The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
CO o GO O c CO November 1981 Vol. 63, No. 9 00 CD > CD CO CD o r 3 What Really Happened to Interest Rates?: A Longer-Run Analysis 15 Trends in Federal Spending: 1955-86 a5 TD CD 25 The Voluntary Automobile Import Agreement with Japan — More Protectionism The Review is published 10 times per year by the Research Department o f the Federal Reserve Bank o f St. Louis. Single-copy subscriptions are available to the public fr ee o f charge. Mail requests fo r subscriptions, back issues, or address changes to: Research Department, Federal Reserve Bank o f St. Louis, P.O. Box 442, St. Louis, Missouri 63166. Articles herein may be reprinted provided the source is credited. Please provide the Bank’s Research Department with a copy o f reprinted material. What Really Happened to Interest Rates?: A Longer-Run Analysis G . J. S A N T O N I A N D C O U R T E N A Y C . S T O N E In te r e s t rate movements have becom e increasingly troublesome in recent years. Chart 1, which shows representative short- and long-term bond yields over the past 27 years, illustrates two perplexing prob lems with interest rate movements during this period. First, interest rates have risen considerably. In 1954, 3-month Treasury bill rates were close to 1.00 percent and long-term government securities yielded around 2.50 percent; during the second quarter o f 1981, the 3-month Treasury bill rate had reached nearly 17.00 percent while the yield on long-term government securities approached 15.00 percent. Second, associated with this rise in their general levels have been larger and more erratic fluctuations in interest rates as well. The purpose o f this article is to discuss the factors primarily responsible for the rise and increased variability in interest rates in recent years. The analysis is not intended, norcan itbe used, to explain every jiggle and jog in interest rates that occurred during this period. Instead, it is meant to uncover those factors that have influenced the longer-term behavior o f interest rates over the past 15 years. A Brief Summary o f Interest Rate Movements: 1954-66 and 1967-81 This article focuses on the changes in the average levels and variability o f interest rates that oc curred between two extended time periods. The first period, 1954 to 1966, was one in which interest rates were both relatively low and comparatively stable. The second period, 1967 to the present, is one in which interest rates have reached relatively high levels and demonstrated considerably greater variability. The major changes in interest rates over these two periods are shown in tables 1 and 2 for four different interest rates: the Aaa corporate bond rate, 20-year Treasury security yield, 90-day commercial paper rate and three-month Treasury bill rate. Whether the interest rate analyzed is short- or long-term, or whether a private or government interest rate is chosen, the general picture remains unchanged. On average, interest rates are considerably higher — from 384 to 420 basis points higher — in the 1967-81 period than they were from 1954 to 1966. As the tstatistics in table 1 indicate, these increases are statistically significant, allowing us to reject the hypothesis that the differences in the average levels o f interest rates in the two periods merely represent sampling error.1 In addition, interest rates have becom e consid erably more volatile in recent years. Their increased variability since 1966 is demonstrated in table 2 using several different measures o f variability. Their standard deviations, the commonly used measure o f ’ F or sam ple sizes used, a t value in excess of 2.00 is sufficient to reject the (null) hypothesis that the ob serv ed differen ce in the m ean valu es eq u a ls zero w ith 95 p e rce n t co n fid e n ce . T h is statistical test is predicated on the use o f in d ep en d en t random sam ples. Since ou r sam ples are n ot random ly ch osen , w e use the t and F tests that appear later in the paper to “ p rovid e the basis for a g ood edu ca ted guess — or what w e m ight term the art o f in feren ce.” (Italics in original.) Thom as H. W onnacott and Ronald J. W onnacott, In trod u ctory S tatistics f o r Business and E con om ics, 2nd ed. (John W iley and Sons, 1977), p. 9. For m ore on t and F tests, see virtually any statistics text. 3 FEDERAL RESERVE BANK OF ST. LOUIS NOVEMBER 1981 R e p re s e n t a tiv e Short-Term a n d Long-Term Interest Rates 1954 1 95 5 1 95 6 1957 1958 1959 1 9 6 0 1961 1 9 6 2 1 96 3 1 96 4 1 96 5 1966 1967 1 9 6 8 1 9 6 9 1 9 7 0 1971 1 97 2 1973 1 97 4 1 97 5 1 97 6 1977 1 97 8 1979 1 9 8 0 1981 L a te st d a t a plotted : 2 n d q u a r t e r variation around the mean, have more than doubled from the earlier to the later period. In addition, their average quarter-to-quarter absolute changes have tripled, as have their standard errors from regres sions o f interest rates on a time-trend variable.2 These various measures indicate that interest rates have been considerably more variable over the past 15 years than they were from 1954 to 1966. These increases in both the average levels and variability o f interest rates are more than merely statistically significant. The rising volume o f public discussion and debate suggests that these changes are econ om ically and p olitically significant as w ell. If we are to devise an effective policy to ameliorate the problems created by interest rate movements, it is important that these changes in the longer-term behavior o f interest rates be explained. 