the length of the test period (1919-69), data prob lems were encountered. For example, the most widely used measure of economic activity (nominal GNP), and the most widely used measure of fiscal policy (high-employment receipts and expenditures of the Federal Government), are not available on a quar terly basis before 1946. These deficiences in the data necessitated developing proxies for these measures. A proxy for nominal GNP was constructed to measure economic activity. The proxy consists of the scaled product of the Industrial Production Index (IPI) and the Consumer Price Index (CPI), both of which are available on a monthly basis in continuous time series back to February 1919. Each is the broad est available measure of real output and prices, and NOVEMBER 1 9 6 9 F E D E R A L R E S E R V E B A N K O F S T. L O U I S C h a rt I R a te s o f C h a n g e Nominal GNP and Economic Activity R a te s o f C h a n g e Q uarterly data at annual rates. [l_Economic A c tiv ity is measured by the scaled product of the Consumer Price Index (CPI) and the Industrial Production Index (IPI) m u ltip lied by Gross N ation al Product (GNP) in the base years 1957-59: /C P H P tV •$457.4 billion=Econom ic A ctivity \ 10 , 000 / Sources: GNP-U.S. D epartm ent o f Commerce; Economic A ctivity-In d u stria l Production Index, Board of Governors of the Federal Reserve System; Consumer Price Index, U.S. D epartm ent of Labor their scaled product is an index of economic activity. To convert this value index into a dollar measure, it was multiplied by the value of nominal GNP in the base years of the value index (1957-59).10 By this method an index of quarterly economic activity, measured in billions of dollars, was constructed for the period 11/1919 to 11/1969. larger swings over the business cycle than does nom inal GNP. However, for the purpose of measuring the changes in economic activity from one quarter to the next, this proxy appears to be both useful and reasonably accurate.11 Chart I shows the quarter-toquarter rates of change in nominal GNP and in our proxy from 1947 to 1969. This proxy for economic activity clearly has a num ber of defects. The service industries, levels of gov ernment and agriculture are excluded. In addition, industrial production traditionally grows at a faster trend rate than overall real output because it is more responsive to increases in productivity. Also, it shows 11 The regressions between rates of change of nominal GNP • • (GNP) and our proxy variable (Y ) appear as follows: 10The formula used to compute this measure of economic activity (Y ) is: Y = F r a r ] • ($457-4 billion) 1/1947 — IV/1952 GNP = 3.78 + .45 Y (3 .53 ) (8 .2 4 ) R2 = .74 D -W = 1.82 1/1953 — 1/1969 GNP = 3.42 + .40 Y (1 1 .4 4 ) (15 .8 0 ) R* = .80 D -W = 1.55 Page 9 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 As a measure of fiscal influence, changes in the national debt ( A D) and actual Federal Government expenditures ( AE) (purchases of goods and services plus transfer payments) were used. Data on tax re ceipts are available, but because of the strong influ ence which economic activity has on the value of tax receipts, it was not used. Because AD is also influenced by changes in tax receipts, only the results using Government spending are reported. The use of Federal Government spend ing as our single measure of fiscal influence is not as serious a drawback as it might first appear. Andersen and Jordan (AJ) found that the strongest measure of fiscal influence was achieved by using Federal Gov ernment expenditures alone. The observed level of Government spending which is used from 1919 to 1945 is not significantly different from the highemployment level of Government spending which AJ used, and which is used here for the subperiods from 1946 to 1969.12 As measures of monetary influence, three varia bles were tested: total reserves of member banks, the monetary base, and the narrowly defined money stock ( currency holdings of the nonbank public and private demand deposits). The separate use of monetary and fiscal variables in these regressions implies that one can think of monetary and fiscal influences as having separate impacts on economic activity. This may not be the case. One well-known fiscal influence on mon etary actions can occur because of “even-keel” ac tions. “Even-keel” is the policy of the Federal Reserve to stabilize money market conditions during periods when the United States Treasury is floating a new issue of securities. Thus, an increase in Government spending financed by an increase in debt could in duce an increase in the money stock because of Federal Reserve “even-keel” actions. This issue can be dealt with only by asserting that all factors which affect the money stock are monetaiy and all factors which affect Government spending are fiscal. This is not unreasonable, since the Federal Reserve could stop even-keel actions if it chose to do so. The tests of monetary and fiscal influences were run using four measures of change: quarterly first differences, quarterly central differences, quarterly first rates of change, and quarterly central rates of change. Only the results with quarterly first differ12The only difference between the observed levels of Gov ernment spending and the high-employment level of Gov ernment spending is an adjustment for unemployment compensation payments. These payments did not start until 1937 and did not amount to a significant figure until after W orld W ar II. See Chart III for sources of data for Gov ernment spending, money stock, and economic activity. Page 10 NOVEMBER 1 9 6 9 ences of the money stock and Government expendi tures are reported in this article. However, alternative measures of change and alternative measures of mon etary and fiscal influences give substantially similar results.13 In each test the form of the equation was estim ated with money alone, fiscal alone, and a combina tion of the two. Alternative time lags between t-1 and t-10 were tried using the Almon distributed-lag technique.14 The form of the equation selected and the time lags to represent each time period were chosen on the basis of minimum standard error of estimate adjusted for degrees of freedom.15 The total period was divided into five subperiods: 1919-29, when economic conditions were generally prosperous; 1929-39, when economic conditions were generally depressed; 1939-46, when the United States was approaching or was in a total w ar situation; 1947-52, the early postwar adjustment period and finally, 1953-69, a period when economic conditions were again generally prosperous. These subperiods cover a sufficiently wide range of economic condi tions to provide an indication of monetary and fiscal influences under a variety of economic circumstances. A summary of the regression results is reported in Table I. For the total period 1919-69, the mone tary variable is statistically significant and the fiscal variable is statistically insignificant at the 95 per cent confidence level. In the five subperiods, the monetary variable is significant in all but the subperiod cover ing the war years, 1939-46. The absence of a statis tically significant monetary variable in this period is probably due more to the inadequacies of the data than to a lack of a relationship. Because of price 13The other results are available upon request. 14The Almon lag technique, by constraining the distribution of coefficients to fit a polynomial curve of n degree, is designed to avoid the bias in estimating distributed-lag coefficients which may arise from multicollinearity in the lag values of the independent variables. The theoretical justifi cation for this procedure is that the Almon constrained estimate is superior to the unconstrained estimate because it will create a distribution of coefficients which more closely approximates the distribution derived from a sample of infin ite size. In order to minimize the severity of the Almon con straint, the maximum degree of the polynomial was used in each case. The maximum degree is equal to the number of lags plus one of the independent variables up to five lags. Following the convention established by Shirley Almon, “The Distributed Lag Between Capital Appropriations and Expenditures,” Econometrica, Vol. XXXIII, No. 1 (January 19 6 5), if there are n lags, t+ 1 and t —n —1 are both con strained to zero. The regressions were also run without constraining the beginning and ending values to zero, and the results are virtually identical. 15For a discussion of this criteria for selecting lags, see Leonall Andersen, “An Evaluation of the Impact of Monetary and Fiscal Policy on Economic Activity,’ Papers and Pro ceedings, Business and Economic Statistics Section, Ameri can Statistical Association, August 1969. NOVEMBER 1 9 6 9 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 Table I INDICATO RS O F M O N E T A R Y INFLUENCES O N AY = (A M ) AND FISCAL E C O N O M IC A C TIVITY do - | - Q i A m (AE) (AY) - f - a2 a e (Q u a rte rly First Differences —- Billions o f Dollars) Time Periods 11/1919 — 1/69 Lags* t-6 ao 1.92 111/1929 — 11/39 t-3 t-5 t-5 .36 5.62 ** (.5 1 ) (3 .1 6 ) — .51 6.32 (1 .3 9 ) 1 /1 9 4 7 — IV /5 2 t-1 0 — .07 (.2 8 ) (.5 4 ) 111/1939 — IV /4 6 2.89 a2 A e (sum) (4 .3 1 ) (2 .3 4 ) 11/1919 — 11/29 ai A m (sum) 3.65 5 .4 0 (3 .4 1 ) — 1.21 (.5 9 ) 13.82 R2 D-W .32 1.15 .35 1.58 -7 .9 7 (1 .9 5 ) .39 1.86 .35 .66 (.8 1 ) 1.60 -3 .3 7 implies that for every $1 increase in the money stock there w ill be a $5.50 increase in economic activity after three to five quarters. These are remarkably stable coefficients. In the postwar subperiods, however, the coefficients are substantially larger, and they are also different with respect to each other. In the 1947-52 period the coefficient on the mon etary variable is 13.82 with a tenr quarter lag in its impact. In the 1953-69 period the coefficient is 8.85 with a four-quarter lag. W hat does this variation in the value of the monetary coefficients imply? .72 The difference in the values of the coefficients between postwar subpe 1.71 (1 .0 7 ) (.7 4 ) (4 .7 0 ) riods is due to the different length of lags. These lags are selected on the N o te : R egression coefficients are the top figu res; their “ t” statistics appear below basis of minimum standard error of each coefficient, enclosed by parentheses. R 2 is the percent o f variations m the dependent variable which is explained by variations in the independent estimate, adjusted for degrees of free variable. D -W is the D urbin-W atson statistic. ♦Lags are selected on the basis o f m inim um standard error, adjusted f o r degrees dom. The results for the 1947-52 sub o f freedom . ** Fiscal variable om itted fo r 1919-29 because it increased the standard error o f period with a four-quarter, rather the estimate. than a ten-quarter lag, had a mone tary variable coefficient of 7.24. This value is quite controls, the measure of economic activity was sub close to the 8.85 value for the 1953-69 subperiod stantially understated between 1939 and 1946. There where the minimum standard error estimate was with fore, it is not surprising that the variables measuring a four-quarter lag. the influences of stabilization actions were not statistically significant in that period. The higher average value of the monetary co The fiscal influence was statistically significant in efficients in the postwar subperiods over the prewar only one of the five subperiods, 1947-52. However, subperiods is due to the weakness in the proxy se the sign of the coefficient is negative due to special lected to measure economic activity. The most com fa c to rs w h ic h a re e x p la in e d b e lo w . p le te m e a s u re o f e c o n o m ic a c tiv ity is n o m in a l GNP. However, it is available on a quarterly basis only In general, the results with respect to both mone since 1946. As previously indicated, our proxy for tary and fiscal variables for the period 1919-69 and economic activity tends on the average to grow more the subperiods conform closely to the results reported rapidly than nominal GNP because its “real” com in the AJ article for the period 1953-68. The co ponent is measured by industrial production. This efficient of determination (R2), which measures the factor did not bias the value of the coefficients in the per cent of variations in AY due to variations in AM prewar subperiods, because the Great Depression in and AE, is lower than that reported by AJ. This result sured that our proxy did not grow significantly is not surprising considering that our proxy is prob between 1919-29 and 1929-39. For the postwar sub ably inferior to nominal GNP as a measure of periods, however, the substantial and continuous in economic activity. creases in economic activity probably have caused an upward bias in the size of the monetary variable Recause of the major importance of the monetary coefficient presented in Table 1. For the 1953-69 influence, it is useful to look at the estimated co period, AJ had a monetary variable coefficient with efficients of the monetary variable during the various a four-quarter lag of 5.63, using nominal GNP. subperiods. In both of the prewar subperiods, 1919-29 This value is almost identical to prewar subperiods and 1929-39, the estimated coefficients on the mone tary variable are almost the same, around 5.50. This when economic activity is measured with our proxy. 