The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
interest rate* and inflation Federal lax and spending reform The discount rate— will it float? Interest rates and inflation 3 Serious em pirical research on the relation b etw een interest rates and inflation resu m ed in the late 1960s after a lapse o f nearly fo u r decades. Federal tax and spending reform CONTENTS In an effort to deal with w id ely felt e co n o m ic pain, the Congress has co n sid e re d several e co n o m ic reform proposals to limit a n d /o r re d u ce the role o f the federal govern m en t in the econ om y. The discount rate— will it float? E C O N O M IC PERSPECTIVES Since the Federal R eserve a d o p te d its n ew reserve s-o rien ted operating p ro ce d u re , the instability o f the sprea d b etw een the federal funds rate and the d iscou nt rate has led many o b servers to suggest floating the discou n t rate. May/June 1981, Volum e V , Issue 3 Economic Perspectives is published bimonthly by the Research Department of the Federal Reserve Bank of Chicago. The publication is produced under the direction of Harvey Rosenblum, Vice President, and is edited by Larry R. Mote, Assistant Vice President, with the assistance of Sandra Cowen (editorial), Roger Thryselius (artwork and graphics), and Nancy Ahlstrom (typesetting). The views expressed in Economic Perspectives are the authors' and do not necessarily reflect the views of the management of the Federal Reserve Bank of Chicago or the Federal Reserve System. Single-copy subscriptions of Economic Perspectives are available free of charge. Please send requests for single- and multiple-copy subscriptions, back issues, and address changes to Public Information Center, Federal Reserve Bank of Chicago, P.O. Box 834, Chicago, Illinois 60690, or telephone (312) 322-5112. Articles may be reprinted provided source is credited and Public Information Center is provided with a copy of the published material. Controlled circulation postage paid at Chicago, Illinois. 13 20 Interest rates and inflation John H. W o o d * If a trader has lent w h e a t. . . at interest, then fo r e very g ur o f wheat he shall take 100 qa as interest. If he has lent silver at in terest, then fo r each sh e ke l o f silver he shall take a sixth part o f a sh ekel, plus six grains, as interest. The Ham m urabi C od e, No. 90, circa 2080 B.C. The relation betw een interest rates and infla tion has attracted much attention in recent years. Serious em pirical research on this sub ject has resum ed after a lapse of nearly four decades, from the early 1930s to the late 1960s. The point of d eparture of this w ork has been Irving Fisher's classic study, The Theory o f In terest [5], published in 1930. Fisherfound interest rates during the period 1890-1927 to respond slow ly and incom pletely to varia tions in inflatio n . The most com m on in te r pretation of these results is that inflationary exp ectations, w h ich in flu en ce current inter est rates, respond slowly to observations of past inflation. Th e results of most recent studies have been consistent with Fisher's. But Eugene Fama [4] has presented results that contradict those of earlier w riters. M ore im portant, Fama's w o rk suggests that interest rates im m ediately and com pletely reflect inflationary expectations. This article com pares the results of Fisher and Fama and places these results in historical p erspective. The small d ifferences between Fisher and Fama appear to be due less to increases in financial m arket efficien cy, im provem ents in statistical m ethods, and better data (as suggested by Fama) than to the spe *Written in collaboration with Scott Ulman. We are grateful for helpful comments by George Kaufman, Larry Mote, and Harvey Rosenblum. Federal Reserve Bank o f Chicago cial, relatively tranquil period chosen by Fama for his em pirical w o rk. The years 1953-71 are unique in A m eri can history for th eir record of stable prices. Charts 1 and 2 m ake clear that there is not, among periods of sim ilar length, a close com petitor after 1894 with 1953-71 for the moststable-price-period prize. Variations in prices since 1971 have been more like those observed by Fisher than Fama. T h e re fo re , an understanding of the co nn ectio n s betw een interest rates and infla tion in the w orld in w hich we live, w hile not neglecting Fama’s contributions, requires that we pay special attention to Fisher's. Real and nominal interest rates The distinction betw een interest rates in terms of m oney (e.g ., silver) and interest rates in terms of goods (e.g ., wheat) has long been recognized. It is useful to think of the form er as nom inal rates of interest, R, and of the latter as real rates of interest, r. In the United States n o m in al rates of in te rest m easure returns in dollars. These are the rates of inter est reported in newspapers and advertised by depository institutions. Real rates of interest, on the other hand, m easure the productivity of investm ent goods (i.e ., the rate of trans form ation of current goods into future goods) and the tim e preferences of households (i.e ., the allocation of consum ption between cu r rent and future goods). D ifferen ces between real and nom inal interest rates ought to be due to expected rates of inflation, i.e ., to expected rates of change in the value of m oney relative to goods. If the expected rate of inflation is denoted p, the equilibrium relation betw een R and r may be expressed as: (1) (1 + R) = (1 + r)(1 + p). 3 Suppose, for exam p le, that the expected real return to an investm ent in a m achine is r = 3 percent per annum . That is, the m achine is expected to produce a net output each year worth 3 percent of the value of the m achine. Further suppose that, due to inflation, the prices of the m achine and its output are expected to rise 5 percent during the next year. That is, exp ected inflation is p = 5 p er cent. The expected nom inal (dollar) rate of return to this real investm ent is therefore (1.03)(1.05) - 1 = 8.15 percent. An investor’s choice between the m achine (or shares in the m achine) and, say, a 52-week Treasury bill depends on the bill's yield , or rate of return, R. If R exceeds 8.15 p ercen t, investors w ill be attracted to the b ill, bidding its rate down until R equals the expected return on alterna tive investm ents of sim ilar risk, including real investm ents.1 If R is less than 8.15 p ercen t, 1 This argument follows Fisher and Fama in abstract ing from complications associated with risk and taxes. investors w ill avoid the bill until its rate becomes com petitive with other investments. This line of argum ent suggests that, given r = .03 and p = .05, the e q u ilib riu m value of R is .0815. An alternative but equivalent way of looking at the real rate of interest, the gain in purchasing pow er from lending m oney, is presented in box 1. U nder the dubious assum ption that the Babylonian b ureaucracy fixed interest on sil ver and wheat at rates com patible with market forces, we can determ ine the expected rate of change of prices in that co untry in 2080 B.C . Since 1 gur = 300 qa of w heat and 1 shekel = 180 grains of silve r, r = 100/300 = 33 1/3 p er cent and R = 36/180 = 20 percent. Using eq u a tion (1), we can surm ise that H am m urabi's subjects expected an annual rate of deflation of 10 p ercen t.2 2Unlike paper money, there is a real return to silver in productive activities. For the sake of completeness, therefore, r should be interpreted in our Babylonian example as the difference between the real returns to wheat and silver. Chart 1. Inflation and real and nominal interest rates 4 Econom ic Perspectives Equation (1) is often expressed as the fo l lowing linear approxim ation: (2) R = r + p. This approxim ation is fairly close for small p and r. In the two exam ples given above, the approxim ate equilibrium nom inal rates are 8 and -23 1/3 percent, com pared with the exact e q u ilib riu m values of 8.15 and -20 percent. A lthough the connections between in terest and inflation have been understood for thousands of years, they have never been d is cussed m ore than during the past 250 years. This is due to the increased use of paper m oney and th erefo re to the increased vo latil ity of inflation. W illiam Douglass wrote in 1740 that large em issions of paper money “ rise the interest to m ake good the sinking principal [2, p. 335].” For exam p le, the Rhode Island issue of 1739, w hich caused a d ep recia tion of paper m oney by 7 percent, required an increase in the rate of interest from 6 to 13 Federal Reserve Bank o f Chicago p ercent. Speaking in the House of Com m ons in 1811, during a period of w artim e inflation, H enry Thornton pointed out that if a man borrow ed m oney at a nom inal interest rate of 5 percent and repaid the loan after a period of 2 percent or 3 percent inflation, “ he would find that he had borrow ed at 2 or 3 percent, and not at 5 p ercent as he appeared to do [11, p. 336].” Fisher’s results The first extensive statistical studies of the relations betw een real and nominal inter est rates and inflation w ere carried out by Irving Fisher. His results w ere stated in their most com plete form in The Theory o f Interest [5, pp. 399-451]. Em pirical tests of equation (1) are difficu lt for a variety of reasons. Most im portant, none of the three variables is directly observable. All are expectations of future events: p is the expected rate of infla tion, r is the expected real return to produc tive activities, and R is the expected nominal 5 return to investm ents in debt to be repaid in dollars. Most researchers