2Average absolute quarter-to-quarter changes and standard errors from tim e-trend regressions w ere used as m easures o f variability to see w hether changes in the trend growth in interest rates in the later p e riod had distorted the usefulness o f the standard deviation as a m easure o f variability. For further discu ssion on this issue, see E dw ard Foster, “ T he Variability o f Inflation, "T h e R eview o f E con om ics and Statistics (August 1978), pp. 346-48. 4 Components o f the Nominal Rate o f Interest: The Crucial Importance o f Expectations Interest rates observed in financial markets are nominal interest rates. They measure the premium that the borrower must pay in a credit transaction involving the exchange o f dollars now in return for a promise to repay dollars at some specified future date.3 The parties involved in such contracts are clearly interested in the expected future value o f money in terms o f goods and services that can be purchased with it when the loan is repaid. Because the value o f money varies inversely with movements in the general level o f prices — falling during periods of inflation and rising during deflationary periods — the expected inflation rate over the period o f the loan is one component o f the nominal rate o f interest. 3F or further discu ssion o f various interest rate con cep ts, see G . J. Santoni and C ourtenay C. Stone, “ N avigating T h rou gh the Interest Rate M orass: Som e Basic P rin cip les,” this R ev iew (M arch 1981), pp. 11-18. NOVEMBER 1981 FEDERAL RESERVE BANK OF ST. LOUIS Table 1 Average Levels of Selected Nominal Interest Rates and Average Annual Growth Rates in Money, Velocity, Output and Prices1 1954-66 Difference 1967-81 t-statistics Aaa corporate bond rate 4.06% 8.26% 4.20 2 16.43 7.62 3.842 14.33 20-year T reasu ry secu rity yie Id 3.78 C om m ercial paper rate 3.45 7.61 4.162 10.46 3-month Treasury bill rate 2 .8 6 6.76 3.902 10.24 7T 2.19 6.43 4.242 11.58 M1B 2.46 6.52 4.062 V 3.55 2 .8 8 -.6 7 -1 .2 6 2.98 - .8 5 -1 .0 5 .16 .27 3.83 Y V - Y -0 .2 7 _ = annualized percentage rate M 1 B = annualized percentage rate growth used p rio r to 1959). V = annualized percentage rate Y = annualized percentage rate of increase in the GNP im p licit price deflator. of grow th in M 1 B money stock (M 1 money stock rates of 77 - 8.38 0 .1 1 of grow th in M1B velocity measure. of grow th in real GNP. 'C alculated from quarterly data at annual rates through 11/1981. S ig n ific a n tly diffe re nt from zero at the 5 percent significance level. Table 2 Measures of Interest Rate Variability 1954-66 1967-81 Aaa corporate bond rate .63% 1.83% 2 0 -year .60 1.94 Commercial paper rate 1.05 2.82 7.231 3-month Treasury bill rate 1.05 2 .6 8 6.45' F-statistic Standard deviations Treasu ry secu rity yield 8.44 1 10.601 Mean quarter-to-quarter absolute changes Aaa corporate bond rate 20 -year Treasury secu rity yield .1 0 % .29% .1 1 .32 Com m ercial paper rate .28 .90 3-month Treasury bill rate .27 .74 Standard errors from regressions of interest rate on time trend .28% .96% .25 .79 C om m ercial paper rate .71 2.38 3-m onth Treasury bill rate .67 2.05 Aaa corporate bond rate 20 -yearTreasury secu rity yield 'T he ratio of the variances is sig nifica ntly greater than one at the 5 percent significance level. 5 FEDERAL RESERVE BANK OF ST. LOUIS The other component of the nominal rate of inter est is the expected or ex ante real rate o f interest. The real rate o f interest is the relative price paid for obtaining the use o f goods now rather than in the future.4 Narrowly interpreted, it represents the expected positive cost to the borrower (or return to the lender) after an adjustment has been made for the expected change in the general price level over the loan period. More importantly, however, the expected real rate o f interest has a pervasive influence throughout the economy because it reflects the premium that individuals place on providing themselves with present consumption goods relative to future con sumption.5 Movements in the real rate o f interest result either from changes in the underlying indi vidual preferences for present consumption goods relative to present capital goods (the sources o f future consum ption goods) or from changes in society’ s ability to transform current goods into future goods via changes in the capital stock. Regardless o f what triggers it, any change in the real rate o f interest will be reflected in the