1/1 953 — 1/69 t-4 (.8 4 ) (3 .5 1 ) (4 .1 2 ) 1.42 8.85 — .84 2.74 .47 Page 11 F E D E R A L R E S E R V E B A N K O F ST. LO U I S NOVEMBER 1 9 6 9 Thus, it is quite possible that if quarterly nominal GNP figures were available back to 1919, the esti mated value of the monetary coefficients would have been close to 5.50 in all subperiods. Testing Propositions The propositions which A J tested were whether monetary or fiscal influences were ( 1 ) stronger, (2) more predictable, and (3) faster in their impact on economic activity. They concluded that the evidence for the 1953-1968 period strongly favored the domin ance of monetary over fiscal influences. These same propositions are tested in this article and provide additional evidence that monetary influences consist ently have been stronger, more predictable, and faster in their effect on economic activity than have fiscal influences. The results are detailed below. Which is Stronger? — To measure the relative strength of monetary and fiscal influences, we need to know which has the largest impact on economic activity. This question can be answered by making an appropriate comparison of the coefficients of the monetary and fiscal variables. If the variables on which these coefficients are estimated have the same dimension and magnitude of variation, then the com parison can be made directly. These conditions, how ever, are not satisfied with these data. Money is a stock variable measured as first differences, and Fed eral Government spending is a flow variable meas ured as first differences at annual rates. Also, the degree of variation in the two variables differs sub stantially. In general, the fiscal variable has fluctuated more than the monetary variable (see Chart III on pages 16 and 17). To make the estimated coefficients of the mone tary and fiscal variables comparable for an assessment of their relative impact on economic activity, they were transformed into beta coefficients. The “sum” Table II BETA COEFFICIENTS Am (s u m ) 11/1919 — 1/69 .3 3 1 * 11/1919 — 11/29 .5 1 5 * 111/1929 — 11/39 .5 9 3 * 111/1939 — IV /4 6 1 /1 9 4 7 — IV /5 2 1/1 953 — 1/69 -.1 5 3 1.7 6 8 * .7 2 6 * A e (sum ) -.0 2 6 — -.8 0 3 .21 9 -2 .3 4 7 * — .159 N ote: "B eta coefficients” are equal to the estimated coefficient tim es the standard deviation o f the independent variable over the standard deviation o f the dependent variable. See A rthur S. Goldberger, E conom ic T h eory (J oh n W iley & Sons, 1964) p p . 197-98. •Significant at the 95% level o f confidence. Page 12 T able III t VALUES Am (sum ) Ae (sum ) 11/1919 — 1/69 4.31 11/1919 — 11/29 3.16 -0 - 111/1929 — 11/39 3.41 -1 .9 5 111/1939 — IV /4 6 - — .28 .59 .81 1 /1 9 4 7 — IV /5 2 3.51 — 4.12 1/1 953 — 1/1969 4.70 -1 .0 7 N ote: A t value is a statistical indicator o f the confidence one may have that the "tru e relation ship" between the independent and dependent variable has the same sign as the statistically estimated coefficient o f that relationship. beta coefficients are presented in Table II. For the whole period the monetary influence is large and statistically significant, while the fiscal influence is negative and statistically insignificant. This result also applies to each of the subperiods, except for W orld W ar II and the early postwar periods (193952). During the W orld W ar II years the monetary influence is statistically insignificant and negative, and the fiscal influence is insignificant and positive. For the early post-World W ar II years the fiscal influ ence is statistically significant and negative. This postwar regression result seems to be due to special factors which are outlined below. Which is More Predictable? — The monetary or fiscal variable with the more statistically significant coefficient is also more reliable in that its relationship to economic activity is more predictable. Statistical significance is measured by the t values of the co efficients of the monetary and fiscal variables when measured against the same dependent variable, which in this case was AY. A t value is a statistical in dicator of the confidence one may have that the “true relationship” between the independent and dependent variable has the same sign as the statis tically estimated coefficient of that relationship. The larger a t value, the more confidence we have that the monetary and fiscal variables are related to eco nomic activity. The t values of the sum coefficients are presented in Table III. For the whole period, the t value of the monetary variable is substantially larger than the t value of the fiscal variable. The same statement also holds with respect to the t values of the monetary and fiscal variables in the sub periods, with the exception of the w ar and early postwar periods (1939-52). Thus, in general, the monetary variable has a more predictable effect on economic activity than the fiscal variable. 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 NOVEMBER 1 9 6 9 C h a r t II Beta Coefficients of M o n e ta ry a n d Fiscal Influences First Differences N o te : B e ta c o e f f ic ie n t s a r e f o r th e M o n e y S to ck (AM ) a n d G o v e r n m e n t E x p e n d itu re s (AE), a n d a re c a lc u la te d a s th e p r o d u c t s o f th e re g re s s io n c o e ffic ie n ts f o r th e r e s p e c tiv e v a r ia b le s tim e s th e r a tio o f th e s ta n d a r d d e v ia t io n o f th e in d e p e n d e n t v a r ia b le s to th e s t a n d a r d d e v ia t io n o f E c o n o m ic A c tiv ity (AY). L a g s w e re s e le c te d o n th e b a s is o f th e m in im u m s t a n d a r d e r r o r o f e s tim a te . T h ese c h a r ts a r e d e r iv e d fro m th e s ta tis tic a l re s u lts w h ic h a re s u m m a r iz e d in T a b le I. W hich Works Faster? —The relative promptness of monetary or fiscal influences can be measured by observing which variable has a shorter time lag in influencing economic activity. This can be seen in the quarterly patterns of the regression coefficients after they have been transformed into beta coeffi cients. The latter are plotted in Chart II and are derived from the same set of statistical results sum marized in Table I. The fiscal variable has about the same impact as the monetary variable in the con temporaneous quarter during the total period 1919 1969. However, in the succeeding quarters the fiscal influence declines and becomes negative, while the monetary influence continues to be positive through the third lagged quarter. The quarterly pattern of the monetary influence in the subperiods is quite similar to that of the total period. The pattern of the fiscal influence varies irregularly over subperiods. However, in all subperiods except the war period 1939-46, the monetary variable has a consistently faster influence on economic activity than the fiscal variable. Page 13 F E D E R A L R E S E R V E B A N K O F ST. LO U I S Historical Review The statistical results reported above are estimated on the basis of the average response of economic activity to monetary and fiscal influences within each of the periods selected. A different w ay of looking at monetary and fiscal influences on economic activity is to investigate specific historic episodes. Chart III on pages 16 and 17 is designed to assist in that inves tigation. In the lower tier of the chart the monetary and fiscal variables are plotted as rates of change on a common axis. In the upper tier of the chart eco nomic activity is also plotted in its rate-of-change form.16 The most interesting comparisons are to be found where the monetary and fiscal influences are operat ing in opposite directions. In those periods the move ment in economic activity w ill indicate which influ ence is dominant. The monetary and fiscal variables move in opposite directions in the periods 1919-21, 1931-32, 1939, 1948-50, and 1966-67. In each of these years economic activity, after a short lag, moved in the same direction as the monetary variable and in the opposite direction to the fiscal variable. As a matter of fact, all cyclical movements in the money stock were followed by proportional cyclical move ments in economic activity. Of the twelve cyclical movements in economic activity from 1919 to 1969, eleven are preceded by corresponding movements in the money stock.17 The single exception is the deceleration in economic activity in 1951, which is discussed below. 1919-1929 — Although this period was one of gen eral economic prosperity, there were three cyclical declines in this ten-year period. The first and most severe occurred in late 1920 and early 1921. During the remainder of the 1920’s two shorter and milder declines occurred; in late 1923 to early 1924, and in 1927. NOVEMBER 1 9 6 9 Federal Government spending showed substantial fluctuation in the earlier part of the period and very little movement in the middle and latter part of the period. This experience reflected the demobilization after W orld W ar I and the conservative spending policies of the Harding and Coolidge administrations. The statistical results reported in Table I, page 11, omit the fiscal variable entirely for this subperiod, be cause its inclusion raises the standard error of the estimate (adjusted for degrees of freedom) and thus contributes nothing to the explanation of move ments in economic activity. This is not true of any other subperiod in this study. 1929-1939 — The first part of this period is un doubtedly the most depressed in the entire economic history of the United States. It was not the sharpness of the decline that was so disastrous. There were more rapid declines in both 1920 and 1937. Its dura tion was disastrous. Economic activity declined at an annual rate of 20 percent or more for ten of the eleven quarters between late 1929 and late 1932. Sustained recovery did not start until the middle of 1933, when 25 per cent of the labor force was un employed and the price level was 24 per cent below its 1929 level. This recovery lasted, with one signifi cant interruption in 1937, through the end of the period. Monetary influences during this period have been characterized by a number of observers as being especially ineffective. The results presented in this article indicate that quite the opposite was the case. Monetary influences played an important role in the declines in economic activity in 1929-33 and 1937-38, and in the recovery in the intervening years. Each of these cyclical movements in economic activity is matched by a corresponding movement in the money stock. Money switched from a 15 per cent rate of increase in the fourth quarter of 1919 to a 16 per cent rate of decline in the first quarter of 1921. This was the sharpest five-quarter deceleration in the money stock recorded during our fifty-year period. The money stock had pronounced, though milder, decelerations in 1923 and late 1926. Although the initial decline in the third quarter of 1929 apparently was not due to tight money in fluence (the money stock did not decline until the fourth quarter of 1929), the fact that the economic decline lasted for more than three years is associated with a decline in the money stock.18 The initial five quarters of decline in the money stock were rela tively mild. After reaching an annual 9 per cent rate of decline in the first quarter of 1930, it slowed to a 3 per cent rate of decline in the fourth quarter of 1930. Then, for the next four quarters, the money stock decelerated substantially and reached an an nual rate of decline of 18 per cent in the fourth 16Rates of change are used to allow comparisons over long time periods on a similar basis. 17Milton Friedman and Anna Schwartz made a similar obser vation in “Money and Business Cycles,” Review of Eco nomics and Statistics, February 1963. 18Milton Friedman and Anna Schwartz, A Monetary History of the United States, (Princeton, New Jersey: Princeton University Press, 19 6 3 ), chapter 7, go into considerable detail describing how Federal Reserve actions dominated movements in the money stock during this period. Page 14 F E D E R A L R E S E R V E B A N K O F ST. LO U I S quarter of 1931. For the next six quarters the de clines became progressively smaller. Finally, at the end of 1933 the money stock registered the first quarterly increase since the third quarter of 1929. The money stock had shown continual quarterly declines for almost four years. Economic activity moved parallel with the money stock pattern in 1929-33. Although the first year de cline was substantial, it was less than the four-quarter decline in 1920-21, and only moderately greater than the four-quarter decline in 1923-24. In the first half of 1931 the rate of decline actually slowed, respond ing to the less restrictive monetary influences. How ever, in the next year the decline in economic activity increased sharply. In the year ending June 1932, it declined by 37 per cent. In late 1932, economic activ ity finally stopped declining, and in early 1933 it started to increase. This increase generally continued until the middle of 1937, when it was temporarily reversed by the tight money influence which de veloped in late 1936. During this period fiscal influences, as measured by changes in Federal Government expenditures, were quite erratic. They were highly expansionary in the years 1931, 1933 to early 1934, 1936, and 1938. On the other hand, they were restrictive in the years 1932, 1935, and 1937. This pattern sometimes con formed with and sometimes opposed monetary in fluences. But in every case economic activity moved consistently with the direction and magnitude of monetary influences. 1939-1946 —Data for the w ar years are presented to make the series complete. However, with compre hensive price controls tending to create a discrepancy between the equilibrium and observed growth rate in economic activity, there is little to be learned about monetary and fiscal influences from this period. Our results indicate that the monetary variable was not statistically significant during this period. The fiscal variable had a strong positive influence in the quarter in which the Government spending took place, but tended to “washout” after five quarters, leaving only a small positive net influence. 