have sim plified the problem by assuming the real return to be constant and by choosing high-grade short term securities w ith the same m aturity as the period of observation to ensure that observed nom inal yields w ere virtually the same as expected nom inal returns. For exam p le, Fisher’s most thorough tests used quarterly data and four- to six-m onth prim e co m m er cial paper. This is not a perfect solution to the problem , but later w riters have been able to use Treasury bills. This leaves expected inflation. It is co n ceivable that the m arket’s expectation of inflation during the period beginning on date t is a weighted average of past inflation rates. Then using the linear approxim ation (2), our model may be w ritten : w h ere R t is the yield on date t on a security m aturing in one p erio d ; the w ’s are weights that indicate the im portance of past rates of inflation (pt_-j, pt_2, . . .) in determ ining exp ectatio n s of in fla tio n for the com ing period, pt; p nt is the weighted average of these past rates of inflatio n , with n being the length of the lag, i.e ., the num ber of past rates in clu d ed ; and et is an unobserved random error term with a mean of zero. There is no time subscript on the real rate of interest because it is assumed to be constant. Fisher did not estimate the w ’s. Regres sions in those days w ere too expensive for such a procedure. Rather, he tried different com binations of the w ’s and n and ran co rre lations of the resulting p nt’s and Rt. Specifi cally, his weighting schem e was: (3) (4) n Pt-1 + (n - 1)pt_2 + • • • + Pt-n Rt = r + w-jpt_-j + w 2pt_2 + . . . + w npt_n P n t " -------------------— - 1)------------------ 1------------------ ' n + (n -- + . . . + ~ • + et = r + Pnt + et Chart 2. Inflation and real and nominal interest rates 6 Econom ic Perspectives Fisher's co rrelatio n s, using quarterly ob servations on the com m ercial paper rate and the w h olesale p rice in d e x, are shown in chart 3. For interest rates observed between 1890 and 1914, the highest correlation was achieved when rates of inflation lagged 30 quarters w ere used. The co rrelation s fell as the lag was lengthened . But betw een 1915 and 1927, the co rrela tio n betw een interest rates and past inflation was co ntinuously im proved as the lag was len g thened — up to the m axim um lag of 120 quarters tried by Fisher. He perform ed sim ilar tests with annual data for the United States and G reat Britain and the results w ere the sam e: assum ing that (a) the real rate of interest is constant, (b) expectations of future inflation are determ ined by past inflation in the m anner show n in equation (4), and (c) the approxim ate eq u ilib riu m relation (2) is satis fie d , the cu rren t interest rate is apparently determ ined by expectations based on past inflation ary exp erien ces as distant as 30 years ago. Fisher w ro te: “ It seems fantastic, at first glance, to ascribe to events w hich occurred last century any in flu en ce affecting the rate of interest today. And yet that is what the co rre lations with distributed effects of p show [5, p. 428]." 3 Recent studies of interest rates and inflation Studies published in 1969 and 1970 by W illiam Gibson [6], Thom as Sargent [10], and W illiam Yohe and Denis Karnosky [12] co r roborated Fisher's results. Based on models sim ilar to equation (3) and data taken from periods both before and after W orld W ar II, these authors, like Fisher, found that interest rates responded slow ly and incom pletely to inflation and that long distributed lags of past inflation rates w ere useful in explaining in terest rates. Sargent described his results as 3Fisher’s notation has been altered to conform to that used in this paper. For a discussion of Fisher’s specu lations about the reasons for these results, see Rutledge [9, pp. 19-21], percent 30 25 20 15 10 lo + 5 5 10 15 20 Federal Reserve Bank of Chicago 7 Box 1 The real rate of interest Suppose you invest an am ount of m oney, V, at the rate of interest R p revailing on dateO. If Pq is the average p rice of goods in your “ norm al co nsum p tio n b u n d le " on that date, you have re lin q u is h e d p u rc h a sin g p o w e r o ve r V / P q goods. A teenager existing e xclusively on Big M acs priced at $1.25 gives up 80 units of curren t consum ption w hen he deposits $100 in the local S&L. He does this, p resum ab ly, in ord er to be able to consum e an even greater num ber of Big M acs in the fu tu re . At a rate of interest of 5 p ercent, his investm ent w ill grow to $105 by next year. This w ill enab le him to consum e 105/P-j units if P-j is the p rice of Big M acs next year. The nom inal (m oney) return on his invest ment is the num ber of dollars gained as a p ro portion of the n um b er of dollars invested (relin q uish ed ) and in this exam p le is 5 percent. The real (ham burger) return is the num ber of Big M acs gained as a p rop ortion of the num ber of Big M acs invested (re lin q u ish ed ). If prices are stable, i.e ., P-j = Pq = $1.25, $105 w ill purchase 84 follow s: “ The re su ltsa re sim ila rto Fisher'sin a couple of ways. Not only do they confirm the im portance of the distributed lag price exp ec tations variable, but they im ply a very long lag in the process of exp ectatio n s form ation [ 10] . " Eugene Fama [4] follow ed a different method and obtained d ifferent results. He used Treasury bill yields and described his estimates as tests of the efficien cy of the Treasury bill m arket.4 In efficient markets observed prices and interest rates correctly reflect all of the inform ation available to market participants. This means that (a) ob served interest rates co rrectly reflect the m arket’s inflationary or deflationary exp ecta tions and (b) those expectations are u n b i ased; i.e ., expectations, on average, are co r 4More generally, Fama described his estimates as tests of the joint hypotheses of (i) market efficiency and (ii) the constancy of the real rate. This article is concerned only with the first hypothesis. For discussions of the latter hypothesis, see Carlson [1], Fama [4], and Nelson and Schwert [8], 8 Big M acs next year and the real rate of return is 4/80 = 5 percent. But if a 20 percent per annum inflation has o ccurred so that P^ is $1.50, $105 w ill be w orth only 105/1.50 = 70 Big Macs and the real return w ill be -10/80 = -12.5 percent. He has gained m oney but lost goods. In general, this real rate of interest, r, may be expressed as fo llo w s: Y(1 + R ) P-i Y_ P„ Letting p = (P-| - Pq)/P q denote the rate of inf lation, the above equation may be rew ritten as: (1 + r) = I ; 0 + P) or (1 + R) = (1 + r)(1 + p), w hich is identical to equation (1). rect. If m arket exp ectatio n s w e re biased and/or w ere not reflected in interest rates, there would exist opportunities to make sub stantial sums either by borrow ing (if interest rates are "to o lo w ") in order to buy goods or by selling goods in order to lend (if interest rates are “ too h ig h "). To the statistician, runs of high or low observed real rates are in d i cated by high autocorrelations, meaning that real rates are highly correlated with their own past values. The concept of efficien t m arkets, w hich is in turn closely related to rational expectations, is discussed in box 2. A u to correlation is discussed in box 3. In general, observed real rates are d iffe r ent from the expected real rates, r, discussed above in connection with equations (1) and (2). This is true for several reasons. Probably the most im portant cause of d ifferences be tween expected and realized (observed) real rates is the inability to forecast inflation, p, accurately. For exam p le, the nom inal interest rate at tim e t might be R = .07 and be based on Econom ic Perspectives Chart 3. Fisher’s correlations between the commercial paper rate and distributed lags of past inflation rates Length of lag in quarters an expected inflation rate of p = .03. This means an expected real rate of approxim ately r = .07 - .03 = .04. But suppose inflation turns out to be 13 p ercent instead of 3 percent. The realized real rate of return to the nominal investm ent— the actual gain or loss in p ur chasing p ow er— is approxim ately .07 - .13 = -.06. Runs of high or low observed real rates w ould suggest that the m arket was systemati cally overp red icting or underpredicting infla tio n . U n d er these circu m stan ces nom inal rates of interest w ould not, on average, be accurate predictors of inflation. Fama's tests of efficien cy w ere lim ited to observations betw een January 1953 and July 1971. He began his sam ple period with 1953 because before 1953 the Federal Reserve interfered with m arket efficien cy by support ing governm ent security prices. He excluded observations after July 1971 because queues, side paym ents, and increases in the various form s of non p rice rationing caused by price controls prevented stated prices from accu r ately m easuring the costs of acquiring goods. Fam a’s tests took two form s. First, he ran auto co rrelations on realized real rates of Federal Reserve Bank o f Chicago Box 2 Rational expectations and efficient markets The co ncept of rational expectations e x tends the notion of rational behavior to the prediction of e co n o m ic events. It assumes that people m ake use of available inform ation in a consistent m anner. In p articu lar, predictions of eco n o m ic events are based upon the p u b lic’s view s of how the econ o m y w orks. O f course, those view s must largely be based upon what has been learned from e xp e rie n ce , including observations