behavior o f the relative prices o f current consumption (short lived) goods in terms o f capital (long-lived) goods. A rise in the real rate o f interest will show up as a general rise in the prices of current consumption goods relative to capital goods; a decline in the real rate o f interest will appear as a decline in these relative prices. In summary, the nominal interest rate consists o f two components that, though not directly observ able, influence the actions of borrowers and lenders. The nominal interestrate (i) can be thought o f simply as the sum o f the expected real rate o f interest (r) and the expected future rate of inflation ( 7t*) over the period o f the loan, or (1) i = r + 77*.6 The important point to rem ember is that the nominal rate o f interest is always forward looking. It NOVEMBER 1981 depends upon the present expectations of borrowers and lenders about future events.7 The “Problem” with Guessing Wrong Of course, expectations can change — sometimes drastically so. These changes are a source o f public concern because lending and borrowing decisions represent bets about the future that involve wealth consequences for individuals. Changes in the nomi nal rate o f interest signal the fact that, in the aggre gate, individuals have revised their assessment of the value o f present consumption goods in terms of capital goods, their expectations o f the future rate o f inflation or both. When these changes occur, lenders, borrowers, investors and consumers face the unanticipated wealth consequences o f their past decisions. These changes capriciously redistribute wealth among individuals. Increased variability in the rate o f interest implies that such revisions are occurring more often and/or are more drastic in nature. As a result, financial market participants are subject to greater and more frequent unanticipated wealth changes and thus are faced with increased risk. To assess the factors that have produced increases in both the average level and volatility o f nominal rates o f interest, we must focus on the behavior o f the two com ponents that make up the nominal interest rate: the expected real rate o f interest and the expected rate o f inflation. Assessing Changes in the Real Rate o f Interest To what extent can the observed rise and in creased volatility o f the nominal rate o f interest since 1967 be explained by changes in the real rate o f interest? This question cannot be answered directly because the expected real rate o f interest is not directly observable. Since relative price movements always result from changes in the real interest rate, however, it should be possible to detect when these changes “Arm en Alchian and W illiam R. A llen , E xch an ge and Production: C o m p etitio n , C oord in ation and C on trol (W adsw orth, 1977), pp. 424-59. 5I b id . S ee a lso I r v in g F is h e r, T he R a te o f I n t e r e s t (T h e M acm illan C o., 1907), p. 88; and Jack H irshleifer, In vestm en t, In terest, and C apital (Prentice-H all, 1970), p. 117. eT he exact relationship is: i = r + 77 * + (r) (7 t*). E quation 1 is an approxim ation that is reasonably close to the exact relation ship for relatively low rates o f inflation and real rates o f interest. Digitized for 6 FRASER 7“ T he rate of interest is always based upon expectation, h o w e v e r little this may b e ju stified b y realization. M an makes his guess o f the future and stakes his action upon it. . . . O ur present acts must be con trolled b y the future, not as it actually is, but as it looks to us through the veil o f ch a n ce.” Fisher, The R ate o f In terest, p. 213. See also, Irving Fisher, The T heory o f In terest (K elley and M illm an, 1954), pp. 13-16, 36-58, 61 and 206-27. FEDERAL RESERVE BANK OF ST. LOUIS have occurred.8 Recall that an increase in the real interest rate will be reflectedby increases in the prices of current goods and services relative to present prices o f durable goods and capital assets. Similarly, a fall in the real interest rate is reflected as a decrease in prices o f current goods and services relative to present prices o f durable goods and capital assets. There are a wide variety o f such relative prices that can be observed, several o f which were ex amined for the 1954-66 and 1967-80 periods. Each price ratio expresses the price o f a present consump tion good relative to the price o f a more durable good or capital asset. For example, the consumer price index (CPI) is heavily weighted in terms o f present consumption goods; as such, it represents an index o f the prices o f these goods.9 The Standard and Poor’ s Stock Price Index is a price index o f the present prices (actually, net values) of capital goods. Therefore, an increase in the ratio o f the CPI to the Standard and Poor’s Index can be interpreted as a reflection o f an increase in the real rate of interest; a decrease in the ratio denotes a fall in the real interest rate. Similar reasoning applies to the other price ratios examined.10 