1947-1952 — There were three cyclical expansions in this period: early 1947-48, late 1949 and 1950, and in 1952. There were cyclical contractions in the inter vening years. The movements in the money stock did a good job of “tracking” the movements of economic activity during this period, with the single exception of the deceleration in economic activity which oc curred in 1951. This is the only deceleration in NOVEMBER 1 9 6 9 economic activity in the fifty-year period which was not anticipated by movements in the money stock. A quite plausible explanation for this phenomena is provided by Friedman and Schwartz.19 In March 1951 the United States Treasury Department and the Federal Reserve reached an “Accord,” which per mitted the latter to abandon its war-induced policy of pegging the price of Government bonds. Even though the Federal Reserve did not take advantage of this increased flexibility in policy actions immedi ately, the public act of abandoning support of the Government bond market greatly reduced the ap parent liquidity of the public. The public was no longer assured that conversions between Government bonds and money could take place at a fixed and known price. This caused a substantial, one-time in crease in the liquidity demand for money balances relative to income, and a decrease in the velocity of money in a period when velocity had typically been rising. This experience suggests not only that permanent changes in the demand for money independent of changes in income can weaken the observed relation between money and income, but that such changes in money demand are relatively rare. Such changes generally have been associated with some specific historic event which changes the previous institutional relations with respect to the liquidity of nonmoney assets. The other unique factor about this subperiod is that the fiscal variable is statistically significant and negative. Weidenbaum has provided a plausible ex planation for this.20 He has shown that Government spending influences economic activity not when the bills are paid and the goods are delivered to the Gov ernment, but when the orders are placed with in dustry, which must then hire employees and open plants to produce the products. This discrepancy does not lead to serious bias in measuring Government spending except when there is a sharp acceleration or deceleration in this variable. This was clearly the case in the Korean W ar, when Government spending went from an annual rate of decline of 38 per cent in the second quarter of 1950 to an annual rate of increase of 83 per cent in the first quarter of 1951 and then fell to an annual rate of increase of 13 per cent in 1952. This “whiplash” 19Friedman and Schwartz, pp. 598 and 612. 20M urray L. Weidenbaum, “The Federal Government Expend ing Process,” Federal Expenditure Policy for Economic Growth and Stability, (Washington, D.C.: Joint Economic Committee of Congress, U.S. Government Printing Office, November 19 5 7 ), pp. 493-506. Page 15 C h a r t III Rates o f C h a n g e C h a n g e s in M o n e y S u p p ly a n d G o v e r n m e n t E x p e n d it u r e s in R e la tio n to C h a n g e s in Economic A c t iv it y Q u a rte rly d a ta a t a n n u a l ra te s . p o s tw a ry e a rs (1947-74) c o n s e q u e n tly has been e n la rg e d to fa c ilita te c o m p a ris o n s a m o n g the th re e series. [L E c o n o m ic A c tiv ity is m e a s u re d b y the sc a le d p ro d u c t o f the C o n su m e r P rice In d e x (CPI) a n d the In d u s tria l P ro d u c tio n In d e x (IPI) Page 16 Rates o f C h a n g e S o urces: M o n e y S u p p ly : 1919-46, M ilto n F rie d m a n & A n n a J .S c h w a rtz “ A M o n e ta ry H is to ry o f th e U n ite d S ta te s ; 1947-19 6 9 "B o a rd o f ♦The m a g n itu d e o f flu c tu a tio n s in th e th re e s e rie s p lo tte d d e c re a s e d c o n s id e ra b ly a fte r W o r ld W a r II. The ra te o f c h a n g e scale fo r the m u ltip lie d b y G ro s s N a tio n a l P rod uct(G N P ) in the ba se y e a rs 19 57-59: C h a r t III /CPI • IPlV $ 4 5 7 .4 b illio n - E co n o m ic A c tiv ity vism) G o v e rn o rs o f the F e d e ra l R eserve S ystem ; F e d e ra l G o v e rn m e n t S p e n d in g : 1917-28, estim a te d fro m U.S. H is to ric a l S ta tis tic s , B u re a u o f C ensus; 1929-1945, e s tim a te d fro m N a tio n a l Inco m e & P roducts S up p le m e n t, U.S. D e p a rtm e n t o f C o m m e rc e ; 19 46-69, Federal Reserve B ank o f St. L o u is; E conom ic A c tiv ity : In d u s tria l P ro d u c tio n In d e x , B o a rd o f G o v e rn o rs o f the F e d e ra l R eserve System ; C o n su m e r P rice In d e x , U.S. D e p a rtm e n t o f L a b o r 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 I S N OV E M B E R 1 9 6 9 1953-1969 —This is the period which was covered in the original AJ study. W hile their measure of economic activity, nominal GNP, differs from the proxy used here and described in footnote 10, their results and ours are similar in most respects, as can be seen in the comparison of summary results in Table IV. In each case the monetary and fiscal measures are the same ( the money stock and government spending). Only the measure of economic activity differs. The first equation is based on our proxy of economic activity and is drawn from Table I. The second equation is based on nominal GNP.22 In each case the monetary variable has a positive coefficient which is statistically significant and the fiscal variable has a coefficient which is statisticially insignificant and close to zero in value. Both equations are suffi ciently well specified to pass the Durbin-Watson ( D- W) test for autocorrelation, and the lag struc tures are the same when selected on the basis of minimum standard error of estimate. The value of 21This is shown by the following regression: 1/1947 - IV/1956 (Quarterly First Differences) AY = 1 .1 1 + 7.82 2 A M t-i - 1.13 S A E t- i ( .48) (3 .3 1) i= o (1.7 6 ) i= o R2 = .27 D -W = .98 where 2 stands for sum of monetary or fiscal influence from period t to period t-3. Lags were selected on the basis of minimum standard error of estimate adjusted for degrees of freedom. 22The results with respect to nominal GNP differ slightly from the original Andersen-Jordan results because of the different lag structure. The lag structure in their original article (contemporaneous and three-lag values) was selected on the basis of minimum standard error of the coefficient attached to the monetary variable. The present lag structure (contemporaneous and four-lag values) was selected on the basis of minimum standard error of the entire equation adjusted for degrees of freedom. In this case, the different criteria did not change the results in any significant way. Page 18 MONETARY A N D FISCAL INFLUENCES O N E C O N O M IC ACTIVITY, MEASURED AS A PROXY (A Y ) A N D AS N O M IN A L G N P (A G N P ) ( 1 /1 9 5 3 — 1 /1 9 6 9 ) Dependent V ariable la g s * <32 AE Ay f-4 1.42 ( -72) 8.85 (4 .7 0 ) - .84 (1 .0 7 ) .47 1.71 A gnp t-4 2.59 (3 .1 9 ) 5.63 (6 .9 4 ) .08 ( .2 4 ) .65 1.78 Constant Term ao M onetary Influence o If we had chosen a somewhat longer time period in which to measure the impact of monetary and fis cal influences, the strong negative offset estimated with respect to AE would have lost; its statistical significance. W e would have had results for the early post-World W ar II subperiod which were comparable to the results of the other subperiods.21 Table IV 5 < movement in Government spending follows by about two or three quarters an equally sharp movement in economic activity. As a result, AY and AE moved in opposite directions in this period. This is the cause of the statistically significant negative coefficient of AE with respect to AY. Fiscal Influence R2 D-W N ote: Regression coefficients are the top fig u re s; their “ t” values appear below each coefficient, enclosed by parentheses. R 2 is the per cent o f variations in the dependent variable which is explained by variations in the independent variable. D -W is the D urbin-W atson statistic. the monetary coefficient is greater with the proxy measure of economic activity ( AY) than with nom inal GNP (AGNP), which is due to the greater aver age value and amplitude of the proxy. The coefficient of determination ( B2) is larger with AGNP than with AY. This is as would be expected if, as seems reasonable, nominal GNP is superior to our proxy as an indicator of economic activity. There were four cyclical declines in this period, each of which was led by a decline in the money stock. Government spending registered three cyclical declines, two of which corresponded to periods of decline in the money stock and one (in 1967) which did not. As noted in our investigation of earlier pe riods, economic activity declined following a decline in Government spending only when accompanied by a decline in the money stock. Determining the Values of the Monetary and Fiscal Variables An assessment of the reliability of the relations presented above will depend upon whether the estimated coefficients for the monetary and fiscal variables are exogenous. This problem arises in all statistical work, and no fully satisfactory solution has been found to test for exogeneity in either single equation or in large structural models.23 However, in the single equation test of monetary and fiscal in fluences on economic activity employed here, one potential source of bias is found in the so-called “reverse-causation” argument. This asserts that the ob served correlation between M and Y is not because changes in M cause changes in Y, but because changes 23In statistical theory, a variable is exogenous if it is un correlated with the “true” error term of the equation. Unfortunately, only the measured error term in any equa tion is observable, so this test cannot be made. 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 in Y cause changes in M. If this possibility cannot be rejected, then a more elaborate statistical test is needed to compare monetary and fiscal influence on economic activity.-4 The fiscal variable used here (total Federal Gov ernment spending, including transfer payments) is generally accepted as being determined by the fiscal authorities and not by the behavior of the public in the marketplace. For the purposes of our test we will assume that the fiscal variable is statistically inde pendent or exogenous. W ith respect to the monetary variable (the narrowly defined money stock), there is considerable controversy as to whether its value is determined by the monetary authorities or by the public. For this reason we w ill concentrate our em pirical investigation on the money variable. It will be shown that the reverse-causation argument is not supported by the evidence. In addition, the available evidence indicates that the actions of the monetary authorities dominate the movements in the money stock. Does Economic Activity Affect Money? In order to evaluate the significance of the reverse-causation argument, w"e need some indicator of the public’s potential influence on the money stock. The indicator chosen is our proxy variable for eco nomic activity (Y). This proxy has two advantages: first, it is the broadest available measure of aggregate economic activity and, as such, most actions of pri vate decision-making units in the economy are re flected in it. Second, it allows us to consider directly the important statistical question of whether move ments in Y lead to movements in M. In order for economic activity to affect the money stock, it must operate through some transmission mechanism.25 The Brunner-Meltzer money stock identity provides a useful structure within which to consider the several ways that economic activity could affect the money stock.28 In this context the money 24A t the least, one would need an equation to explain the monetary and fiscal variables by factors which themselves were independent of income. 25The approach used here to test for the influence of the public on the money stock was suggested by the work of Leonall C. Andersen, “Additional Empirical Evidence on the Reverse-Causation Argument,” this Review, August 1969. 26For a systematic exposition of this approach, see Albert Burger, “An Analysis and Development of the BrunnerMeltzer Nonlinear Money Supply Hypothesis,” Working Paper No. 7, Federal Reserve Bank of St. Louis, May 1969. NOVEMBER 1 9 6 9 stock (M ) is defined as the product of the money multiplier (m ) and monetary base (B ): M = mB The sources of the monetary base consist of various kinds of credit extended by the monetary authorities to the rest of the economy. The use of the monetary base is divided between currency holdings of the nonbank public and reserves of commercial banks. The money multiplier, which is defined as m = — ------- 1 + k ---------r (l+T + g)+k , is largely determined by the behavior of the public, including commercial banks; k represents the ratio of private currency holdings to private demand depos its; t represents the ratio of private time deposits to private demand deposits; g represents the ratio of Government deposits in commercial banks to private demand deposits; and r represents the ratio of total bank reserves and total bank deposits.27 Economic Activity and the Monetary Base — The influence of economic activity on the money stock could operate either through the monetary base (B) or the money multiplier (m). To test whether eco nomic activity has influenced the monetary base, regressions were run for the total period (1919-69), and for each of the five subperiods reported above. The results are presented in Table V. For the entire 50-year period changes in the monetary base have a statistically significant relation with changes in eco nomic activity. However, economic activity explains at most only 4 percent of the variance of the changes in the monetary base; that is, the R2 was .04. For every $1 billion increase in economic activity, there is associated only an $8 million increase in the base in the same quarter. Equally weak relations between AY and AB were found in the subperiods. Only the first (1919-29) and the last (1953-69) subperiods had statistically significant coefficients, while the R2 varied between .01 and .15. These results imply that the public, operating through economic activity, has only a small effect on the monetary base, and that this effect has varied 27For a detailed discussion of the determinants of the multi plier and its influence on the money stock, see Philip Cagan, Determinants and Effects of Changes in the U.S. Money Stock, 1875 -19 60 , (New York: National Bureau of Economic Research, 19 6 5 ); and Jerry L. Jordan, “Elements of Money Stock Determination”, this Review, October 1969. 