of past events. Presented this w ay, nothing could be less o bjectio nable than rational expectations. There h ard ly, thus far, seem s room for debate. The debate enters w ith the frequent assum ptions by advocates of rational expectations that (i) peo ple have the “ c o rre ct” view of the econom ic stru ctu re, i.e ., that they “ kno w e ve ryth in g ,” and (ii) new inform ation is instantly and fully reflected in econom ic decisions. Com bined, these assum ptions im ply no transactions costs and free info rm atio n about all kinds of processes, sim ple and co m p lex. The result is efficient markets, in w hich prices and interest rates alw ays fu lly and co rre ctly reflect all publicly available inform ation. By “ know ing e ve ryth in g ,” i.e ., having the “ c o rre ct” view of the econo m ic structure, advo cates of efficient m arkets do not mean that people predict the future with certainty. It is assum ed, rather, that people know the econ o m y’s statistical tend en cies. They cannot predict exactly what the rate of inflatio n , for exam ple, w ill be. But th eir pred ictio n s w ill be correct on average. Predictions w ill not be biased in the sense of being consistently high or low. interest. Even in efficient m arkets, p red ic tions w ill almost always be w rong. But they will not be biased in the sense of being co n sistently too high or too low. An extrem e exam ple of an inefficien t market is the period from May 1942 to July 1947, when the Federal Reserve pegged the discount rate on threemonth bills at 0.375 percent. During this period the consum er and w holesale price indices rose at average annual rates of 6.2 percent and 7.9 percent, respectively. M onthly changes in the CPI and W P I, at annual rates, 9 Box 3 Autocorrelation Suppose xt is the sales of a com p any in period t. The au to co rrelation s of this tim e series are the co rrelations of x w ith its own past values. For exam ple, the first-order au to co rrelatio n , p , of x is the co rre la tio n betw een x{ and xt If the co m p any’s sales increase by some constant am ount c so that xt = x{ ^ + c , then p = 1. In this case, xt and x{ -j are p erfectly c o rre la te d ; one co nsequence of this relation is that x{ may be predicted with certainty if we kno w x , . , . O n the other hand, if x varies in a co m p letely random fashion such that a kno w ledge of x ^ conveys no inform ation about xt, then p =0. Tim e series that are subject to sm ooth cyclical fluctuations have highly positive auto co rrelations. Those w hich rapidly change directio n have highly negative au to co rrelatio n s. M ost A m erican eco nom ic tim e series, inclu d in g interest rates and inflation, belong to the form er category. exceeded 0.375 percent 40 and 42 tim es, respectively, during these 62 m onths. That is, realized real rates of interest w ere predom i nantly negative and, in ad d ition, w ere highly autocorrelated. Using the CPI and Treasury bill rates, Fama fo u n d , as exp ected , very differen t re sults for the 1953-71 period. The first-order autocorrelation coefficients of one-, two-, and three-m onth real returns on one-, two-, and three-m onth bills w ere .09, .15, and .00, respectively. Because these autocorrelations are “ close to z e ro ,” Fama interpreted his results as consistent with the hypothesis that the Treasury bill m arket is efficient. The second series of tests perform ed by Fama involved estimates of regression equa tions sim ilar to the fo llo w in g :5 (5) pt = -r + Rt + et. 5Fama actually used the rate of change in the pur chasing power of money, A t, which is approximately equal to -pt> in his regressions. But his results are fully consistent with regressions of the form shown in equa tion (5), which has been used for ease of comparison with the work of Fisher and others. 10 This is sim ilar to Fish er’s equation (3) except that (i) the actual rate of inflation, pt, is used instead of a weighted average of past inflation rates, p nt, and (ii) the positions of inflation and Rt in the equation have been reversed. The nom inal rate of interest is now the explanatory variable instead of the d e pendent variable. As in equation (3), et is a random error term with mean zero. Equation (5) asserts that, given the assumed constant real rate, r, the m arket's expectation of the rate of inflation during the period beginning on date t, pt, is fu lly reflected in the nom inal rate of interest, Rt, observed on that date. Fama reported regressions for the period 1953-71 on one- to three-m onth bill yields. In every case the co efficien t of Rt did not differ significantly from unity, as suggested by equa tion (5). His correlation co efficients (between pt and Rt) w ere statistically significant and ranged from .54 to .70. Fama also interpreted these results as consistent w ith the hypothesis that the Treasury bill market is efficient. Earlier w riters had observed that interest rates responded slow ly to inflatio n , i.e ., that Rt did not fu lly reflect expected pt . This sug gests, if sim ilar relations prevailed in the Treasury bill m arket during 1953-71, that the results obtained by means of equation (5) might be im proved by adding past inflation rates as explanatory variables. Fama estimated regressions sim ilar to (6), w hich he rep re sented as tests of this hypothesis: (6) pt = -r + R t + w-j pM + et. But the addition of pt_-j failed to im prove the correlations significantly and the esti mates of the co efficien t (w-j) of pt_-| w ere statistically insignificant, leading Fama to claim these regressions as fu rth er eviden ce of the efficien cy of the Treasury bill m arket. He dis missed the results of Fisher and others that suggested market inefficien cy as probably having been caused by poor price data. Back to Fisher Fama's paper elicited critical com m ents by w riters w ho com bined his approach with Econom ic Perspectives Fish er’s. Douglas Joines [7] and Charles N el son and W illiam Schw ert [8] pointed out that regression (6) could not fairly be com pared with Fisher’s results, w hich depended on many— not on e— past rates of inflation. Using Fam a’s data, they estimated regressions of the form : Table 1 Inflation and real and nominal rates of interest A* • Four- to six-month prime commercial paper and the wholesale price index Standard deviations (7) p, = r + b R , + w 1p ,_1 + w 2pt _2 + . • . + w nPt-n + et and found that past rates of inflation co n tained significant inform ation about future inflation in addition to that reflected in Rt> Fu rth erm o re, their estimates of the co effi cient (b) of Rt w ere significantly different from unity. A p p aren tly, during 1953-71 as during 1890-1927, interest rates responded slowly and incom pletely to inflation. Variations in inflation and interest rates during the Fisher and Fama periods of obser vation are shown in chart 1. The chart shows the rate of change in the wholesale price index and real and nom inal returns on fourto six-m onth prim e com m ercial paper begin ning in June 1894, w hen four- to six-m onth prim e com m ercial paper rates w ere first re ported on a regular basis. O bservations are at five-m onth intervals expressed in p ercent ages at annual rates. O ne of the most striking characteristics of this chart is the stability of inflation in Fam a’s p eriod, 1953-71, com pared with the very large fluctuations before 1953 and after 1971. A p p aren tly, Fama’s sam ple is not typical of A m erican exp erien ce. This co n clusion is supported by chart 2, w hich shows the rate of change in the consum er price index and real and nom inal returns on threemonth Treasury b ills, beginning with the reg ular reporting of bill yields in February 1930. O bservations are at three-m onth intervals expressed in percentages at annual rates. Now let’s apply Fam a’s test of market efficien cy to the data shown in the charts. The co lum ns headed p in table 1 list the firstorder auto co rrelations of observed real rates during our co m plete sam ple periods and selected su bperiods. W e have separated the Federal Reserve Bank o f Chicago Time period P P R r 6/94 - 4/80 .317 13.79 2.65 16.31 - 6/29 - 10/52 - 2/71 - 4/80 .264 .522 .226 -.078 17.82 13.45 2.10 7.32 1.09 1.10 1.89 3.41 22.98 12.28 2.23 5.88 7/71 - 1/74 6/74 - 4/80 .038 -.555 8.83 6.78 2.42 3.69 7.30 5.10 6/94 11/29 3/53 7/71 B Three-month Treasury bills and the consumer price index Standard deviations P P R r 2/30- 5/80 .569 6.66 2.80 5.66 2/30 - 11/52 2/53 - 5/71 8/71 - 5/80 .570 .178 .270 8.78 2.04 3.80 .70 1.68 2.50 8.07 1.40 1.97 8/71 - 2/74 5/74- 5/80 .259 .279 3.88 3.59 1.95 2.54 2.47 1.73 Time period p is the first-order autocorrelation of r. Standard deviations are in percentages at annual rates. post-Fama period into observations falling w ithin the period of price controls, August 1971 to April 1974, and those occurring after price controls w ere abandoned. The p rin ci pal results shown in the table may besum m arized as fo llow s: • The first-o rd er auto co rrelatio n s are high and significant for both of the full sam ple periods (1894-1980 in table 1A and 1930-80 in table 1B) and for the subperiods dominated by the Great Depression (with runs of high real rates) and W orld W ar II (with runs of negative real rates). 