Because it is always possible that special factors may cause individual price ratios to change for reasons other than a change in the real 8O n e alternative approach e m p loy ed has b een to estimate the exp e cte d inflation rate and subtraetthis from the nom inal interest rate. In essen ce, this procedu re transposes equation 1 to re = i — tr|, w here tt% den otes an estim ate o f the exp ected inflation rate and re is the derived estim ate o f the real rate. Unfortunately, these em pirical estimates are subject to tw o major criticism s. First, during certain periods, these real interest rate estimates have b e e n negative. This is a nonsensical result. T h e exp ected real rate o f interest is always positive. See Friedrich A. H ayek, The Pure T heory o f C apital (The University o f C hicago Press, 1941), pp. 223-24; and Fisher, The T heory o f In terest, pp. 186-94. S econ d, co m m o n ly d erived estimates ol the exp ected inflation rate are biased if the real rate is changing, leadin g to erroneous estim ates o f the real rate if this m ethod is em p loyed . See W . W. Brow n and G . J. Santoni, “ Unreal Estimates o f the Real Rate o f Interest,” this R eview (January 1981), pp. 18-26. 9Bureau o f L a b or Statistics, H an d book on M eth od s, Bulletin 1910 (1976). For a discu ssion o f the various price in dices cu r rently used, see W illiam H. W allace and W illiam E. C ullison, M easuring Price C hanges: A Study o f the Price Indexes, 4th ed. (Federal Reserve Bank o f R ichm ond, 1979). 1“T h ese ratios are n o t equal to the ex ante real rate o f interest. H ow ever, they are functions o f the real rate o f interest. T o see this, let P0 equal the price o f a present con su m p tion g o o d and let Pc equal the present price ol a capital g ood . Pc can be represented as follow s: 11 Pt <lt TTT75 NOVEMBER 1981 rate o f interest, the behavior o f several such price ratios must be examined.11 Has the Real Rate o f Interest Changed? The price ratios examined provide evidence that the rise in the average levels o f nominal rates of interest was not produced by a rise in the average level of the real rate o f interest.12 Data on the average level o f four price ratios are presented in table 3; differences in the average levels o f the price ratios between 1954-66 and 1967-80 are presented in column 4. Two of the price ratios, the ratio o f the CPI to the Standard and Poor’ s Stock Index and the ratio o f the price o f lamb to the price o f sheep, declined on average, presumably an indication that the real rate of interest declined. Neither o f these changes, however, is statistically significant, which means that we cannot reject the hypothesis that the average real rate o f interest was actually unchanged between the two periods. Two price ratios, the ratio o f the nondurable goods com ponent o f the CPI to the durable goods component and the ratio o f the price o fb e e f to the price o f cattle, rose, on average, osten sibly signaling a rise in the real rate o f interest. H ow ever, the rise in the average ratio o f b e e f to cattle prices is, again, not statistically significant; w here qt represents the quantity o l good s and services pro du ced at time t, Pt are the prices (net o f costs o f production) at tim e t, and i is the nom inal rate o f interest. Inflation can b e in trod uced into the analysis as follow s (using the exact relationship show n in footn ote 6): Qt ( ii)P c -2 ^ tt o (1r * /(1PI oiT-\t /I + r ) f (1 )tqV -L + "* )* r o | t io (1 + r )4 w here r and rr* are the ex ante real rate and the exp ected rate o f inflation, respectively. This results in (iii) Index = Pq /P c = ^ t=o qt (1 + r)t This index rises w h en ever r rises and falls w h en ev er r falls. "C h a n g e s in special circum stances tend to occu r random ly and are as lik ely to raise as to lo w e r the price ratio. 12O f course, year-to-year fluctuations in the real rate w ill occur. For exam ple, there is e v id e n ce that the real rate o f interest rose in 1973-75. See B row n and Santoni, “ U nreal E stim ates.” T h ese fluctuations are not critical to the present question. W e are interested here in w h eth er the average level around which these fluctuations occu r is different during the m ore recent p e r io d and, i f so, to w hat extent this ch ange exp lains the d ifferen ce in the average level o f the nom inal interest rate b etw een the tw o periods. 7 NOVEMBER 1981 FEDERAL RESERVE BANK OF ST. LOUIS Table 3 Table 4 Average Levels of Proxies for Changes in the Real Interest Rate1 Standard Deviations of Proxies for Changes in the Real Interest Rate 1967-802 1954-66 1967-80 F-statistic Ratio 1954-66 R1 1.599 1.549 -.0 5 0 -.3 0 R1 .450 .404 1.24 R2 .935 1.074 .139 7.87 R2 .025 .059 5.571 R3 .163 .166 .003 .39 R3 .017 .023 1.83 R4 .385 .352 -.0 3 3 -1 .7 3 R4 .044 .053 1.45 Difference t-statistic R1 = R a tio o fth e c o n s u m e rp ric e in d e x (C P I)to th e S ta n d a rd and P oo r’s Index of common stock prices. 