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 I S NOVEMBER 1 9 6 9 substantially over time in both Table VI degree and significance. Allow THE INFLUENCE OF STABILIZATION (FR), EVEN-KEEL (A D ) A N D ing for the influence of lagged F IN A N C IA L (r-rn) OBJECTIVES OF THE M O NETARY AUTHORITIES values of AY on AB does not O N THE M O NETARY BASE (A B ) change the results presented in Table V, except for 1947-52 A B t = c i FR + C2 A d C3 (r-rn ) (Q u a rte rly First Differences — B illions o f Dollars) when the R2 increases to .24 Time S tabilization Even-Keel Dummy and the coefficient becomes Financial Periods Objective Objective Objective V ariab le* R2 D-W statistically significant. e i FR C2 A d C3 (r-rn ) Monetary Authorities and the 11/1929 — IV /1 9 3 9 .33 — .219 .401 .553 .59 1.75 Monetary Base — Another po (6 .7 6 ) (2 .8 6 ) (3 .6 5 ) (1 .9 5 ) _ 1 /1 9 4 0 — IV /1 9 5 2 .01 .070 .46 9 .46 1.90 tential source of control of the (6 .3 2 ) (.3 7 ) (.8 1 ) monetary base is through the 1/1 953 — IV /1 9 6 8 .19 .018 .69 1.84 .123 .415 actions of the monetary au (2 .2 8 ) (1 -7 3 ) (2 .6 9 ) (5 .5 8 ) thorities. There have been a N ote: Regression coefficients are the top figu res; their “ t” values appear below each coefficient, number of studies which have enclosed by parentheses. R 2 is the p ercent o f variations in the dependent variable which is explained by variations in the independent variable. D -W is the Durbin-W atson statistic. related policy targets of the ♦In 1929-39 the Dumm y V ariable is designed to accou nt fo r the im pact on the m onetary base o f the rise in the price o f gold in February 1934. I t assumes the value o f 1 fo r the first and second monetary authorities, such as quarters o f 1934 and zero fo r all other quarters. In 1953-68 the Dumm y V ariable accounts f o r the change in presidential adm inistration. It assumes a value o f zero from 1/1953 to 11/1962 and a income stabilization, to various value o f one from III/1962 to IV /1968. indicators of monetary actions, such as the money stock (Dewith respect to income, employment, and prices, re wald and Johnson), total member bank reserves flected in the Federal Reserve Open Market Commit (Dewald), free reserves (W ood), and the monetary tee policy statements as proxied by the level of free base (Keran and Babb). A ll these studies conclude reserves (FR); an even-keel objective with respect that the monetary authorities have dominated move to Government debt financing, measured by ments in the money stock or some closely allied vari changes in the national debt ( A D) ; and a financial able. The last named study will be briefly reviewed objective with respect to stability of the financial here because it deals explicitly with control of the system, measured by deviations of Corporate Aaa monetary base by the authorities. bond yields from “normal” yield levels (r-rn).28 In Keran and Babb found that a large proportion addition, economically “random” events, such as of the movements in the monetary base can be ex changes in the price of gold in 1934 and changes in plained by the desire of the monetary authorities to presidential administrations, have also influenced the achieve three objectives: a stabilization objective actions of the monetary authorities with respect to Table V changes in the monetary base ( AB) . These events THE INFLUENCE O F E C O N O M IC ACTIVITY (A Y ) are represented by “dummy variables” in Table VI. O N THE M O NETARY BASE (A B ) ABt = bo + b i A Y t (Q u a rte rly First Differences — Billions o f Dollars) Time Periods bo bi Ay !? . D-W 11/1919 — 1/1 969 .31 (8 .1 7 ) .008 (2 .9 8 ) .04 .58 111/1919 — 11/1929 -.0 1 2 (.6 8 ) .009 (2 .4 2 ) .12 .64 111/1929 — 11/1939 .20 (3 .3 9 ) .004 (.4 1 ) .02 .85 111/1939 — I V / 1946 .99 (7 .3 0 ) -.0 1 2 (1 .1 4 ) .01 1.09 1 /1 9 4 7 — IV /1 9 5 2 .12 (1 .4 5 ) .012 (1 .7 6 ) .08 .87 1/1 953 — 1 /1 969 .34 (6 .2 4 ) .012 (3 .4 5 ) .15 .87 N ote: Regression coefficients are the top figures: their t values appear below each coefficient, enclosed by parentheses. R 2 is the percent o f variations in the dependent variable which is explained by variations in the independent variable. D -W is the Durbin-W atson statistic. Page 20 Two of the three subperiods considered by Keran and Babb were approximately the same as subperiods in the present study (1/1953 to IV/1968) and 1/1940 to IV/1952). Another subperiod in that study was reestimated to match the 1929-39 subperiod in this study. The results are presented in Table VI. In each case, fifty percent or more of the variations in AB are explained by the actions of the monetary authori ties. In contrast, where the actions of the public were assumed to operate, the best results explained fifteen percent or less of the variance in AB (see 28They have also shown that in the 1953-68 period, Federal Reserve open market operations (adjusted for changes in reserve requirements) were also explained by the same three objectives plus an additional money market objective, which in effect offset the noncontrolled sources of the monetary base. 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 NOVEMBER 1 9 6 9 Table V ). The acceptable Durbin-Watson statistics in Table VI suggest that no important explanatory varia bles have been omitted from the monetary authori ties’ explanation of AB. On the other hand, low Durbin-Watson statistics in Table V imply that im portant explanatory variables have been omitted from an economic activity explanation of AB. Economic Activity and the Money Multiplier — Another channel through which economic activity could influence the money stock would be through its influence on the money multiplier. As indicated above, most of the ratios which are involved in determin ing the multiplier depend upon the behavior of the public, including commercial banks. Table VII presents the results relating changes in the money stock ( A M ) to changes in the monetary base ( AR) and economic activity (AY) . Assuming that the monetary authorities determine movements in the base, and that the public operating through economic activity influences the money multiplier, our results indicate that for the total period (1919-69) both the monetary authorities (AR) and economic activity ( AY) explain 67 per cent of the variance in AM. However, the beta coefficients, which indicate the “typical” influence of each independent variable on the dependent variable, show that the monetary au thorities operating through the base ( AR) have an impact on the money stock ( A M) which is 3V2 times as large as the public influence operating through economic activity ( A Y ). The results for the subperiods are substantially the same as for the total period. The coefficient for the monetary base is statistically sig nificant in all subperiods, while that for economic activity is statistically significant in only the first two subperiods (from 1919 to 1939). There was one sub period (1929-39) where the beta coefficients in dicated that economic activity was more important than the monetary base in determining movements The values of the coefficients in Table VI for the prewar (1929-39) and postwar (1953-68) subperiods were similar with respect to the income stabilization objective (FR) and the financial stabilization objec tive (r-rn),29 supporting the hypothesis that the mon etary authorities have acted in a largely consistent manner in controlling the monetary base (AR) . Dur ing the war and early postwar period (1940-52), the Federal Reserve followed a single-minded policy of supporting the Government bond market. The results in Table VI reflect this, with only the even-keel vari able statistically significant in that subperiod. The results presented here indicate that it is the behavior of the monetary authorities (Table VI) rather than economic activity (Table V ) which have dominated movements in the monetary base (AR). There is no evidence that the reverse-causation argu ment holds with respect to AR. 29For an explanation of all variables used in Table VI and of the difference in the even-keel sign between (19 2 9 -3 9 ) and (19 5 3 -6 8 ), see Keran and Babb, pp. 9-15. Table VII RELATIVE INFLUENCE OF E C O N O M IC ACTIVITY (A Y ) A N D THE M O NETARY BASE (A B ) O N AMt = d, + THE M O N E Y STOCK (A M ) d, A Y t + d2 ABt (Q u a rte rly First Differences — B illions o f Dollars) Time Periods Beta Coefficients do d iA Y t d 2A B t R2 D -W Ay Ab 11/1919 — j/1 9 6 9 .094 (1 .4 4 ) .023 (4 .6 1 ) 1.89 (1 7 .7 8 ) .67 1.32 .198 .755 11/1919 — 11/1929 .075 (1 .7 2 ) .025 (2 .4 9 ) 2.41 (5 .5 7 ) .61 1.43 .288 .64 0 111/1929 — 11/1939 .064 (.6 3 ) .063 (4 .4 5 ) .71 (2 .8 6 ) .42 1.21 .54 6 .351 111/1939 — IV /1 946 .82 (2 .5 2 ) .022 (1 .4 5 ) 1.69 (6 .3 5 ) .57 1.73 .181 .791 1 /1 9 4 7 — IV /1 9 5 2 .37 (2 .2 7 ) .019 (1 .4 1 ) 1.46 (3 .6 3 ) .46 .9 7 .230 .598 .032 (-2 4 ) .014 (1 .9 2 ) 1.94 (8 .0 0 ) .59 1.56 .166 .696 1 /1 953 — 1/1 969 N ote: R egression coefficients are the top fig u res; their “ t ” values a p p ear below each coefficient, enclosed by parentheses. R 2 is the p ercent o f variations in the dependent variable which is explained by variations in the independent variable. D -W is the Durbin-W atson statistic. 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 NOVEMBER 1 9 6 9 Table V III RELATIVE INFLUENCE OF E C O N O M IC A C TIVITY M O NETARY BASE (A B ) O N THE M O N EY Amt = eo (AY) A N D THE MULTIPLIER (A m ) 61 A Y t -)- b 2 A B t (Q u a rte rly First Differences — B illions o f Dollars) Time Periods H /1 9 1 9 — 1/1 969 Beta Coefficients e0 e iA Y t e2ABt R2 D-W Ay Ab .001 (•1 4 ) .001 (4 .0 3 ) -.0 3 (4 .2 0 ) .12 .96 .19 4 — .288 11/1919 — 11/1929 .01 1 (1 6 2 ) .004 (2 .5 3 ) -.1 6 (2 .3 7 ) .16 1.45 .438 -.4 0 3 111/1929 — 11/1939 — .004 (.3 1 ) .0 0 7 (4 .0 4 ) — .16 (5 .1 6 ) .50 .98 .441 — .575 111/1939 — I V / 1 946 .038 (3 .0 5 ) .001 (1 .3 2 ) -.0 3 4 (3 .3 4 ) .32 1.27 .274 — .528 1 /1 9 4 7 — IV /1 9 5 2 .008 (2 .2 4 ) .0001 (1 .4 5 ) — .028 (2 .9 8 ) .24 .95 .062 — .308 .0001 (2 .1 8 ) -.0 1 1 (2 .6 6 ) .09 1.56 .10 0 — .332 1/1 953 — 1/1 969 .0001 (.1 5 ) N ote: Regression coefficients are the top figu res; their " t ” values a p p e a r below each coefficient, enclosed by parentheses. R 3 is the p ercent o f variations in the dependent variable which is explained by variations in the independent variable. D -W is the Durbin-W atson statistic. in the money stock. The strength of the economic activity variable in that period reflects the substantial decline in the multiplier. The multiplier declined dur ing the early part of that period (1929-33) due to a change in the currency-deposit ratio (k), which reflected the run on banks by households as they at tempted to convert their bank deposits into currency. These results are not changed when lagged values of AY and AB are added to explain AM. W ith four lags the statistical significance of the coefficient for AY disappears in 1919-29, while in 1953-69 the co efficient for AY becomes negative. This latter result is inconsistent with the usual reverse-causation argu ment, which asserts a positive relationship. The results presented in Table VII imply that eco nomic activity has had some influence on changes in the money stock, presumably through its influence on the money multiplier, especially in the important 1929-39 period. However, the observed influence of economic aotivity on the money stock would over state its true influence if offset by the actions of the monetary authorities operating through the monetary base. For example, if part of the actions of the mone tary authorities had been designed to offset the in fluence of economic activity on the money multiplier, then the observed association of economic activity and the money stock would be, at least, statistically ambiguous. Table VIII indicates this is the case. It shows the relative impact of the public operating through eco nomic activity ( AY) , and the monetary authorities Page 22 operating through the monetary base ( AB) , on the money multiplier ( Am) . In the total period and in all subperiods the influence of economic activity ( AY) is positive and the influence of the monetary base ( AB) is negative. The beta coefficients indicate that in all subperiods (including 1929-39), the mone tary authorities offset or more than offset the influence of the public on the money multiplier. Thus, the significance of the association of economic activity and the money stock reported in Table VIII is weak ened, because those movements in the money multi plier induced by economic activity have been offset by changes in the monetary base. The conclusions which can be drawn from these statistical tests are (1 ) that the monetary base is the dominant factor in determining movements in the money stock, both directly (Table VII) and indirectly (Table V III), by offsetting other influences on the money stock; and (2) that the monetary authorities are the dominant factor in determining movements in the monetary base (Table VI). Thus, for the pur poses of the single equation regressions used in this article, there are no statistical reasons for not treat ing the money stock as substantially controlled by the monetary authorities in all subperiods (including 1929-39). Summary The intent of this article is to measure the impact of monetary and fiscal influences on economic activ ity over as long a period of American histoiy as available data permit (1919-69), and for selected F E D E R A L R E S E R V E B A N K O F ST. LO U I S subperiods. This was done to see if different financial institutions, Government involvement in the economy and general economic conditions which existed dur ing this long period have substantially affected the relative impacts that monetary and fiscal influences have had on economic activity. For the whole period and for each of the sub periods (except the war years 1939-46), the rela tive impacts of monetary and fiscal influences have been remarkably stable. Changes in the money stock (the indicator of monetary influence) have consist ently had a larger, more predictable, and faster im pact on changes in economic activity than have changes in Federal Government spending (the in dicator of fiscal influence). This basic relationship is observed in the economically depressed period of 1929-39 and in the prosperous periods 1919-29 and 1953-69.30 A historical investigation of the past fifty years reveals that in every case where the monetary varia ble and the fiscal variable moved in opposite direc tions, economic activity moved in the direction of the monetary variable and opposite in direction to the fiscal variable. Every cyclical movement in the money NOVEMBER 1 9 6 9 stock since 1919 has been followed by a proportional cyclical movement in economic activity. Both the statistical results and the historical in vestigation provide strong support for the case that monetary influences have a significant impact on eco nomic activity over the business cycle. An important implication of these results is that monetary policy should be given a central role in any economic stabilization program. 