11 • The 1974-80 subperiod yielded a posi tive autocorrelation in part B of th etab le but a negative auto co rrelation in part A. This may be due to the freq u en t acceleratio ns and decelerations of inflation during this period. Perhaps the five-m onth observational period used in A allow s tim e fo r reversals in p, and therefo re a negative serial correlation of r, not accounted for by the three-m onth period used in B. W e should not put much stock in these results, h o w e ver, due to the short period of observation. • Perhaps surp risin g ly, in view of the d if ferent results that have been associated with these subperiods, the first-order a uto co rrela tion of r during 1894-1929, w hich is largely co incid ent with the period of Fisher's analy sis, is only slightly greater than that for 195371, the period of Fam a’s analysis.6 M arkets did rem arkably w ell in forecasting the large fluctuations in inflation during the earlier period. • N otice, ho w ever, that the standard 6Note that the first-order autocorrelation for 1953-71 (.178) reported in table 1B is higher than the 0.00 reported by Fama. This may be due to the use in the present study of yields that are monthly averages of daily figures. Fama used yields on the first business day of each month. Since price indices are compiled from data that are collected throughout the month, and are therefore monthly aver ages of a sort, it was felt that comparability between interest rates and inflation required the former also to be expressed in terms of monthly averages. deviation of R during 1894-1929 was much less than the standard deviation of p. This co n trasts with the 1953-71 p erio d, in w hich the volatility of R was only slightly less than that of p. But this was due less to the greater respon siveness of R during the later period than to the sm aller volatility of p. Conclusions It should be stressed that much work rem ains to be done in this area and that none of the results presented in this paper—whether Fisher's, Fama's, or those in table 1— have ju s tified any firm co nclusio ns about the p ro cesses that determ ine observed relations b e tween inflation and real and nom inal rates of interest. About all that can be said on the basis of the available data is that, during most p eri ods (excepting especially 1929-52), the T rea sury bill and com m ercial paper markets have not appeared to be highly in e fficien t. A u to correlations in real rates are not usually very high. Yet nom inal interest rates persistently fail to respond fu lly to inflation w hen in fla tion is vo latile. This was true before 1953 and after 1971. A p parently, like other reasonably effective but im perfect processes, the short term securities markets perform w ell if not asked to do too m uch. They can keep up with inflation if the pace is not too fast or too variable. References 1. John A. Carlson, “ Short-Term Interest Rates as Predictors of Inflation: Co m m ent,” Am erican Econom ic R eview , vol. 67 (June 1977), pp. 469-75. 2. W illiam Douglass, A D iscourse C on cern in g the C u rrencies of the British Plantations in Am erica (S. Kneeland and T. G reen, 1740). (Reprinted in 1897 for the Am erican Eco nomic Association by M acm illan, New York). 3. C h ilperic Fdwards, The Hamm urabi C o de (Watts and Co m pany, 1921). 4. Fugene F. Fama, “ Short-Term Interest Rates as Predictors of Inflation,” Am erican Econom ic R eview , vol. 65 (June 1975), pp. 269-82. 8. Charles R. Nelson and G. W illiam Schwert, “ Short-Term Interest Rates as Predictors of Inflation: On Testing the Hypothesis that the Real Rate of Interest is Constant,” Am erican Econom ic Review , vol. 67 (June 1977), pp. 478-86. 9. John Rutledge, A M onetarist M o d e l o f Inflationary Expecta tions (D .C. Heath, 1974). 10. Thomas J. Sargent, “ Com m odity Price Expectations and the Interest Rate,” Q uarterly Journal o f Econom ics, vol. 83 (February 1969), pp. 127-40. 6. W illiam F. G ibson, “ Price-Expectations Effects on Interest Rates,” Journal o f Finance, vol. 25 (M arch 1970), pp. 19-34. 11. Henry Thornton, “ Speech in the House of Com m ons on the Report of the Bullion Com m ittee,” May 7,1811. Reprinted in Appendix III to Thornton’s An Enquiry into the Nature and Effects o f the Paper Credit o f Great Britain. Reprinted in 1939 by George A llen and U n w in , Fondon, with an introduction by F.A. Hayek. 7. Douglas Joines, “ Short-Term Interest Rates as Predictors of Inflation: Co m m ent,” Am erican Econom ic Review , vol. 67 (June 1977), pp. 476-77. 12. W illiam P. Yohe and Denis S. Karnosky, “ Interest Rates and Price Level Changes,” Review , Federal Reserve Bank of St. Louis (Decem ber 1969), pp. 19-36. 5. Irving Fisher, The Theory o f Interest (M acm illan, 1930). 12 Econom ic Perspectives Federal tax and spending reform W. Step h en Sm ith The eco n o m ic legacy of the 1970s has been the co ntinuous upward spiral of inflation, un em p loym ent, and interest rates. The federal governm ent's inability todeal effectively with these problem s has placed eco nom ic reform at the top of the nation's agenda for the 1980s. In an effort to deal with these problem s, the Congress has considered a w ide variety of eco n o m ic reform proposals in recent years. Several of these proposals have a com m on th em e: the role of the federal governm ent in the eco nom y should be lim ited and/or re duced. Four of th em — three proposed co n stitutional am endm ents and one tax reform b ill— have received significant attention from prom inent politicians and the press: • The balanced budget am endm ent, which w ould req u ire that federal expenditures not exceed federal revenues. • The spending cap am endm ent, w hich would lim it federal expenditures to some specified portion of GN P. • Th e re ve n u e cap a m en d m en t, w h ich w ould lim it federal revenues to some specified portion of GN P. • The Kem p-Roth b ill, w hich would reduce personal incom e taxes 30 percent over the next three years. This article presents an overview of these proposals and discusses their im plications for the nation's eco nom ic future. The real sources of economic pain The first rum blings of the taxpayers' revolt w ere heard in the late 1960s, as the “ go-go" years drew to a close and, partly as a result of the d eficit fin an cin g of the Vietnam W ar, inflation began to heat up. The focus of atten Federal Reserve Bank of Chicago tion at the tim e was reform of the local prop erty tax, but few of the organized initiatives met with success. H o w ever, the severity of inflation in the late 1970s undoubtedly added strength to the psychology of the taxpayers' revolt m ovem ent, w hich drew widespread attention in 1978 with the passage in C alifo r nia of Proposition 13. As a result of the national attention that was focused on Propo sition 13 and its p roponents, a num ber of other states considered and im plem ented fis cal reform s.1 In less than a decade, the taxpayers' revolt m ovem ent has been transformed from a sm all, ineffective lobby to a dom inant force on the A m erican political scene. What factors w ere prim arily responsible for this change? Econom ist Lester Thurow has argued that popular support for p olicies that would bring about a dram atic shift in the distribution of eco nom ic resources arises only from intense econom ic p ain .2 An obvious source of such pain was the apparently d eclining standard of living in A m erica. Yet, in the six years (1972-78) of eco nom ic tu rb u len ce that gave rise to the w id e spread popularity of tax reform , real per cap ita disposable personal incom e rose just under 16 p ercen t, almost as much as during the “ gogo” years 1966-72. (See table 1.) To be sure, real hourly earnings in the private nonagricultural sector w ere slightly low er in 1978 than they had been in 1972, but the decline was m ore than offset by a sharp rise in the proportion of the total population that is ’At least 15 states have adopted fiscal limitations since the passage of Proposition 13. The recent recession, however, has apparently reversed this trend, as taxlimitation proposals were defeated in six states in 1980. For an argument that Proposition 13 was not the result of a basic shift in taxpayer attitudes, see James M. Buchanan, “The Potential for Taxpayer Revolt in American Demo cracy,” Social Science Q uarterly, vol. 59 (March 1979), p. 691. 2Lester Thurow, “The Real Sources of Economic Pain,” Wall Street Journal, July 6,1978. 13 Table 1 Growth in real per capita disposable income Year Real per capita disposable incom e Percent change (1972 dollars) 1966 1968 1970 1972 1974 1976 1978 # o CO 5 r- 3,290 3,493 3,668 3,880 4,050 4,216 4,487 4,567 6.2 5.0 5.8 4.4 4.1 6.4 1 .8 •Estimated. SOURCE: Econom ic Report o f the President, Jan uary 1981. em ployed. H ow ever, to many households the loss of leisure may have constituted a decline in their standard of living. In any case, the 16 percent rise in real per capita disposable incom e might conceal large disparities between groups w ithin the popu lation; it might reflect substantial real gains made by some w h ile many others suffered real econom ic losses. "N o t so ,” Thurow co n cludes. " In the six years from 1972 to 1978 there have been no significant shifts in the distribution of incom e. The gap between rich and poor, black and w h ite, male and fem ale has rem ained unch an g ed .” 3 Since relative incom es have not changed significantly, all groups have benefited from the real eco nom ic growth. O f course, it is still possible that there w ere large d ifferences between individuals within each of these broad groups. A nother freq u en tly m entioned source of eco nom ic pain is that the governm ent has taken an ever-increasing share of the average citiz e n ’s earnings. Again, the eco nom ic evi dence does not support the popular assum p tion. W h ile the ratio of total tax revenue at all levels of governm ent to personal incom e increased from 17.7 percent in 1950 to 30.1 percent in 1980, most of this increase took Jlbid. 