1The ratio o f the variances is sig nifica ntly greater than one at the 5 percent significance level. For footnotes, see table 3. R2 = Ratio of the nondurable good com ponent of the CPI to the durable good com ponent of the CPI. R3 = Ratio of the price of beef per cwt. to the price of cattle per head. R4 = Ratio of the price of lamb per cwt. to the price of sheep per head. 'C alculated from annual data. 2Data fo r R3 and R4 are currently available only through 1979. Sources: U.S. Departm ent of Labor, Bureau of Labor Statis tics; Standard and P oor's; U.S. D epartm ent of Commerce, Statistical Abstract o f the U.S.; and U.S. Department of Agriculture, A g ricultu ral Sta tistics Annua! (1979). we are unable to determine whether the change in the average level o f the ratio o f the nondurable to the durable good s co m p o n e n t o f the CPI is significant.13 Thus, these ratios provide no statistical evidence that the real rate of interest has changed, on average, between the two time periods. Therefore, the increase in the average level o f nomi nal interest rates in recent years cannot be attributed to an increase in the average level of the real interest rate. Nominal interest rates were more volatile during the 1967-81 period than during the 1954-66 period as shown by the statistically significant increase in their variances in the more recent period (see table 2).14 This increase in volatility however, can not be 13W h en em p loyin g the t test for d ifferen ces in m eans for sm all sized sam ples (less than 30 observation s), the sam ples are assum ed to b e drawn from n orm ally distributed populations having the same standard deviations. This assum ption is v io lated in the case o f R2 (see the F-statistic in table 4) and, thus, the t-test for the significance o f the differen ce in means is inappropriate. 14T h e F-test is the ratio o f the larger variance to the smaller variance, w h ere the variance is the square o f the standard deviation. F or the sam ple sizes used, an F-ratio in excess of 1.60 is sufficient to reject the (null) hypothesis that the variances are equal w ith 95 percent con fid en ce. 8 explained by increased volatility in the real rate o f interest. As shown in table 4, only one relative price ratio, the ratio o f the nondurable to the durable goods component o f the CPI, demonstrates any significant increase in variance. For three price ratios, there is no statistically discernible change in their variances between the two periods. Again, the preponderance o f evidence suggests that the increased volatility in nominal interest rates did not arise from greater variation in the real rate o f interest. Therefore, the solution to the puzzle o f nominal interest rate b e havior in the more recent period lies elsewhere. Expected Inflation, Actual Inflation and Measured Inflation The econom ic theory summarized by equation 1 suggests that if changes in the real rate o f interest do not explain the increases in the mean level and volatility o f nominal interest rates between the two periods, then changes in the expected rate o f inflation mustbe responsible. To analyze properly the impact o f inflation on nominal interest rates, however, three different concepts regarding the rate o f inflation must be distinguished: 1) the expected rate o f inflation in the general level o f prices, 2) the actual or true rate o f inflation in the general level o f prices and 3) the measured rate o f inflation. It is the expected rate o f inflation ( 77*) that is impor tant in explaining the behavior o f nominal interest rates; this is the inflation variable that appears in equation 1. Unfortunately, the expected inflation rate is not directly observable. The expected inflation rate represents the public’ s FEDERAL RESERVE BANK OF ST. LOUIS best “ guess” about what the actual or true future rate o f inflation in the general level o f prices will be. This guess generally will be incorrect for any specific time period; the future is, after all, uncertain and, therefore, subject to a variety o f random shocks. However, econom ic theory suggests that because there are large wealth consequences associated with these predictions, estimates o f the true rate o f inflation made by individuals will be unbiased; that is, although these predictions generally will be wrong due to the occurrence o f unexpected future events, they will not consistently over- or underpredict the true rate o f inflation.15 As a result, if we knew about changes in the true rate o f inflation, we could use this information as a proxy for changes in the (unobservable) expected rate o f inflation. However, the true rate o f inflation in the general price level is also unobservable. We have no direct information on this rate; instead, we have informa tion on several different measured rates o f inflation. These are typically derived from changes in various price indices, such as the CPI, the GNP deflator and the Index o f Producer’ s Prices. Unfortunately, the measured rate o