30The author was surprised at the consistency of the mone tary influence during the various subperiods. Before con ducting the research reported in this article, he considered that monetary influences on economic actiyity were strongly significant only during periods of generally strong business conditions like the 1920 s and 19 60 ’s, while fiscal influences were dominant in periods of generally weak business condi tions like the 1930’?. In the March 1967 issue of this Review (page 1 4 ) , he said: “during the 1930’s business expectations of the future were so badly impaired by the depression ex perience that even large change in financial variables like money, . . . would not be sufficient to induce new invest ment and consumption.” In the November 1967 issue (page 8) he made the same statement in a slightly different context: “If the forces which create strong private demand should disappear, i.e., loss of optimistic expectations by firms and households, the rate at which money is made available to the economy may not result in a predictable change in income.” The results reported in this article do not support the above quotations. Monetary influences have dominated fiscal influences on economic activity in both periods of secular boom and periods of secular recession. This article is available as Reprint No. 47. The Appendix to this article, which begins on the next page, considers the fiscal influence in more detail. APPENDIX M ost of the readers of this article w ill not be sur prised w ith ou r results, w h ich indicate th at m onetary influences h ave had a m ajor im pact on econom ic activity. M ost econom ists b elieve th at m oney m atters. H ow ever, th ey m ay be surprised at the consistently w eak or non existent fiscal influence w h ich our results im ply. This A pp endix w ill explore one possible explanation fo r these surprising results. T able IX shows the im pact o f fiscal influences (m easured b y changes in F ed eral G overn m ent spending) on eco nom ic a ctivity w ith ou t taking m on etary influences into consideration. N otice in T able IX that the sign and statistical significance o f the fiscal influences differ sub stan tially fro m the results presented in T able I, w h ere m on etary influences are explicitly accounted for. In T a b le IX, the fiscal influences are positive and statistically significant fo r the entire period and fo r each o f th e sub periods. T he single exception is the period 1 9 4 7 -5 2 , w h ere special factors explained in the text tend ed to bias the fiscal m easure. On the oth er hand, the fiscal influences m easured in T able I w e re statistically insigni ficant and n eg ative fo r th e en tire p erio d and fo r each o f the sub-periods, again w ith the single exception o f the p erio d 1 9 4 7 -5 2 . A com parison o f the results in T able I and T able IX in dicates th a t fiscal influences on econom ic a ctivity m ay be strongly dep enden t upon how G overn m ent spending is financed. F ed eral G overn m ent expenditures can be paid fo r either b y tax receipts from the public, b y issuing bonds to the public, or b y expansion o f the m oney supply. The results in T able IX, w h ere the fiscal influences are positive and statistically significant, do not d iffer entiate b etw een the altern ative m ethods of financing G overn m ent spending. T able I, h o w ever, b y explicitly including changes in the m oney stock, im plies that G overn m ent spending financed b y m oney creation is accounted fo r b y the m on etary variable. O nly th at portion of G overn m ent spending w h ich is financed b y taxation and selling bonds to th e p u b lic is m easured b y the fiscal v a ria b le .1 These results ind icate that th e strength o f the fiscal influences on econom ic a c tivity is dep enden t upon the m ethod o f financing G overn m ent spending. If spending is financed b y increases in the m oney stock, it has a m easurable effect on econom ic activity. If it is financed b y taxes or selling bonds to the public, there is no m easurable fiscal influence on econom ic activity. Table IX IM PACT OF FISCAL INFLUENCES ON E C O N O M IC ACTIVITY A y = f„ + (AE) (AY) fi AE (First Differences — Billions o f Dollars) R2/D-W Lags* fo f l A E (sum ) H /1 9 1 9 — 1 /1 969 t-6 3.83 (5 .1 6 ) .81 (3 .0 1 ) .13 .96 11/1919 — 11/1929 t-7 1.73 (2 .4 4 ) 3.2 7 (2 .7 8 ) .41 1.58 111/1929 — 11/1939 t-1 0 -3 .4 5 (2 .2 5 ) 18.28 (2 .1 8 ) .14 1.49 111/1939 — IV /1 94 6 t-4 4.29 (2 .9 1 ) .53 (2 .1 1 ) .61 1.43 1 /1 9 4 7 — IV /1 952 t-4 9.38 (3 .9 1 ) — 1.49 (1 .7 6 ) .09 .81 1 /1 9 5 3 — M /1 9 6 9 t-3 6.08 (2 .9 5 ) 1.71 (2 .1 5 ) .08 1.13 N ote: R egression coefficients are the top figu res; their “ t” values appear below each coefficient, enclosed by parentheses. R 2 is the p ercent o f variations in the dependent variable which is explained by variations in the independent variable. D-W is the D urbin-W atson statistic. ^Selected on the basis o f m inimum standard error o f estimate, adjusted fo r degrees o f freedom . 24 Digitized for Page FRASER M ost w riters on pu blic finance and fiscal p o licy2 h ave g enerally asserted th at th e im pact o f G overn m ent spend ing on th e econom y is influenced b y h o w th e spending is financed. The most expansionary m ethod is through increasing the m oney supply, and the least expansionary m ethod is through increases in taxes. T he results p re sented in this A pp endix are consistent w ith those assertions. 'The role of commercial banks is ambiguous in this analysis. If banks buy Government bonds and induce the Federal Reserve to increase total reserves, it is treated as an increase in the money supply by the mon etary authorities. However, if the com mercial banks buy Government bonds, independent of any increase in reserves, it is treated the same as a purchase of bonds by the general public. -Richard Musgrave, The Theory of Public Finance, (New York: McGraw-Hill, 19 5 9 ); Alvin Hansen, Monetary Theory and Fiscal Policy, (McGraw-Hill Company, 19 4 9 ); and O. H. Brownlee and E. D. Allen, The Economics of Public Finance, Second Edi tion, Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 19 5 6 ). The Effects of Inflation (1960-68) by ALBERT E. BURGER I HE MAJOR GOALS of economic stabilization policy are a full-employment level of real output and a stable price level. Over the first five years of the current economic expansion which began in early 1961, the goal of full employment dominated discus sions of policy. In January 1966 the interim target goal of 4 per cent unemployment was reached. Since early 1966 most studies of unemployment have been concerned with unemployment by specific classes or groups rather than total unemployment. At an aggregate level, the problems of achieving a stable price level have increasingly dominated the attention of the policymakers and the public. The objective of a full-employment level of real output is a desirable goal of economic stabilization policy. If the economy is operating at less than its potential level of real output there is waste, not only from the standpoint of individuals who are un employed, but from an aggregate viewpoint. There is less real output being produced than the economy could produce, given its endowment of factors of production, the degree of skill and training of the labor force, and the available technology. By mov ing from a position of underemployment to one of maximum utilization of resources, a larger flow of real goods may be made available for all members of the economy. Most people can see the inherent dangers of socalled hyperinflation. Germany in the post-World W ar I period, when prices rose by a factor of 100 billion in one year, the Eastern European countries of Poland and Hungary in the 1921-23 period, and China in the post W orld W ar II period, stand out as very clear examples of the severe political as well as economic consequences of hyperinflation. However, to the individual, effects of inflation are less immediately clear when the rate of change of prices increases from 1.3 per cent to 5 per cent over a period of four years, as was experienced in the United States during 1964-68. Indeed, sometimes there is even confusion as to just what the term inflation means. A careful distinction must be made between changes in relative prices of assets and changes in the same direction of prices of all assets except money.1 Changes in relative prices play an important role in a dynamic growing economy. In a market-directed economy such as ours, changes in relative prices of goods and services and classes of factors of production are the mechanism by which resources are directed to produce the real goods and services that maximize the satisfaction of individuals in the economy. Individuals purchase real goods and services be cause the consumption of these items yields satisfac tion (or as economists would say, utility) to the in dividual purchaser. Exactly defined, inflation refers to a situation where an individual can no longer pur chase as large an amount of utility for a given money outlay. Because a satisfactory means has not been de veloped to quantify the utility that individuals receive from consuming goods and services, a less exact definition of inflation must be used. The term inflation is applied operationally to a situation where the exchange value of the medium of exchange (money), in terms of real goods and services, is decreasing. W e attempt to measure whether the general level of prices has increased, or whether there has only been a change in relative prices, by the use of a price index. Changes in the price index reflect changes in the total cost of a rep resentative market basket of goods. For example, if a price index rises from 100 to 105 over a period of time, w e say that the exchange value of money in terms of this representative market basket of goods is 4.8 per cent less. The purpose of this article is to examine the effects of inflation on individuals in their separate roles as income earners and holders of financial and real as]An individual’s holdings of assets, the current dollar value of which measures his nonhuman wealth, may be divided into two broad classes — real assets and financial assets. Real assets are items which yield a direct flow of consumption or production services to the asset holder. Financial assets are items that represent a claim on real assets or other financial assets. Page 2.5 NOVEMBER 1 9 6 9 F E D E R A L R E S E R V E B A N K O F ST. LO U I S sets. The analysis is limited to the eight-year period 1960 through 1968. No attempt is made to discuss long-run trends. For comparison purposes, the period from 1960-68 is divided into two four-year periods: 1960-64 when overall prices remained relatively sta ble, and 1964-68 when the rate of increase of prices accelerated. The analysis is limited to the effects of inflation on individuals in their separate roles on the average. Any one individual is not exclusively an income earner, not just a homeowner, nor just a holder of financial assets. Quite likely, he is all three. W e can judge whether a particular individual “benefited” or “lost” in a given period of time only by examining his total balance sheet. During the latter comparision period some individuals experienced greater increases in real income flows and in the real value of the stock of assets they held than during the 1960-64 period. Others fared worse with respect to these items than during the earlier comparison period.2 An Overall Look at the Period 1960 Through 1968 A t the aggregate level, both of the comparison pe riods show remarkable economic expansion. Over the first four-year period, real GNP increased by $93.4 billion, a 19 per cent increase. During the next four years, real GNP grew by an additional $126.5 billion, up 22 percent.3 Per capita real GNP also rose mark edly, by 12 per cent from 1960 through 1964, and then by 16 per cent from 1964-68. The two periods were dissimilar in at least two important aspects: prices and unemployment. The first period, 1960 through 1964, was characterized by a period of prevailing price stability: the consumer price index rose at an average rate of 1.2 per cent, wholesale prices showed almost no change, and the broader index, the GNP deflator, rose at an average annual rate of only 1.3 per cent. The second fouryear period was characterized by an accelerating price level. From an increase of only 1.3 per cent in 1964, the consumer price index increased at an aver age rate of 1.7 per cent in 1965, increased to a 2.9 per cent rate for 1966, slowed in the mini-recession of the 2Since the household sector is a net monetary creditor (its monetary assets exceed its monetary liabilities), this sector loses real wealth in periods of inflation to net debtor sectors such as the government sector and business sector. 3The 22 per cent increase in real GNP in the latter four years is an even more remarkable rise when one considers that this increase was achieved starting in 1965 from a much higher level of resource utilization than prevailed in 1960-61. 