14 place during the 1950s and 1960s.4 (See table 2.) G o vernm ent exp enditures exh ib it almost the same growth pattern. Table 3 shows that w h ile the ratio of total exp enditures at all levels of governm ent to G N P has risen from 23 percent in 1950 to 33 p ercent in 1980, virtu ally all of the increase took place during the 1950s and 1960s. Th ere has been little growth in the ratio of governm ent exp e n d i tures to GN P during the 1970s. Still another possibility has been sug gested by Lester T h u ro w . He argues that the primary source of pain is the so-called "m oney illu sio n ” created by the enorm ous gap be tween the growth of real and m oney incom es that has resulted from inflatio n . W h ile real incom e grew 16 p ercent from 1972 to 1978, m oney incom e grew 72 percent. People think what life w ould be like if their incom es had risen by 72 percent w ith no inflatio n . Some people may even co nvin ce them selves that their real standard of living has fa lle n .5 Most people fo rget, h o w ever, that in fla tion raises incom e as w ell as prices. Every price increase is a reduction in the real living standard of some purchaser of a good or ser vice , but it is also a real incom e increase for some p rovider of that good or service. M ore im portantly, most people suffering from the "m o n ey illu sio n ” do not realize that if there had been no inflation from 1972 to 1978, real incom es w ould have grown by 16 p ercen t, not 72 percent. It is not easy to identify clearcut reasons for the widespread perception of econom ic stress. It may have resulted in part from a 4Richard A. Musgrave has argued that this evidence countersthe widely held belief that inflation has resulted in an increasing tax burden. See “The Tax Revolt: Causes and Cure,” Social Science Q uarterly, vol. 59 (March 1979), p. 699. 5 This is clearly not the type of “ money illusion” that is so familiar in the literature of monetary economics. There “ money illusion” refers to the temporary failure of people to realize that their real incomes have not kept pace with their nominal wages. In Thurow’s use of the term it is precisely the public’s awareness, perhaps belated, that their real incomes lag behind their nominal incomes that is the source of pain. The illusion consists in their belief that somehow their real incomes could be made to rise as rapidly as their nominal incomes. Econom ic Perspectives Table 2 Tax revenue as a percentage of personal income (all levels of government) 1950 Incom e tax C o rp oratio n tax Payroll tax Property tax O ther Total 1960 1970 1980* 6.6 3.1 1.5 2.5 4.0 10.5 3.6 2.8 4.2 3.4 12.4 3.3 4.9 4.5 3.9 13.0 3.1 7.4 3.1 3.5 17.7 24.5 29.0 30.1 * Estimated. SOURCE: U.S. Bureau of the Census, Governm ent Finance, various years. Econom ic Report o f the President, January 1981. “ m oney illu sio n " that confuses nom inal and real m easures of eco n o m ic p erfo rm ance. N evertheless, d esp ite evidence that the gov ernm ent sector did not continue to grow in the 1970s at the rapid rates of the 1950s and 1960s, many people feel that reducing the role of governm ent in the econom y provides a prescription for relief from the econom ic problem s facing the nation in the 1980s. Balanced budget amendment O f the proposed reform s, the balanced budget am endm ent is probably the most popular with the general electorate. In a spring 1979 C BS-N ew York Tim es p oll, 73 perTable 3 Public expenditures as a percent of GNP 1950 Federal State and local Total 1960 1970 1980* 13.4 7.9 21.3 17.1 9.8 26.9 18.2 13.4 31.6 19.5 13.5 33.0 ♦Estimated. NOTE: Federal grants-in-aid to state and local gov ernments are included at the level of the recipient. SOURCE: Econom ic Report o f the President, Jan uary 1981. Federal Reserve Bank of Chicago cent of the respondents favored a constitu tional am endm ent to requ ire the Congress to balance the budget every year. Legislators in 30 of the necessary 34 states have approved resolutions asking the Congress to call a con stitutional convention to consider such an am endm ent. Th ree other states have adopted resolutions that urge the Congress to adopt a balanced budget am endm ent, but do not call for a co nven tio n . The convention m ovem ent, h o w ever, has met with significant opposition from several key political figures who fear a “ run aw ay" co nventio n that w ould attempt to adopt am endm ents on other issues such as busing and abortion. The balanced budget amendment is some thing of a m isnom er, because the proposal w ould bar the federal governm ent from in curring d eficits, but not from attaining sur pluses. O f co urse, surpluses have been few and far betw een in recent years. A more sub stantive shortcom ing of the am endm ent is that, even if it achieved the goal of elim inat ing deficits, it w ould not necessarily limit or reduce the role of the federal governm ent. The governm ent could co ntin u e to increase spending, in absolute term s and in relation to G N P, as long as it increased tax revenues to keep the budget balanced. Aside from its inappropriateness as a means to ach ieve the goals of some of its proponents, the balanced budget amendment might severely im pair the governm ent's abil ity to in flu en ce the econom y. Prior to the Depression the governm ent pursued the “ fis cally resp onsib le" policy of balancing the federal budget. This “ old-fashioned doctrine,” according to econom ist Robert J. G o rdon, “ did co nsid erab le harm to the econom y and has since been abandoned by all econom ists, monetarists and nonmonetarists a lik e ."6Why? D uring a recession G N P declines along with personal and corporate taxable incom es. If tax rates and governm ent expenditures re main constant, and the budget was in balance just before the recession, the budget w ill now 6Robert J. Gordon, M acroeconom ics (Little, Brown and Company, 1978), p. 480. 15 show a d eficit. To rebalance the budget, the federal governm ent must either raise tax rates or reduce governm ent expenditures, either of w hich w ill exacerbate both the recession and the deficit. Attem pting to balance the actual budget during a recession ignores the fact that w hile the budget affects the eco nom y, the eco n omy has a feedback effect on the budget. Although, in p rin cip le, the governm ent could stim ulate the econom y by raising both e x penditures and tax rates during a period of slack private d em and, thereby m aintaining the budget in balance, the adm inistrative and political d ifficu lties of d oin g so probably p re clude such an approach. The balanced budget am endm ent, in and of itself, is u n like ly to enhance the governm ent’s ability to control inflation. A l though governm ent deficits have some shortrun im pact on d em and, they are not in fla tionary in the long run unless they are fi nanced by increases in m onetary growth in an attempt to hold down interest rates. O ther things being eq u al, a governm ent deficit financed by the sale of Treasury securities, accom panied by rising interest rates and no increases in m oney su pply, w ould not add to inflationary pressures. In this context the bal anced budget am endm ent appears to be n eu tral in its expected impact on inflation. Even with such an am endm ent in place, the Congress and the President would have many ways of m aintaining expenditures. Offbudget outlays could be increased. The fed eral governm ent in fiscal 1980 allocated $12 billion to off-budget entities and $19.1 billion to governm ent-sponsored agencies, roughly 6 percent of the total budget. Loan guaran tees,an o ther method of avoid ing the budget ary process, am ounted to $284 b illio n in 1980. In sum , the available evid en ce suggests that a balanced budget am endm ent may create as many problem s as it solves. W h ile it might, in a nom inal sense, elim inate future federal deficits, it is not u neq uivocally clear that this is a desirable goal. Even if it w ere, there is no assurance the governm ent would not circum vent the intent of the proposal. 76 M o reo ve r, the am endm ent w ould take away an im portant recession-fighting tool of the federal government and, in fact, might deepen any future recessions. Spending cap amendment This proposed constitutional am endm ent w ould limit spending by the federal govern ment to a certain percentage of G N P. The specific percentage varies with alternative proposals, but virtually all have fallen into the 18 percent to 21 percent range. The spending cap am endm ent has been advocated most strongly by econom ist M ilton Friedm an and the National Tax Lim itation Co m m ittee. Their am endm ent includes provisions that would limit off-budget outlays, allow the spending limit to be exceeded in national em ergencies, and protect grants to state and local govern m ents. It w ould lim it the growth of the fed eral governm ent’s share of eco nom ic activity, but is silent on the question of deficits. By lim iting governm ent spending, the am endm ent w ould tend to w eaken the gov ernm ent’s recession-fighting capabilities. This effect w ould not be as serious as under the balanced budget am endm ent, h ow ever, since the governm ent w ould retain the authority to cut taxes during recessionary periods. In the long run the am endm ent might be helpful in fighting inflation caused by m onetization of deficits because it w ould hold the growth of governm ent expenditures below that of nom inal GNP. O ver the past decade federal exp en ditures grew at an annual rate of 10.5 p ercent, w ell above the 9.5 percent growth rate of GNP. The am