f inflation may, under certain circumstances, differ significantly from the true rate o f inflation in the general level o f prices. This potential divergence occurs because the price indices mentioned above typically include a relatively narrow sample o f all o f the goods that are available for purchase. In particular, they generally exclude the prices o f existing capital assets, thereby placing greater w eight on the prices o f current 15“ B efore p r o ce e d in g to sp ecific statistics, it is im portant to em p h asize the broad fact that . . . b usin ess foresight exists and that the accuracy and p o w e r o f this foresight is greater today than ev e r b e fo re . . . . E very chance for gain is eagerly w atch ed. An active and in telligent speculation is constantly g o in g on , w h ich . . . perform s a w ell-k n ow n and providen t fun ction for society. Is it reasonable to b e lie v e that foresight . . . has an excep tion as a p p lied to falling or rising p rices? Or, if so, can the academ ic . . . assume h im self possessed o f a fore sight o f w h ich he says the practical man is in capable? It is the practical m an’ s business to foresee. It is he w h o first gathers the facts and statistics. . . . It is he w h o w atches the trends. . . . A n d it is in his trade journals that w e find the first discussions o f the proba b le e ffe ct o f gold discov eries or silver legislation on prices and trade. T h e theorist can aid in these predictions o n ly b y supplying the p rin cip le on w h ich they are con stru cted .” Irving Fisher, A p p recia tion and In terest (Augustus M. K elley, 1965), pp . 36-37. For further discu ssion o f efficien t markets, see E u gen e F. Fama, “ E fficient Capital Markets: A R ev iew o f T h eoretical and Em pirical W ork,” Journal o f F in ance, Papers and P roceedin gs (M ay 1970), pp. 383-417. 16Arm en A. Alchian and Benjam in Klein, “ O n a C orrect Measure o f Inflation,” Journal o f M on ey, C red it and Banking (February 1973), pp. 173-91. NOVEMBER 1981 consumption goods.16 Measured rates o f inflation will therefore produce biased estimates o f the true rate o f inflation in the general level o f prices on those occasions when the prices o f present consumption goods are changing relative to the prices o f capital assets. This specific bias occurs whenever the real rate o f interest changes. Consequently, the mea sured inflation rate overstates the true rise in the general level o f prices when the real rate o f interest rises, and understates the true rate o f inflation when the real rate o f interest declines. Fortunately, it is possible to “ link” the unobserv able average rate o f expected inflation to the observ able average rate of measured inflation, if expectations o f individuals regarding the future rate o f inflation yield unbiased predictions o f the true rate o f inflation and if, in addition, the average level o f the real in terest rate is unchanged over the period o f analysis. The first “ i f ’ enables us to link the expected to the actual rate o f inflation. The second “ i f ’ lets us relate the actual to the measured rate o f inflation. Note the critical importance o f our previous find ing that the average level o f the real interest rate did not change between the 1954-66 period and the 1967-80 period. Because o f this, we would not expect the average measured rate o f inflation over these two periods to differ from the true rate o f inflation.17 Because we believe that the expected rate o f inflation is an unbiased estimate o f the actual rate o f inflation, we can directly link the average rate o f expected inflation to the average rate o f measured inflation. The Relationship Between Measured Inflation and Nominal Interest Rates Inflation statistics using the GNP implicit price deflator are reported in tables 1 and 5. The data in table 1 indicate that, on average, the measured rate o f inflation ( 77) was 4.24 percent higher during the 1967-81 period than during the 1954-66 period. This increase closely parallels the 384 to 420 basis-point increases in the average levels o f the nominal inter est rates reported in table 1. In fact, none o f these 17W h ile it is true that year-to-year changes in the real rate o f interest have occu rred and thus have in troduced a bias into the m easured rate o f inflation at those points in tim e, these fluctua tions have apparently averaged out. As a result, apart from other p r o b le m s , any o b s e r v e d d iffe r e n c e b e tw e e n the a v era g e m easured rate o f inflation b etw een the tw o subperiods can not b e explained by m easurem ent error in trod uced by a change in the average level o f the real rate o f interest. 