26 Digitized forPage FRASER Consumer Price Index R atio S cale R a tio S cale 1957-59=100 1961 1957-59=100 1962 1963 1964 1965 1966 1967 1968 1969 Source: U.S. D epartm ent o f Labor Percentages are annual rates o f change com puted from a nnual a verages o f m on thly data. Latest d a ta p lotted: September first part of 1967, and then again began its upward movement. Over the last half of 1967, the consumer price index rose at a 3.8 per cent annual rate and then accelerated to a 4.7 per cent rate for 1968. The first of our four-year periods was characterized by unemployment above 5 per cent. In 1962 the un employment rate was 5.5 per cent, a sharp drop from 6.7 per cent in the previous year. Over the next three years the unemployment rate remained at ap proximately the 1962 level. The second period, 1964 through 1968, is characterized by another sharp break in the per cent of the labor force unemployed. In 1965 the unemployment rate fell to 4.5 per cent and in 1966 reached an average of 3.8 per cent, a low level believed almost unattainable in a growing eco nomy in the early 1960’s. U n e m p lo y m e n t Rate 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 Let us now turn to the consideration of each of several different groups and see how, on the average, each of these groups actually fared in our two com parison periods. How Do We Judge Whether Individuals Have Benefited or Lost? If over a period of time an individual’s ability to command real output increases, we say he has “benefited.” If over a period of time an individual’s ability to command goods and services decreases, we w ill say he has “lost.” Real versus Nominal Benefits and Losses An individual’s ability to command goods and services depends upon his ability to command money balances. Over time, however, the exchange ratio be tween a given amount of goods and services and a given amount of money balances may change. A change in the amount of money balances a person can command is referred to as a change in his nominal command over goods and services. A change in the amount of goods and services a given amount of money balances commands is called a change in the individual’s real command over goods and services. A careful distinction must be made between nominal and real gains and losses. For example, if a person’s holdings of money balances rises from $100 to $200, his nominal gain is a doubling of his money balances. If, however, over the same period of time, prices of all goods and services double, then his real gain is zero. The confusion of nominal gains with real gains is called a “money illusion” by economists. Economists attempt to strip away the veil of money by adjusting nominal changes with a price index. Two Measures Two closely related measures of an individual’s command over real output are used to decide whether he has benefited or lost during a period of time. The first measure is income, which is defined as the flow of money payments an individual receives over a period of time. A person’s income is one major de terminant of the amount of goods and services he can command over time. NOVEMBER 1 9 6 9 Assets and liabilities are divided into two major classes — real and monetary. Monetary assets refer to assets exchangeable only for a fixed amount of dollars. Real liabilities are obligations to deliver a real asset whose exchange value in terms of money may vary. The individual’s balance sheet appears as: Assets Liabilities Monetary Assets Real Assets Monetary Liabilities Real Liabilities Net W ealth If the dollar value of the items on the left-hand side of the ledger equals the dollar value of the items, excluding net wealth, on the right-hand side of the ledger, then net wealth equals zero. To the extent that the dollar value of assets exceeds the dollar value of liabilities, the net wealth of the economic unit is greater than zero. If the net wealth of an economic unit expressed in current dollars rises between two periods in time, then the economic unit’s available command over money balances has risen. However, we would say the economic unit benefited only if this greater potential command over money balances represents command over a larger set of real goods and services. Income Flows Most individuals, when looking back over two pe riods of time and attempting to judge whether they fared better in the first period or in the second period, consider not only changes in their holdings of assets, but also consider how their flow of nominal income changed. To many individuals this last consideration is the more important of the two. When asked in which of two periods he benefited more, an individual’s answer will probably depend upon the answers to these two questions: (1 ) In which period did his income increase the most? (2) In which period did changes in his income allow him to command the greatest increase in flows of real goods and services? The second measure used is net wealth. An eco nomic unit’s net wealth is defined as: Net W ealth = Assets — Liabilities.4 Although questions (1 ) and (2) are related, they are not identical. Referring to the discussion of nominal versus real changes w e note that the period which shows the greatest increase in nominal terms is not always necessarily the same period that reflects the greatest increase in real terms. 4The difference between assets and liabilities may also be represented by the terms net worth or equity. In the 1960 through 1964 period, employment of production and non-supervisory workers in nonag- Page 27 F E D E R A L R E S E R V E B A N K O F ST. LO U I S NOVE MB ER 1 9 6 9 N o m in a l a n d Real Gross A v e ra g e W e e k ly Earnings o f P rod uctio n o r N o n s u p e rv is o ry W o rk e rs „ .. D o lla r s on P riva te N o n a g ric u ltu ra l P a yro lls , „ 5 r ---------- --------------------------------- ' Ann“ ° " >° '° D o lla r s -----------------— -----------------------115 To gain some insight into the relative income per formance of different categories of workers, let us examine selected working groups as presented in data of the Bureau of Labor Statistics. Skilled Compared to Unskilled Workers 7 5 '------------------------------------------------------------------ -------------------------------------------1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 75 (J.E a rn ing s in c u rre n t d o lla rs . 12 N o m in a l e a rn in g s d e fla te d b y th e c on sum e r p ric e in d ex 1957-59=100. As illustrated in Table I, both skilled and unskilled workers experienced a more rapid rise in nominal wages in the most recent period than in the previous four-year period. However, changes in real wages show quite a different picture. Since 1964, as the rate of increase of prices rose year after year, the per centage increase in real wages of both skilled and unskilled workers was only about one-half as large as in the 1960-64 period of relatively stable prices. Neither skilled nor unskilled workers benefited more in inflation than in the comparison period of price stability. S o u rc e : U.S. D e p a rtm e n t o f La bo r ricultural establishments expanded by 5.4 per cent, and gross average weekly earnings rose by 13.2 per cent in nominal terms and 8 per cent in real terms. Over the next four years employment in this area accelerated, increasing by 14.2 per cent. Average weekly earnings grew by 18 per cent. However, con sidering the rapid rise of prices, the real gain in purchasing power was about 5.2 per cent, only about two-thirds as great as that experienced during the previous four years of general price stability. Comparatively, unskilled workers benefited least from the recent period of inflation. In 1964 their flow of real wages permitted them to command 8.7 per cent more real goods and services than in 1960. In 1968, after four years of inflation, real wages of un skilled workers had risen only 4.1 per cent. Union Workers Table II presents average hourly wage rates for classes of union workers in selected trades. Table III shows the percentage changes in nominal and real EARN IN G S OF SKILLED A N D UNSKILLED WORKERS (1961 = 100) Skilled8 (M aintenance) Year Nom inal Index Average Hourly Earnings Per Cent Change ( 1960 Average Hourly Earnings 1960-64 \ 1964-68 ) 96.5 Unskilledb (P lant) Real Index Per Cent Change ( N om inal Index Average Hourly Earnings 1 9 6 0-64 \ 1964-68 ) 9 7 .57 Per Cent Change Real Index Average Hourly Earnings / 1 9 6 0-64 \ \ 1964-68 ) 96.50 1 9 6 0 -6 4 \ 1964-68 ) ( 9 7 .5 7 1961 100.00c 100.00 100.00d 100.00 1962 103.10 101.88 103.20 101.98 1963 105.90 1964 108.80 1965 111.40 105.59 113.20 107.30 1966 115.50 106.45 116.80 107.65 1967 120.30 107.80 121.80 1968 126.90 103.42 1 2.7% 16.6 104.92 109.11 106.60 7.5% 4.0 110.00 128.40 104.10 14.0% 106.08 8.7% 109.14 16.7 110.40 a Includes carp enters; electricians; m achinists; m echanics; m echanics autom otive; p a in ters; pip efitters; and tool and die makers. ^Includes janitors, porters and clea n ers; and laborers, material handling. cThe dollar figure fo r 1961 is approxim ately $2.89. d The dollar figure fo r 1961 is approxim ately $1.93. S ourer: U. S. D epartm ent o f Labor, Bureau o f Labor Statistics, H andbook o f L abor Statistic«, 1969, Table 94. pp . 200-201. Digitized forPage FRASER 28 Per Cent Change 4.1 NOVEMBER 1 9 6 9 F E D E R A L R E S E R V E B A N K O F ST. LO U I S Table II AVERAGE U N IO N W A G E SCALES® FOR SELECTED TRADES (N O M IN A L HOURLY W A G E S ) Printing B uilding Year Local Transit Local Trucking Journey man Helpers & Laborers Book & Job News papers Drivers Helpers 1960 $3.86 $2.88 $3.08 $3.48 $2.68 $2.38 $2 .37 1961 1962 1963 1964 4.02 4.15 4.31 4.4 6 3.0 6 3.15 3.26 3.4 0 3.18 3.24 3.3 7 3.4 7 3.58 3.66 3.75 3.84 2.78 2.89 3.02 3.14 2.48 2.55 2.68 2.79 2.46 2.55 2.65 2.76 1965 1966 1967 1968 4.6 4 4.83 5.09 5.43 3.54 3.67 3.83 4.05 3.58 3.69 3.81 4.0 0 3.9 4 4 .0 7 4.2 7 4.4 7 3.2 6 3.3 9 3.5 9 3.78 2.90 3.0 0 3.21 3.36 2.88 3.00 3.22 3.4 4 a The scales represent the m inimum wage rates (exclu din g holiday and vacation p aym ents). S ou rce : U. S. Departm ent o f Labor, Bureau of Labor Statistics, H andbook o f L a bor Statistics, 1969, Table 83, p. 170. wages for each of these classes of union workers in the two four-year comparison periods. These tables illustrate that many of the union groups covered ex perienced more substantial percentage increases in nominal wages over the 1964-68 period than during the 1960-64 period. White Collar Workers Table IV shows that a broad class of workers in white collar jobs experienced a more rapid rise in nominal wages in the latter period than in the 1960-64 period. Nevertheless, as was the case with most of the union groups surveyed, white collar workers re ceived a substantially smaller percentage increase in real wages in the more recent four-year period. In the 1964-68 period real wages of white collar workers included in Table IV rose only 3.5 per cent, less than one-half the increase in the 1960-64 period of widespread price stability. However, looking at changes in real wages in Table III, it appears that most union groups received smaller percentage increases in real wages in the recent pe riod of rapidly rising prices than in the 1960-64 period. The increases in the payments received for productive services by union workers in local truck ing, building, and printing trades in the 1964-68 pe riod represented substantially smaller percentage increases in their command over real goods and serv ices than what they experienced in the previous fouryear period. Only local transit workers, of the groups considered, received the same percentage rise in real wages in both periods. Professional Workers A third category of workers is labeled professional workers. Examining Table V w e see that, unlike the skilled or unskilled categories, or union and white collar categories, the selected groups of professional workers in Table V received substantially larger in- Table III C H AN G ES IN U N IO N W A G E S a FOR SELECTED TRADES Building Journeyman 1960-64 1964-68 Printing Helpers and Laborers Book 8< Job N om inal Real N om inal Real 1 5.5% 21.7% 10.4% 8.5% 18.1% 19.1% 12.5% 6.2% Nom inal 1 2.7% 15.3% Newspapers Real Nom inal 7.5% 2.7% 10.3% 16.4% Local Trucking Drivers 19 60-6 4 1964-68 Real 5.2% 3.8% Local Transit Helpers Nom inal Real N om inal Real Nom inal Real 17.2% 20.4% 11.9% 7.0% 17.2% 20.4% 11.8% 7.5% 16.5% 24.6% 11.0% 11.4% “Real wages are calculated by deflating nominal wages with the consum er price index, using 19fi0 — 100. Page 29 NOVEMBER 1 9 6 9 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 Table IV EAR N IN G S OF W HITE COLLAR WORKERS8 (1961 = Year 100 ) N om inal Index Real Index Per Cent Change / 1 9 6 0 -6 4 \ y 1964-68 ) Average W eekly Earnings 1960 96 .80 97.88 1961 1 0 0 .0 0 1’ 100.00 102.08 1962 103.30 1963 106.20 1964 109.20 Per Cent Change / 1 9 6 0 -6 4 \ ^ 1964-68 / Average W eekly Earnings rapid rate. For example, real wages of engineers rose 5.4 per cent in the three-year period 1964-67, slightly less than the 5.6 per cent increase in real wages dur ing the previous three years. From 1967 to 1968 average salaries of engineers in nominal terms rose by 7.6 per cent. This sharp rise in nominal terms offset the continued upward movement in prices, and as a result, their real wages showed an increase of 8.9 per cent from 1964-68. 103.71 105.30 12.8% 1965 1 12.30 106.45 1966 1 15.90 106.82 1967 120.90 1968 126.80 108.33 16.1 Wealthholders 7.6% 109.03 3.5 aO ffice clerical category which includes a broad range o f w hite collar workers. Included are typists, stenographers, different classes o f clerks, keypunch operators, bookkeeping and office boys and girls. bThe dollar figure fo r 1961 is approxim ately $77.10. S ou rce: U. S. Departm ent o f Labor, Bureau o f Labor Statistics, Handbook o f L abor Statistics, 1969, Table 94, p . 