endm ent has other disadvantages. First, since certain expenditures rise autom a tically (for exam ple, unem ploym ent insurance benefits during a recession), other exp en d i tures w ould presum ably have to be reduced. This w ould create a great deal of uncertainty with regard to the planning of certain exp en diture program s. Second, it w ould lim it the ability of the governm ent to target exp en d i tures during a recession toward certain hardhit or disadvantaged areas, industries, or indi- Econom ic Perspectives viduais since any increased expenditures in these areas w ould have to be offset by d e creases elsew here. Revenue cap amendment A third proposal, less frequently discuss ed , is the reven u e cap am endm ent, w hich w ould lim it fed eral revenues to a certain p er centage of G N P. This am endm ent w ould not elim inate deficits as it controls only govern m ent reven u es, not governm ent spending. Although it w ould w eaken the governm ent’s ability to fight recession to a lesser extent than the balanced budget am endm ent, the re ven u e cap am endm ent w ould som ewhat in hibit the governm ent's range of inflatio n fighting strategies. O n e co nsequence of our tax structure is that federal revenues tend to grow faster than the general econom y during in f la t io n a r y p e r io d s . T h e a m e n d m e n t w ould force the governm ent to cut tax rates to hold dow n tax revenues during inflatio n ary tim es. As a result, fiscal policy would be of little use in attenuating inflationary trends. Im plem entation of the proposal would be relatively easy, unless the cap were set at a m uch low er level than present tax co lle c tions. As noted e a rlie r, federal taxes did not rise dram atically during the 1970s; had the am endm ent been adopted in 1970, with the cap set at the then-p revailing level, it would not have been exceeded to date. W hile the econom y and federal spending grew at an nual rates of roughly 9.5 percent and 10.5 p ercen t, resp ectively, over the past decade, federal revenues grew at roughly 8.5 percent. Th u s, unless the cap w ere low ered to a pre1970 share of the G N P , the revenue cap am endm ent is not likely to reduce sig nifi cantly the g overnm ent’s share of econom ic activity. O n the o ther hand, it should prevent that share from growing significantly in the future. Kemp-Roth bill A fourth reform proposal, and the one w hich has received the most political atten Federal Reserve Rank of Chicago tio n , is the Kem p-Roth b ill, w hich calls for a 30 percent reduction in federal incom e taxes over the next three years. Proponents of the bill have argued that reducing personal in com e taxes w ill increase incentives to w ork and w ill expand the tax base so that, even at the low er rates, no tax revenues are lost. The theory behind this argum ent is sum m arized by the Laffer C u rv e , named for its originator, econom ist A rth ur Laffer (see box). Prior tax cuts. M any proponents of the bill have argued that prior tax cuts, particu larly the Kenn ed y tax cu t, provide em pirical confirm ation of the Laffer C u rve hypothesis. W a lte r H e lle r , th e ke y a rc h ite c t of the Kennedy tax cu t, has responded that the supply-siders’ argum ents are flaw ed. Taxes w ere cut by about $12 b illio n ($10 billion in d i vidual and $2 b illion corporate) in 1962-64; H eller notes that “ the record is crystal clear that it was its stim ulus to d em and . . . that powered the 1964-65 expansion and restored a good part of the initial revenue loss.’’7 U nem ploym ent was reduced from 5.6 p er cent in January 1964 to 4.5 percent in July 1965, and utilization rates in m anufacturing increased, draw ing on existing excess capa city. Since inflation rose only slightly over the same perio d, from 1.4 percent to 1.6 percent, most of the increase in demand was co n verted into m ore output, not higher prices. H o w ever, the prem ise that any change in eco nom ic activity after a tax cut is a result of the tax cut ignores the m ultiple causal rela tions in a co m plex eco nom y. O th er fiscal fac tors playing a critical role in the 1963-68 exp ansion, for exam p le, w ere the huge (over) stim ulus of Vietnam exp en d itu res, the four increases in payroll tax rates and base in those years, and the $6 billion of revenues from the 1966 Tax Act. M o reo ve r, m onetary policy also played some role in the expansion. After slow ing in 1962, m oney supply growth accel erated in 1963 and 1964. Sim ilar difficulties in isolating the effects of tax cuts from other influences plague the other historical exam7 Walter W. Heller, "The Kemp-Roth-Laffer Free Lunch,” Wall Street Journal, July 12,1978. 17 pies, the M ellon tax cuts in the 1920s and the West Germ an cuts in 1948, w hich the KempRoth proponents use to support their theory. E co n o m ic e v id e n ce . The econom ics pro fession has been studying questions related to the Laffer hypothesis for several years. There is little eco n om ic evid en ce, how ever, to support the co nclusion that current levels of tax rates create disincen tives to w o rk and save. Studies of w o rk er response to changes in take-hom e pay have yielded ambiguous results. Some people w ill w o rk harder if a tax cut or some other change makes each hour of w ork worth m o re; others choose to take additional tim e off and enjoy m ore leisure w hile earning the same incom e. After review ing the available evid ence, the Congressional Budget O ffice concluded that hours w orked w ould increase if after-tax real wages rose, largely because of the impact of m arried wom en entering the labor m arket. The net effect, h o w e ver, w ould be sm all— perhaps a 1 p ercent to 3 p ercent increase in the labor supply as a result of a 10 percent rise in disposable in co m e.8 This estimate falls short of the m inim um 10 percent increase in the labor supply w h ich w ould be necessary for the Kem p-Roth cuts to be self-financing .9 The effect of changes in the after-tax rate of return on savings is also an em p irical q ues tion, and available evid en ce is also am bigu ous.10*Some people w ill save m ore if they earn a higher rate of retu rn ; others w ill save less and m aintain a constant level of assets. M any econom ists have long accepted what has becom e know n as D enison's Law, that the saving rate is virtu ally constant and unaf fected by changes in the tax structure or the The Laffer Curve Laffer argues that taxes create a “ w ed g e” betw een salary and take-ho m e pay and be tw een pre-tax and after-tax investm ent p ro fits. As the tax rate rises, people begin shift ing out of p ro d uctive activities (w hich are taxed) into less p ro d u ctive, freq u e n tly leis u re, activities (w hich are not taxed) and tax revenu e drops. If the governm ent w ere to tax 100 percent of all earnings, Laffer argues, no one w ould w ork and there w ould be no revenue from taxes. D u e, in part, to its in tu itive ap p eal, the Laffer C u rve has enjo yed a m odicum of suc cess in po litical circle s. M ost econom ists, h o w e ve r, have argued that the th eo ry does not necessarily support the co nclusio n that tax cuts w ill be self-fin an cin g . Th e true shape of the Laffer C u rv e is an e m p irical q uestio n. Th e re is little e vid e n ce available to show that the curve ever bends backw ards, m uch less that it is sym m etrical. M o re o ve r, there is v ir tually no eviden ce w hich dem onstrates that o u r present tax structu re is an yw h ere near the backw ard bending portion of the cu rve. tax revenues (dollars) Congressional Budget Office, An Analysis o f the Roth-Kem p Tax Cut Proposal (Government Printing Office, 1978), pp. 14-16. 9 This conclusion is based on the liberal assumptions that actual output is currently 4 to 5 percent below poten tial production and that capital-output and labor-output ratios are constant. Ibid., pp. 8-9. 10Richard A. Musgrave and Peggy B. Musgrave, Pub lic Finance in Theory and Practice (McGraw-Hill Book Company, 1973), p. 478. 78 real after-tax rate of return on cap ital.1 Pro 1 ponents of Kem p-Roth respond with the recent findings of Stanford's M ichael Boskin that the total elasticity of private saving with "Edward F. Denison, “ A Note on Private Saving,” Review o f Econom ics and Statistics, vol. 40 (August 1958), pp. 261-7. Econom ic Perspectives respect to incom e is 0.3 to 0 .4.1 The m eth 2 odology of the study, how ever, has been strongly c ritic iz e d .1 M o reo ve r, even if Bos3 k in ’s findings are accepted, the resulting increase in saving falls far short of the m in im um 10 percent increase necessary for the Kem p-Roth cuts to be self-financing.1 4 E co n o m etric studies. Laffer and his sup porters criticized the m ajor m acroeconom ic forecasting m odels for excluding the eco nom ic responses w hich they describe. Argu ing that no present model could accurately capture the eco n om ic impact of the KempRoth proposal, Laffer constructed his own model that dem onstrates the revenue feed back effects of his cu rve. The report des cribing the m odel has been quoted as ac know ledging that “ the task of quantifying the theoretical Laffer C u rve is unach ievab le.” M o reo ve r, the only group that responds in accord with Laffer’s theory is the working poor. For all other groups, the model suggests that the governm ent should increase tax rates to increase reven u e.1 5 O th er m odels, sp ecifically reform ulated to inclu d e the Laffer hypothesis, have co n cluded that the revenue feedback effects anticipated by Laffer w ould not occu r. C o n gressional co m m ittees co m m issioned two consulting firm s, Data Resources, Inc. and Evans Econom ics, to build models to test the Laffer theory. The m odels, although quite d if ferent in co nstru ctio n , reached sim ilar co n clu sio n s: im plem entation of a 30 percent across-the-board cut in tax rates, w ithout sig 12Michael J. Boskin, “Taxation, Savings, and the Rate of Interest,” Journal o f Political Econom y, vol. 86 (March/April, 1978), pp. S3-S27. 