9 NOVEMBER 1981 FEDERAL RESERVE BANK OF ST. LOUIS increases in the average levels o f interest rates are significantly different from the increase in the average level o f inflation.18 The rise in the average rate o f inflation fully “ explains” the average increases in these nominal interest rates. Data for various measures o f variability in the measured rate o f inflation for the two periods are reported in table 5. Tw o of the three measures indi cate increased variability in the rate o f inflation. The standard deviation of the measured rate of inflation in the more recent period is almost twice that in the earlier period; this closely parallels the increase in the standard deviations of the nominal interest rates reported in table 2. The standard error associated with regressing the rate o f inflation on a simple time trend has increased by 43 percent. Only the mean quarter-to-quarter absolute change in the rate o f inflation shows no increase in the later period. This evidence suggests that the rise in the average level o f nominal interest rates since 1967 and, to a lesser extent, their increased volatility can be ex plained by the increase in the level and volatility o f the rate o f inflation. What remains, therefore, to complete the analysis o f interest rate movements in recent years is to determine why the rate of inflation has changed. Determinants o f the Rate o f Inflation The increase in the average level and volatility o f the rate o f inflation during the 1967-81 period could have been produced by changes in the b e havior of the growth in the money stock (Kl), in the growth o f its velocity o f circulation (V), or in the growth of real output (Y). The relationship between the rate o f inflation and these variables is given by the following identity: (2) 7T = Id + V — Y. The rate of inflation is positively related to changes in the growth rates o f the money stock and its velocity o f circulation, and inversely related to changes in the growth rate o f output. Econom ic theory converts this identity into a hy pothesis about the long-term relationship between the growth rate in money and the rate of inflation by 18T h e t-statistics for the (null) hypothesis that the change in the interest rates b etw een the tw o periods is exactly 4.24 percent are —0.16, —1.49, —0.20, and —0.89, for the interest rates show n in table 1, respectively. Digitized for 10 FRASER Table 5 Measures of Variability in Growth Rates of Prices, Money, Velocity and Real Output1 1954-66 1967-81 F-statistic Standard deviations 3.422 7r 1.31% 2.42% M1B 2.32 2.76 1.41 V 3.85 3.79 1.03 Y 4.20 4.30 1.05 V -Y 2.62 3.58 1.872 Mean quarter-to-quarter absolute changes TT 1.44% 1.38% M1B 1.63 2.49 V 3.45 4.19 Y 3.49 4.14 V -Y 2.05 2.80 Standard errors from regressions against time trend TT 1.32% 1.89% M1B 2.22 2.72 V 3.88 3.74 Y 4.10 4.33 'C a lcu la te d from qu arterly data at annual rates throu gh 11/1981. 2The ratio of the variances is sig nifica ntly greater than one at the 5 percent significance level. arguing that the long-term growth rates in velocity and output are essentially unaffected by changes in the long-term growth rate in the stock o f money.19 As a result, changes in the long-term rate o f growth in the money stock will be directly matched by changes in the rate o f inflation if the long-term rates o f growth in velocity and output remain unchanged. Since our analysis covers two relatively long time periods, we should be able to determine the extent to which each o f these factors has influenced the long-term changes in the rate o f inflation and, by extension, the changes in nominal interest rates. 19Irving Fisher, The Purchasing P ow er o f M on ey (Augustus M. K elley, 1963), p. 14, notes that: “ This theory, though often cru d ely fonnulated, has b een a ccep ted by L ock e, H um e, Adam Smith, Ricardo, M ill, W alker, M arshall, H adley, Fetter, Kem merer, and m ost writers on the subject. T h e Rom an Julius Paulus, about 200 A .D ., states his b e lie f that the value o f m on ey d ep en d s u pon its quantity.” See also pp . 157-59. O n page 296-97, F ish er states that the “ Law [em phasis in original] o f direct proportion b etw een [the] quantity o f m on ey and the price l e v e l . . . is as im portant to the theory o f m on ey as B o y le ’ s Law is to the physical theory o f gases.” FEDERAL RESERVE BANK OF ST. LOUIS Has the Long-Term Growth Rate in Real Output Changed? The long-term growth rate in output is primarily determined by the rate at which capital is accumu lated.20 The average growth rate in real output was 3.4 percent per year over the 1954-81 period. O f course, the growth rate o f output fluctuates consid erably from quarter to quarter around its long-term average rate. The pattern o f the short-term variations in output growth and the average rate o f growth over the entire 1954-81 period are shown in the first tier o f chart 2. The quarter-to-quarter growth rates in real output vary considerably, from a positive 11 percent during the first quarter o f 1973 to a negative 10 percent during the second quarter o f 1980. However, as the summary statistics in tables 1 and 5 indicate, both the average rate o f growth in output and its variance are essentially unchanged between the two periods. The reported differences