198. creases in both nominal and real wages in the three year period 1964-67 of rapidly rising prices than in the 1961-64 period of price stability. Most of these groups made marked gains in real terms in 1968 when prices were rising at their most An individual has available to him a wide array of real and financial assets in which he can hold his wealth. He may acquire real assets such as land, houses, or other real goods, or he may acquire one of the many different types of financial assets. Financial Assets Financial assets may be divided into two broad classes, money and other financial assets. The basis for distinction used here is that money is the financial asset which yields no nominal return to the holder. Other financial assets yield, or are expected by wealthholders to yield, a nominal or money rate of return. Table V EARN IN G S OF SELECTED CLASSES OF PROFESSIONAL WORKERS Accountants Year Average Annua] Salary Attorneys Realb Nom inal Average Annual Salary Per Cent Change $6 ,324 1964 6,8 40 1967 7,820 1968 8,277 Average A nnual Salary 4.3% 7 ,1 1 7 7.9 $8 ,136 8,532 4.9% 10,293 $6,684 1964 7,3 20 Average Annual Salary 1967 1968 8,432 8,931 Per Cent Change ( 7,0 59 7,5 56 22.0 7,6 79 Average Annual Salary Realb Per Cent Change 8.8 8,004 9,078 9,771 ( 9.5% 22.1 7,718 1 9 6 1 -6 4 \ 1964-6 8 ) 5.6% 8,134 8,402 “ Figures fo r 1960 n ot available. Figures based on the “ II ra tin g” in H andbook o f L abor Statistics. DReal wages are calculated by deflating normal wages with the consum er price index, 1961 — 100. S ou rce: U . S. D epartm ent o f Labor, Bureau o f Labor Statistics, H andbook o f L abor Statistics, 1969, Table 89, p . 191. Page 30 Per Cent Change $7 ,308 $ 7,308 5.6% Average A nnual Salary / 1 9 6 1 -6 4 \ \ 1964-6 8 ) 1 9 6 1 -6 4 \ 1964-68 ) $6,684 9.5% 7.6 Engineers / 1 9 6 1 -6 4 \ \ 1964-6 8 ) 1961 1.1% 8,8 50 20.6 N om inal Real1’ Per Cent Change 8,228 8,658 Chemists Average Annual Salary Per Cent Change / 1 9 6 1 -6 4 \ \ 1 9 6 4 -6 8 ) 9,662 7 ,0 0 7 N om inal Year ) $ 8,136 6 ,5 9 6 21.0 Average Annual Salary Per Cent Change /1 9 6 1 - 6 4 \ 1 96 4 -6 8 / 1 9 6 1 -6 4 \ \ 1 9 6 4 -6 8 / $6 ,324 8.2% Realb Nom ina! Per Cent Change ( 1961-64 \ \ 1964-68 ) 1961 (1961-68)' 8.9 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 N OV E MB E R 1 9 6 9 For an analysis of the effects of inflation on the holders of financial assets we shall distinguish effects on the wealthholder’s nominal wealth from effects on his real wealth; and the impact of inflation on his flow of returns in nominal and real terms. For purpose of analysis seven widely held financial assets were selected: (1) savings and loan shares (2) mutual savings bank deposits (3) time deposits at commercial banks (4) corporate bonds (5) U. S. Government bonds (6) municipal bonds (7) common stock Several major differences exist between groups of these assets. The first three, savings and loan shares, mutual savings bank deposits, and time deposits at commercial banks, represent legal claims to fixed amounts of money. In most cases, this claim may be exercised on demand by the holder of the claim or after only a short period of time. The next three items, corporate, U.S. Government, and municipal bonds, represent rights to a fixed amount of money only at maturity, usually much longer into the future. From the time they are issued until maturity, their magnitude of exchange value in terms of money depends upon the valuation which market partici pants place on the future flow of money payments they offer. Our last financial asset, common stocks, does not represent a claim to any fixed money pay ment, either currently or in the future. Table VII PERCENTAGE INCREASES IN YIELDS O N FIXED DOLLAR VALUE F IN A N C IA L ASSETS N om inal Percentage Increase 1960 to 1964 Savings and Loan Shares Savings Deposits at M utual Savings Banks 8.0% 1964 to 1968 1 2.0% Real Percentage Increase 1960 to 1964 1964 to 1968 3.1% - 0 .2 % 17.0 18.5 11.6 5.6 Time and Savings Deposits at Commercial Banks 33.6 31.2 27.5 17.0 S ou rce: U. S. Savings and Loan League Savings and Loan Factbook, 1969, Table 6, p. 17. year period the consumer price index rose 4.8 per cent. As a result of the rise in the price level, the command over real goods and services represented by the $1,000 in fixed dollar financial assets was re duced to $954.20 in 1964. During the second com parison period the holder of this type of asset suffered a much greater real loss. As the price level rose 12 per cent from 1964 through 1968, the asset holder found that his initial $1,000 in 1960 dollars repre sented only about $850.34 in real purchasing power in 1968. From the flow side, holders of this type of asset experienced substantial increases in nominal payments in both periods. In the first comparison period the nominal return on all three fixed dollar value finan Fixed D ollar Value Financial Assets —To give some cial assets rose more rapidly than the price level, and initial comparisons, let us assume that in mid-1960 hence the real rate of return received by holders of an individual bought $1,000 of one of the fixed all three of these assets rose. For example, as shown dollar value assets. At the end of each period his in Table VI, the holder of a $1,000 in savings and wealth would still be $1,000 in nominal terms. How loan shares received a flow of $41.70 in dividends in ever, the asset holder lost real command over goods nominal terms in 1964 compared to $38.60 in 1960. and services in both periods. During the first fourAs illustrated in Table VII this represented an 8 per cent increase in nominal terms and a 3.1 per cent increase in real terms. Table VI During the second comparison YIELDS O N FIXED DOLLAR VALUE F IN A N C IA L ASSETS® period, although the nominal yields 1960 1964 1968 on these assets rose sharply, much Nom inal Real Real Real Nom inal N om inal of the increase in nominal terms was Yield Yield Yield Yield Yield Yield taken up in price increases.3 As Savings and lo a n Shares $ 3 8.60 $38.60 $ 4 1.70 $39.79 $ 4 6 .7 0 $39.71 Table VII shows, in nominal terms Savings Deposits at the yields on savings and loan M utual Savings Banks 3 4 .70 34.70 38.74 40 .60 48 .10 40 .90 shares rose by an additional 12 per Time and Savings Deposits a t Commercial Banks 25.60 25.60 3 4 .20 32.63 44 .90 38.18 “These yields are calculated using the exam ple o f $1000 in assets. S ou rce: U . S. Savings and Loan League Savings and Loan Factbook, 1989, Table 6, p. 17. 5The much smaller increases in real yields on fixed dollar financial assets in the sec ond period to a significant degree reflects the fact that their maximum nominal yields are fixed by regulatory authorities. NOVEMBER 1 9 6 9 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 cent, raising nominal yields on $1,000 of shares to $46.70. However, in real terms the yield on savings and loan shares fell slightly from $39.79 in 1964 to $39.71 in 1968. The nominal yield on time deposits at commercial banks rose more rapidly than the price level from 1964 through 1968, and holders of this financial asset experienced a 17 per cent increase in real yields on these assets. However, these results were much less than the 27.5 per cent increase in real yields holders of debt obligations of commercial banks obtained dur ing the four years of prevailing price stability.6 Financial Assets with Money Values Fixed Only at Maturity — In this section we examine how holders of three widely held classes of bonds fared during our two comparison periods. The dollar value of these assets is fixed only at maturity. As the market places different valuations on the flow of money payments offered by these assets, the money exchange value of these assets varies over the life of the bond. This revaluation of money flows offered by bonds is represented by inverse movements of their prices and yields. As the public places a lower valuation on the stream of payments which the claim represents, the market price of the bond falls and its effective yield rises. The relationship between prices and ef fective yields can be seen in Table VIII. In the 1964-68 period of accelerating price infla tion, bond holders suffered much larger capital losses in nominal and real terms. From 1964 through 1968 the price of U.S. Government bonds fell 14.4 per cent, municipal prices fell 16.1 per cent, and corporate bond prices declined 19.7 per cent. Not only did bond holders experience a drastic decline in nominal values of their assets, but as the rise in the price index ac celerated, they found the exchange value in real goods of their declining nominal values fell about 2% times as rapidly as during the 1960-64 period of relatively stable prices. Common Stocks — Of the broad classes of financial assets being considered, only common stocks in creased both the nominal and real wealth of the asset holder in both periods. In the 1964-68 period the Standard and Poor’s 500 Stock Index (a broad measure of the magnitude of the money exchange value of common stocks) rose 21.3 per cent. How ever, this was only 60 per cent as great a percentage increase as the 36 per cent rise in the Stock Index in the 1960-64 period.7 A much larger rise in the consumer price index in the more recent period meant that holders of com mon stocks not only fared worse in the second period in nominal terms, but they also fared considerably worse in real terms. It is interesting to note that in the second period the real wealth of holders of com mon stock rose only 8.1 per cent, compared to 29.8 per cent in the period 1960-64. The holder of U.S. Government bonds over the 1960-64 period would have suffered about a 2 per cent decline in the nominal value of his bonds. Hold 7The market prices of common stocks are heavily influenced by ers of corporate bonds would have found in 1964 that the level of business activity. 1960 was a trough in business activity. To remove some of the cyclical influence on our the price of these bonds was approximately the same analysis, an average of 1959, 1960, and 1961 was used for as in 1960. Holders of municipal bonds would have 1960. been able to sell these assets in 1964 at a 7.3 per cent higher price than in 1960. In real Table V III terms, the rise of the consumer AVERAGE A N N U A L YIELDS A N D PRICES O F SELECTED BO NDS price index reduced the real Corporate U nited States State and Local value of any given money claim. Aaa Government M unicipal Bonds Bonds Year Bonds Hence, in real terms holders of Average Average U.S. Government and corporate Average Price Yield Price Yield Yield Price8 bonds were worse off in 1964 $ 1 0 3 .9 0 4.41% $ 9 4 .6 0 3.69% 4.01% $8 6.22 1960 than in 1960; only holders of 4.35 9 5 .2 0 3.60 107.80 3.90 87.55 1961 municipal bonds gained in real 96 .20 86.94 3.30 112.10 4.33 3.95 1962 terms. 111.30 4 .2 6 96 .80 3.28 4.00 86.31 1963 6The above discussion should not lead to a confusion between levels and percentage changes in levels. The person who held $100 0 of savings and loan shares over the 1960-68 period would have received, on the average, a much larger flow of returns in nom inal and real terms than if he had held a time deposit account at a commercial bank. Page 32 1964 4.15 84.46 3.28 111.50 4.4 0 9 5 .1 0 1965 4.21 83.76 3.34 110.60 4.49 9 3 .9 0 1966 4.66 78.63 3.90 102.60 5.1 3 86.10 1967 4.85 76 .55 3.9 9 100.50 5.51 81.80 1968 5.25 72.33 4.48 9 3 .5 0 6.18 76 .40 3 The dollar price per $100 bond. Source: Federal Reserve Bulletin 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 In the first comparison period the dividend pay ments received by stockholders rose from $13.4 billion to $17.8 billion, a 33 per cent increase. In the 1964-68 period the flow of dividend payments rose to $24.6 billion, a 38 per cent increase. In nominal terms, the percentage increases in the flow of payments to stockholders was somewhat greater in the latter pe riod. However, the increased flow of dividends in the latter period represented a somewhat smaller increase in real purchasing power, 23 per cent, compared to 27 per cent in the earlier period. The assertion that common stocks are a better hedge against inflation than the other types of finan cial assets we considered is borne out by the evi dence. However, although stockholders fared better in inflation relative to holders of the other financial assets we discussed, holders of common stock bene fited much more in the earlier period of extensive price stability than in the latter period of rapidly rising prices. It seems difficult to support an assertion that stockholders benefited more in inflation when the percentage increase in their real wealth was much greater under four years of generally stable prices than under four years of rapidly rising prices. Real Assets Sometimes the general assertion is made that infla tion destroys the incentive to save by wiping out the real value of wealth accumulated by past acts of saving. This statement fails to take into account that saving may occur by additions to wealth in the form of real assets as w ell as financial assets. As the magni tude of the exchange value of money in terms of real goods and services falls, holders of real assets benefit from inflation in nominal terms and suffer no loss in real terms. To examine the effects of inflation on holders of real wealth w e have selected two real assets, land and houses. Land —As shown by Table IX, the average mar ket value of an acre of farmland increased in both comparison periods. In the 1964-68 period of inflation, the average market value of an acre of farmland rose by 29 per cent, compared to an 18.5 per cent in crease over the previous four years. When we adjust both nominal increases for price level changes, the spread between the two periods is reduced. However, in contrast to all the financial assets discussed, the real wealth of landowners showed a somewhat larger percentage increase in the period of rapidly rising prices, 15 per cent, com pared to 13 per cent in our comparison period of widespread price stability. NOVEMBER 1 9 6 9 Table IX AVERAGE MARKET VALUE OF FARM LAND Year Average Price Per Acre 1960 $1 16.48 1961 118.22 1962 124.19 1963 129.59 1964 138.00 1965 145.75 1966 157.00 1967 167.00 1968 178.00 S o u rce: U. S. Departm ent o f A griculture, Agricultural Finance R eview , V olum e X X I X . Supplem ent, A p ril 1969, Table 39, p. 65. Houses — Over the eight-year period we are con sidering, the price of houses rose sharply. From $13,800 in 1960, the average construction cost of new homes increased to $15,550 in 1964, and then rose to $18,675 in 1968. To analyze the effects of inflation on home owners the net wealth measure is used. W e must take into account at least two other fac tors in order to use the net wealth measure. The ma jority of the funds used for the purchase of most homes are borrowed. W hen a person borrows to pur chase a house, he generally agrees to pay a fixed monthly payment to the mortgage lender for a period of years; he agrees to give up a fixed amount of nom inal purchasing power each month until the mort gage is paid off. The homeowner decides to forego present and future command over real output for present and future command over the flow of serv ices from a specific real asset, a house. The second factor we must consider is that over time a real asset is used up, or depreciates. In gen eral, as a house is used over a period of time, the flow of services it can provide decreases, hence the market value placed on the flow of services offered by the house also declines. To take into account the two factors, mortgage buying and depreciation, three assumptions are made in the following example: (1) A house is bought in 1960 on a 25-year mort gage with 20 per cent of the purchase price as a down payment. (2) The mortgage is repaid in equal monthly install ments over the 25-year period. (3) The house depreciates at the same rate as the mortgage is paid off. Page 33 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 Using assumptions (1) through (3) in our example, we have: Purchase price of a house in 1960 =, $13,800 Downpayment = 20 per cent of $13,800 = $ 2,760 Mortgage = 80 per cent of $13,800 = $11,040 Yearly mortgage repayments = $11,040 h- 25 years = $441.60 Yearly depreciation of house = $441.60 1960 BALANCE SHEET Assets Liabilities $13,800 = real assets = market value of house $11,040 = monetary liability = mortgage $ 2,760 = net wealth A t the end of 1964 the balance sheet has changed. Over the four years 1961 through 1964, the home owner repays $1,776.40 of his mortgage (4 x $441.60), and his house depreciates by this amount. The mar ket value of his house in 1964 equals the construction cost of a new house ($15,500) less the depreciation of his home ($1,766.40). 