1 Boskin defines personal saving to include consu3 merdurables; thus, increased saving does not necessarily mean that more funds are available for investment. His omission of the inflation rate from the estimating equa tion and the particular time period studied may have strongly influenced his results. Finally, the presence of substantial serial correlation may affect the statistical sig nificance of his findings. See Congressional Budget Office, p. 18. 1 Congressional Budget Office, p. 9. 4 1 “ The Impact of a Reagan-Style Tax Cut,” Business 5 W eek (June 9, 1980), pp. 90, 95. Federal Reserve Bank o f Chicago nificant exp en ditu re reductions, would add between $85 billion and $135 billion to the annual budget deficit by 1985 and would add at least 2 percentage points to the inflation rate.1 6 W h ile Kem p-Roth may have more p oliti cal support than any of the proposed consti tutional am endm ents, it also has the most potential for eco n om ic harm . W hile it will unquestionably reduce tax rates, there is little evidence to support the conclusion that it will pay for itself in the short run. In fa c, nost of the eco n om etric m odels, including those that make Laffer C u rve assum ptions, forecast that the bill w ill sim ply produce larger deficits and increased inflation unless accom panied by significant spending cuts. Summary The Congress has considered a variety of proposals for eco n o m ic reform over the past few years, several of w h ich seek to limit and /o r reduce the role of the federal govern ment in the eco nom y. O f the four plans ana lyzed here, the spending cap am endm ent appears to be the one best-suited to achieve these ends. H o w ever, econom ic evidence suggests that reducing and/or lim iting the federal g o vern m en t’s role may not elim inate the true sources of eco n om ic pain. Sim ilarly, available evid en ce casts some doubt on the reasonableness of the incom e and revenue effects predicted by proponents of the KempRoth bill. But a real test must await the tax cut's actual adoption and im plem entation. Any determ ination as to w hich of the four pro posals is “ best” is ultim ately a value judge ment and w ill vary w ith the social, political, and eco nom ic predilections of each individ ual. The purpose of this article has been sim ply to synthesize some of the econom ic in fo rm a tio n n e ce ssa ry to d e te rm in e the tradeoffs. 16lbid. See also Stephen Brooks and Otto Eckstein, "Economic Analysis of the Kemp-Roth Proposal,” Data Resources U.S. Review (August 1978), pp. 1.12-1.15. 79 The discount rate—will it float? Paul L. Kasriel Since the Federal Reserve adopted its new reserves-oriented operating p ro ced u re on O cto b er 6, 1979, the role of discount policy has com e under greater scrutiny both inside and outside the System. O f special interest has been the spread betw een the federal funds rate and the d iscount rate. Th ere is a strong positive association betw een the fed e ra l fu n d s r a te - d is c o u n t ra te s p re a d a n d the am ount of reserve adjustm ent borrow ing from the Federal R eserve.1 Some analysts have criticized the Fed for allowing this spread to w iden as m uch as it has on certain occasions. They argue that the w ide spread induces depository institutions (hereafter referred to as banks) to borrow reserves from the Fed at what might be co nsi dered a subsidy rate. In their view banks' increased incen tive to borrow from the dis count w ind ow w hen the discount rate is low relative to m oney-m arket rates dim inishes the Fed's control over total reserves and, thus, the m oney supply. In ord er to keep the spread between m oney-m arket rates and the discount rate sm aller and m ore stable, it has been suggested that the discount rate be allowed to float with a m oney-m arket rate such as the federal funds rate. This article explains the discount m echanism and dis cusses the im plications of a floating discount rate w ithin the cu rren t fram ew ork of reserve a c c o u n tin g and o p en m a rk e t o p e ra tin g p ro ced u re— nam ely, lagged reserve account ing and nonborrow ed reserve targeting. The discount mechanism U n d e r th e system of lagged reserve accounting adopted in 1968, the average level of reserves that a bank is req uired to hold as a deposit at the Fed a n d /o r in vault cash in a Reserve adjustment borrowing excludes seasonal and special borrowing. 20 As reserve adjustment borrowing from the Fed rises . . . billion dollars . . . the spread between the federal funds rate and the discount rate increases percentage points 1979 1980 1981 given reserve settlem ent w eek is determ ined by the reserve requirem ent ratios (set by the Fed) applied to the average level of the bank's reservable liabilities two w eeks prior. Thus, changes in a bank's deposits and other reser vable lia b ilitie sd u rin g th e c u rre n t reservesettlem ent w eek cannot change its required reserves for this w eek. Upon entering the set tlem ent w eek, each bank knows the average level of reserves it must hold in order to satisfy its reserve requirem ents and the Fed knows what average level of reserves it must supply so that the banking system can satisfy its reserve requirem ents. Reserves can be supplied in two ways— through Fed open market operations (no n borrow ed reserves)2 and through discount w indow lending (borrow ed reserves). Any shortfall in n onborrow ed reserves com pared 2So-called market factors such as float also provide nonborrowed reserves. The Fed attempts to offset unde sired changes in nonborrowed reserves caused by market factors through open market operations. Econom ic Perspectives to req u ired reserves must be made up by borrow ed reserves.3 U nder the Fed’s new operating p ro ced u re, open m arket op era tions are co nd ucted so as to hit a targeted level of n on b orrow ed reserves on a w eekly average basis. Since required reserves are p redeterm ined in any given w eek because of lagged reserve accou n tin g , the choice of a w eekly level of n onborrow ed reserves largely determ ines the w eekly level of borrowed reserves.4 Although the w eekly am ount of b or rowed reserves for the banking system is d eterm in ed o n ce the Fed chooses a n onb or rowed reserve target, borrow ings by ind ivid ual banks from the discount w ind ow are not. An individual bank can obtain reserves in several alternative ways, including purchas ing federal fu n d s, selling C D s, or selling a security from its portfolio. These alternatives redistribute the existing quantity of reserves among banks. Th ey do not increase the re serves of the banking system as a w hole. In contrast, borrow ing from the Fed increases both the borrow ing bank's reserves and those of the banking system. If a bank could borrow from the Fed as m uch and as often as it desired at the discount rate, then the discount rate w ould serve as a cap to the fed eral funds rate. The fact that the federal funds rate is usually above the dis count rate w hen nonb orrow ed reserves are less than req uired reserves is prim a facie e vi dence that the discount rate does not m ea sure the full cost of b orrow ing from the Fed. The full cost of borrow ing from the Fed, or 3 This abstracts from reserve carryover, the privilege banks have of carrying over a surplus or deficiency of up to 2 percent of required reserves into the following reserve settlement week. 4lf banks' demand for excess reserves (i.e., reserves in excess of those required) were zero or constant, then the choice of a weekly level of nonborrowed reserves completely determines the weekly level of borrowed reserves. To the degree that banks' demand for excess reserves varies, then a given weekly level of nonbor rowed reserves does not completely determine a weekly level of borrowed reserves. Because excess reserves tend to be relatively small and stable, the analysis is not mate rially affected by them and, therefore, it will be assumed that they are zero. Federal Reserve Bank of Chicago the effective discount rate, is the sum of the quoted discount rate plus the nonpecuniary costs resulting from discount w indow adm in istration. Because the Fed tries to lim it the am ount and duration of borrow ing by indi vidual banks, by subjecting th eir lending and investm ent practices to “ su rveillan ce,” and because banks wish to assure themselves access to the w in d o w in the future w hen they may face liq u id ity p roblem s, the n o np ecun iary costs of borrow ing an additional dollar rise with the quantity and frequency of bor row ing by an ind ivid u al bank. These costs w ould rise even w ith an unchanged level of borrow ing if the adm inistration of the dis count w indow w ere to get “ tou gh er.” To m inim ize its costs, an individual bank w ill manage its reserve position in such a way that the effective discount rate on an addi tional dollar borrow ed from the Fed w ill be equal to the cost of acquiring reserves from alternative sources. A tth e m argin, then, there is no subsidy involved in borrow ing from the Fed when the effective rather than the quoted d iscount rate is com pared with the cost of alternative sources of fu n d s.5 Because bor * rowing federal funds is a substitute for bor row ing at the discount w in d o w , this cost can be m easured by the federal funds rate. Thus, the e ffe ctiv e d isco u n t rate tends tow ard equality with the federal funds rate, and the spread betw een the federal funds rate and the quoted discount rate measures the mar ginal non p ecu n iary cost of borrowing from the Fed. If the Fed provides less nonborrow ed reserves than req u ired , then those banks for w hich the effective discount rate is higher than the costs of alternative sources of funds w ill attem pt to obtain reserves from these sources, thereb y driving up their interest rates. Some reserve-deficien t banks w ill be induced to increase th eir borrowings from 5However, there is a subsidy on average because the full nonpecuniary costs of borrowing are incurred only on the last dollar borrowed; on the intramarginal bor rowing the bank incurs below-market costs. For mone tary policy purposes, of course, it is only the marginal cost that is relevant. 