1964 BALANCE SHEET Assets Liabilities $13,784.60 — market value of house $9,273.60 = outstanding mortgage $4,511.00 = net wealth A t the end of 1968, the average homeowner has repaid an additional $1,766.40 on his mortgage, and his house has depreciated by this amount. The market value of his house in 1968 is $18,675 (the construc tion cost of a new house) minus $3,532.80 (eight years of depreciation). 1968 BALANCE SHEET Assets Liabilities $15,142.20 = market value of house $7,507.20 = outstanding mortgage $7,635.00 = net wealth Examining the balance sheets we see that over each comparison period the homeowner experienced a substantial rise in nominal net wealth. In the first period nominal net wealth rose by 63.4 per cent, and in the second period by 69.3 per cent. Table X illustrates the contribution of factors act ing to change the net wealth position of the home owner. The use of the house over time decreases the market value of its flow of services and hence opera tes to decrease the net wealth of the homeowner (decreases the dollar value of his real asset). The repayment of the mortgage decreases the homeown er’s monetary liabilities and hence operates to in Digitized for Page FRASER 34 NOVEMBER 1 9 6 9 crease his net wealth. Assuming no change in the general price of houses there would be no change in his net wealth. However, due to the marked increase in the gen eral construction cost of houses over each period, the market value of the real asset held by the home owner increased. In the 1960-64 period the rise in the construction cost of houses resulted in an increase of $1,751 in the homeowners nominal net wealth. At the end of the first four-year period, the homeowner in our example could have sold his house for approxi mately what he paid for the house in 1960. In 1968, because the construction cost of houses advanced more rapidly in the 1964-68 period than over the pre vious four years, he could have sold his house at a price almost 10 per cent above what he paid in 1960. W hen we adjust the changes in nominal net wealth for changes in prices over our two periods, we also record substantial gains for the homeowner. In real terms, the homeowner’s net wealth rose by $1,544 in the four-year period 1960-64 and then showed an even greater increase in the 1964-68 period, increas ing by $2,188. Again to clarify the meaning of the term “benefited,” we mean that the homeowner’s real command over goods and services increased. For example, between 1960 and 1964 his real net wealth was augmented by $1,544. If he had sold his house at its market value of $13,784.60 in 1964 and repaid his outstanding mort gage, his command over real goods and services would have been $1,544 greater than if he sold his house and repaid his mortgage in 1960. Over the more recent four-year period, the homeowner in our example benefited even more in the sense that his real command over goods and services rose by $2,188. For the potential home buyer, the purchase price of a house is only one consideration. Another major consideration is the cost of borrowing funds to make Table X CO M P O N E N TS OF C H AN G ES IN H O M E O W N E R ’S N O M IN A L NET W EALTH Per ods Changes in N om inal N et W ealth 1960-6 4 1964-6 8 Due to depreciation o f house -$ 1 ,7 6 6 .4 0 -$ 1 ,7 6 6 .4 0 Due to general rise in prices of houses + $ 1 ,7 5 1 .0 0 + $ 3 ,1 2 4 .0 0 Due to repaym ent o f mortgage + $ 1 ,7 6 6 .4 0 + $ 1 ,7 6 6 .4 0 Total + $ 1 ,7 5 1 .0 0 + $ 3 ,1 2 4 .0 0 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 purchase. During the first of our comparison periods the contract rate on conventional first mort gages remained fairly stable at around 5.75 to 5.85 per oent. On balance, it would not have cost our average individual more in nominal terms to obtain funds to finance the purchase of a home in 1964 than earlier in the first period. The second period shows quite a different picture. Over this period the cost of financing a home in creased along with other market interest rates. Whereas the average cost of financing a new home by a conventional first mortgage was 5.78 per cent in 1964, this rate rose to an average of 6.83 per cent for 1968 and was at 7.08 per cent in the second half of 1968. On balance, if instead of buying a home in 1964, an individual had delayed buying a home un til the second half of 1968, not only would the aver age construction cost of the home have been about 20 per cent higher, but the financing costs would have risen by 22.5 per cent. In real terms, if a person had financed a home on a conventional first mortgage during the period 1961 through 1965, he would have gained in real terms during the following three years. However, after 1965 the current cost of mortgage financing rose faster than the consumer price index. If a person delayed buying a house in 1965, when mortgage rates were about 5.74 per cent, until 1966 when mortgage rates rose to an average of 6.14 per cent, he would not have experienced a reduction of his real financing costs to the 1965 level until late 1968. Suppose our average individual became an average homeowner in 1960. Using our example, we assume he purchased a $13,800 house in 1960 with a 20 per cent downpayment and the balance financed over 25 years with a financing cost of 5.75 per cent. On this basis, his monthly payments would be $69.48. Our individual has decided to give up $69.48 a month in nominal command over goods and seivices in ex change for the flow of services from a house. By 1964 the homeowner would still be giving up $69.48 a month in nominal purchasing power. How ever, since the consumer price index rose by 4.8 per cent over these four years, he would be giving up slightly less in real purchasing power each month, about $66.30 in real purchasing power. In contrast, the average homeowner found that, in the four-year period since 1964, the real purchas ing power he was giving up each month decreased about 2 V2 times as rapidly as over the previous four NOVEMBER 1 9 6 9 years of general price stability. In our illustration the real purchasing power of $69.48 in 1960 dollars fell to $59.08 in 1968. Retired Persons One of the common maxims in most discussions of the effects of inflation is that people on fixed in comes, especially retired persons, “lose” during peri ods of inflation. Since inflation is a situation where the magnitude of the exchange value of money in terms of real assets declines, individuals whose flow of money payments remains fixed find their income commands a smaller flow of real goods and services in an inflationary situation. The truth of the assertion that retired persons lose during inflation depends upon the assumption that their income payments re main “fixed” and that net nominal increases in the value of their other assets do not offset their loss of real income. One form of income flow to retired persons is social security benefits. As illustrated by Table XI the aver age amount of monthly benefits received by retired persons did not stay fixed over the 1960 through 1968 period. Beflecting several increases in social security bene fits, average monthly payments to retired workers rose from $74.04 in 1960 to $98.86 in 1968. In the second half of the period, when prices began to rise rapidly, benefits rose by 27.4 per cent compared to an increase of only 4.8 per cent over the 1960 through 1964 period. Considering the rise in the consumer price index over each period, the purchasing power in real terms of Government transfer payments to retired persons Table XI AVERAGE A M O U N T OF M O N T H LY O A S D H I3 BENEFITS T O RETIRED WORKERS" Year A verage M o nthly Benefits 1960 $ 7 4.04 1961 75.65 1962 76.19 1963 76.88 1964 7 7 .57 1965 83.92 1966 84.35 1967 8 5 .37 1968 98 .86 a01d-age, Survivors, Disability, and Health Insurance P rogram . ^Persons aged 65 and over (and aged 62-64, begin n in g in 1956 fo r women and 1961 fo r m e n ). A verage benefits in current-paym ent status at end o f period. S o u rce: U. S. Departm ent o f Health, Education and W elfare, Social S ecurity Bulletin, Septem ber 1969, p. 42. Page 35 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 was about the same in 1964 as in 1960, then increased by approximately 13.6 per cent from 1964 through 1968.8 For many retired persons, social security payments comprise only a portion of the income they depend upon after retirement. A portion of their income derives from returns from financial assets they have purchased over a period of years. Such assets include the value of life insurance, savings and loan shares, bank deposits, bonds and common stock. In the sec tion on the effects of inflation on holders of financial assets, we saw that holders of these assets did not fare as well in the recent 1964-68 period of rapid price inflation as in the 1960-64 period of much more stable prices. To the extent that retired persons held real assets, they were made no worse off by inflation and, de pending on the asset, may have benefited. Many people, when they reach retirement age, have paid off the mortgage on their home. As the price of homes has risen, the magnitude of the exchange value of this asset in terms of other real assets has increased. Alternatively, since they are no longer net debtors with regard to their house, they do not benefit as much from inflation as those individuals who are still net debtors on their homes. On balance, rapidly rising social security benefits offset part of the effects of inflation on retired persons. However, unless they were solely dependent upon such payments for retirement income, it does not ap pear that retired persons made any real gains in the period from 1964 through 1968. In fact, compared to the previous four years of price stability, in many cases retired persons may have suffered a decline in their ability to command real output. 8The change in real purchasing power for retired persons may have been less than indicated by deflating by the consumer price index. The prices of many services, which might be expected to weigh more heavily in retired person’s budgets, such as medical care, physicians’ fees, property taxes, and public transportation, increased over 2 0 per cent during this period compared to a 12 per cent rise in the total index. NOVEMBER 1 9 6 9 Conclusions We have concluded that, of all the classes of workers considered, only certain groups of profes sional workers, such as accountants, attorneys, en gineers, and chemists, could be said to have benefited more with respect to income flows in the 1964-68 period of inflation than in the previous period of general price stability. The broad classes of skilled workers, unskilled workers, union workers, and white collar workers that were examined all benefited less in the 1964-68 period than during the 1960-64 period. Individuals, on the average, in their separate roles as asset holders, benefited more in the 1964-68 period than in the 1960-64 period only in their roles as owners of real assets — land and houses. W ith respect to all financial assets considered except common stocks, individual asset holders lost during the latter period. Holders of common stocks, although they did not lose in the 1964-68 period, benefited substantially less than during the 1960-64 comparison period. Finally, to the extent that retired persons are solely dependent upon social security benefits, the sharp upward revisions of these transfer payments resulted in a rise in retired persons’ real command over goods and services. However, to the extent that retired per sons also depended upon insurance payments, fixed dollar value securities, and bonds, they lost real pur chasing power in the 1964-68 period. Any particular individual can only determine in which of our comparison periods he fared better by examining his total balance sheets for both compari son periods. This article has attempted to give a gen eral framework in which the individual can complete this analysis. Rather than making broad assertions about the nebulous “evils of inflation,” a definition of benefit and loss has been presented for the reader. If the individual is alerted to the dangers of confus ing nominal benefits with real benefits, he has a means of judging the effects of the recent inflation on his own ability to command real output. This article is available as Reprint No. 48