21 the Fed as the alternative cost of funds rises to the level of th eir effective discount rates. Interest rates w ill co ntinue to rise until banks are induced to b orrow enough from the Fed to meet th eir req uired reserves. This rising cost of reserves w ill eventually cause banks to curtail the expansion of their assets and, hence, slow the growth of the money supply. It can be seen, th en , that a penalty dis count rate p olicy— i.e ., a policy w hereby the Fed always m aintains the quoted discount rate above the cu rren t federal funds rate— is theoretically inconsistent with lagged reserve accounting and nonborrow ed reserve target ing. The way in w hich the m arket for bank reserves com es into eq u ilib riu m w hen the level of nonborrow ed reserves is set below that of required reserves is for the federal funds rate to rise to a level above the quoted discount rate such that individual banks are induced to b orrow enough reserves from the Fed to elim in ate the reserve d eficien cy for the banking system. A penalty discount rate would prevent the reserves m arket from reaching eq u ilib riu m because no bank would be w illing to b orrow from the Fed as long as it could obtain reserves in the federal funds m arket at a rate below the discount rate. The federal funds rate w ould co ntinue to ratchet upward until the Fed provided additional nonborrow ed reserves (i.e ., above the tar geted level) to eliminate the reserve deficiency.6 W ith contem poraneous reserve account ing where the current week's required reserves are determ ined by the cu rren t w eek's reservable liabilities, a penalty discount rate w o uld, in theory, be feasible and would be eq uiva lent to the Fed closing down the discount w indow for reserve adjustm ent borrow ing. As the federal funds rate rose and banks sold securities to the nonbank p ub lic in ord er to acquire reserves, reservable liabilities of the banking system w ould d eclin e and, thus, reduce the cu rren t w eek's required reserves. The federal funds rate would continue to rise until reservable liabilities declined to the point w here required reserves w ere reduced to a level equal to nonborrow ed reserves.7 In * p ractice, sharp increases in the federal funds rate might occur so as to induce banks to make the portfolio adjustm ents necessary to reduce required reserves to the targeted level of nonborrow ed (and in this case, total) reserves in as short a time as a w eek. alternatively, at some level of rates, banks might bid so aggressively for deposits as to attract currency out of circulation, which the banks could then ship to the Fed to meet their reserve requirements in the current week. 7Under lagged reserve accounting, the rising federal funds rate would also cause the current week’s reserva ble liabilities to fall, but this would have no effect on the current week’s required reserves. 22 Floating the discount rate As m entioned at the outset, some ana lysts have suggested that the quoted discount rate be allowed to float with a particular m oney-m arket rate (e.g ., the federal funds rate) or some com posite index of moneym arket rates in order to keep the spread b etw een m arket rates and th e d isco u n t rate more stable. For exam ple, the quoted disco un t rate in the c u rre n t w eek might be set at 50 basis points (0.5 percentage points) above the previous w eek's average federal funds rate. But it has been seen that the spread betw een the federal funds rate and the quoted discount rate in the current w eek depends critica lly on the am ount of borrow ing forced on the banking system by the Fed's choice of a n on b orrow ed reserves target and the attendant n onp ecuniary costs of such borrow ing. Floating the discount rate would have no impact on the stability of the federal funds rate-discount rate spread. Floating the discount rate w o uld , how ever, have im portant im plications for the behavior of the federal funds rate and other related interest rates in a fram ew ork of lagged reserve accounting and nonborrow ed reserve ta rg e tin g . C o n s id e r th e fo llo w in g tw o relationships: (1) (RFFt - RDt) = cBR t + e, c > 0, BRt > 0 (2) R D t = RFFt- l + K, Econom ic Perspectives w h ere RFF is the federal funds rate, RD is the discount rate, c is the co efficient reflecting the m arginal non p ecu n iary costs of b orro w ing from the Fed, BR is borrow ed reserves (dollars), e is an erro r term , K is a constant (percentage points) and t refers to the tim e period (e.g ., w eek). The first relationship says that the spread (in percentage points) between the cu rren t federal funds rate and current discount rate is an increasing function of the am ount of reserves borrow ed by the banking system from the Fed. The second relationship is a form ula for floating the discount rate. The constant K can be assigned positive, negative, or zero values. Substituting (2) into (1) and ignoring e yields the follow ing relationship: the federal funds rate and the discount rate w ould start to ratchet up again.9 A som ewhat differen t result is possible if K is negative, i.e ., the cu rren t p erio d’s dis count rate is less than the previous period's federal funds rate. If borrow ed reserves are less than (or equal to) some critical value, then the federal funds rate need not rise co n tinuously but could d eclin e (or rem ain co n stant) from period to period. This critical value can be obtained from relationship (3a). These conditions may be stated as: (4) RFFt -R F F t- i ^ 0 as BRt f — • c This can be rearranged as: The critical valu e, th en , is - K / c (rem em ber, in this case, K < 0 so - K > 0). If borrow ed re serves had been above the critical value and, for some reason, fell b elo w , then the federal funds rate could d eclin e. (3a) RFFt - RFFt— = cB R t + K. 1 Conclusion If K is p ositive, i.e ., the current period's disco unt rate is set above the previous p eri od's federal funds rate, then relationship (3a) Floating the discount rate, then, would not necessarily keep the federal funds ratediscount rate spread more stable, but because of its possible upward ratcheting effect on the federal funds rate, it could tend to produce q u icker bank portfolio and deposit responses than w ould be the case under a more co n stant discount rate policy. There rem ains the question w hether the benefits of the m ore rapid deposit responses resulting from a floating discount rate policy w ould be greater than the costs of the conse quent increased interest rate vo latility.1 0 (3) RFFt - RFFt- l - K = cB R t. im p lie s th a t th e c u r r e n t p e rio d 's fe d e ra l fu n d s rate w ill always be higher than the previous p eriod's so long as the banking system is forced to b orro w from the Fed. This im plica tion also applies if K is zero. Thus, even if borrow ing from the Fed w ere d eclining , the federal funds rate w ould ratchet upward until banks' deposits and other reservable liab ili ties slowed enough to cause required reserves to fall below the Fed's nonborrow ed reserve path, at w hich point the federal funds rate w ould fall rapidly toward ze ro .8 The discount rate, being tied to the federal funds rate, w ould also p lum m et. As soon as banks w ere on ce m ore fo rced to borro w from the Fed, Relationship (1) above applies only when borrowed reserves are greater than zero. For a detailed discussion of why the federal funds rate falls rapidly toward zero as nonborrowed reserves are increased above required reserves, i.e., borrowed reserves are zero, see Robert D. Laurent, “ A Critique of the Federal Reserve’s New Oper ating Procedure,” Staff M em oranda No. 81-4 (Federal Reserve Bank of Chicago, forthcoming). Federal Reserve Bank o f Chicago 9 This analysis assumes that the Fed adheres to its nonborrowed reserves path without any interest rate constraints being self-imposed. Notice that even with a constant as opposed to floating discount rate, a nonbor rowed reserves targeting policy implies a very sharp drop in the federal funds rate once nonborrowed reserves are greater than required reserves. However, the constant discount rate policy does not imply a continuously rising federal funds rate when nonborrowed reserves are less than required reserves. 1 For a discussion of the social costs of interest rate 0 volatility, see Paul L. Kasriel, "Interest Rate Volatility in 1980,” Econom ic Perspectives, Federal Reserve Bank of Chicago, (January/February 1981), pp. 16-17. 23 Public Information Center Federal Reserve Bank of Chicago P.O. Box 834 Chicago, Illinois 60690 Please attach address label to correspondence regarding your subscription.