in both the mean growth rates and variances are not statistically significant. Thus, the statistical evidence does not support the claim that changes in the average growth in real output and its variability are responsible for the higher and more variable inflation rate. Has the Long-Term Growth Rate in Velocity Changed? The velocity o f money measures the relationship between total spending and the stock o f money. The larger the velocity, the greater the amount o f spend ing that a unit o f money will finance during any given period.21 This is why, given the quantity of money and real output, an increase in velocity is associated with a rise in prices. Like real output, velocity has been increasing. The average rate o f growth in velocity was about 3.3 percent per year over the 1954-81 period, virtually cancelling out the effect on prices resulting from the average growth in real output as shown in equation 2. 20F or a com p lete discu ssion, see Fisher, The Purchasing P ow er o f M o n ey, pp. 74-111. F ish er in clud es the quantity o f natural resources, the div ision o f labor, tech n iq u e o f produ ction , variety o f wants, facilities for transportation, institutional arrangements regarding the rights o f in dividuals to contract, etc. “ V elocity dep en d s u pon such things as the e x p ected rate o f inflation, the rate at w hich ch ecks are cleared and the use o f credit. In addition to these, Fisher, The Purchasing P ow er o f M on ey, p. 79-89, adds preferen ces to hoard, the tim ing and regularity o f receipts and disbursem ents, population density and transportation facilities. NOVEMBER 1981 The second tier o f chart 2 shows quarter-to-quarter growth rates in velocity and the average growth rate for the entire 1954-81 period. These quarter-toquarter growth rates fluctuate considerably around the long-term average. However, as the data in tables 1 and 5 indicate, there has been no significant change in the average rate o f growth in velocity or in its volatility between the two periods. Differences in the mean growth rates and the variances between the two periods are not statistically significant. Consequently, the increase in the average level and variability o f the rate o f inflation can not be explained by changes in the growth rate of velocity. Has the Relationship Between Velocity and Real Output Growth Changed? Although the increase in the level and volatility o f inflation since 1967 is not explained by changes in the growth rates o f velocity and real output when analyzed separately, it is really the difference b e tween these two growth rates that measures the longer-run, non-monetary influence on inflation (see equation 2). Therefore, an explicit analysis o f this difference may reveal significant changes that do not appear when velocity and real output growth rates are examined individually. Evidence bearing on this issue is presented in tables 1 and 5; the difference is depicted graphically in the third tier o f chart 2. As shown in table 1, there has been no statistically significant increase in the average difference be tween the growth rates for velocity and real output. Not only is this difference not significantly different from zero in each period shown, the 0.16 rise in the difference in the later period is explainable by sampling error alone.22 Thus, the increase in the average inflation rate since 1967 can not be attributed to significant changes in the growth o f velocity and real output, whether taken separately or in tandem. The results in table 5 suggest, however, that the increased volatility o f inflation since 1967 has been accompanied by a significant increase in the vola tility o f the difference between velocity growth and real output growth. The rise in the standard devia tion is statistically significant; the mean absolute quarter-to-quarter changes have risen 37 percent in the 1967-81 period. Therefore, the greater volatility in this difference since 1967 has contributed to increased variability in the rate o f inflation. 22T h e t-statistics for the d ifferen ce betw een velocity and output grow th in each p eriod are —0.74 for 1954-66 and —0.23 for 1967-81. 11 FEDERAL RESERVE BANK OF ST. LOUIS NOVEMBER 1981 C h art 2 G r o w t h Rates of Real O u t p u t a n d V e lo c ity Perceit 12 erence b e tw e e n velocity g ro w th rate and re al output g ro w th rate 10 8 6 Quarterly velocity growth l1 ■minus real output growth 4 ferem 2 0 -2 -4 -6 -6 1954 1955 1956 1957 1958 1959 1960 11 C o m p o u n d e d a n n u a l r a t e s o f c h a n g e , la t e s t d a t a plotte d : 2 n d q u a r t e r Digitized for12 FRASER 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 NOVEMBER 1981 FEDERAL RESERVE BANK OF ST. LOUIS C h art 3 G r o w th Rates of G N P Implicit Price De flato r a n d M1B 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 U_ C o m p o u n d e d a n n u a l ra te s of c h a n g e . L atest d a t a plotted : 2 n d q u a rte r Money Growth and Inflation As we observed above, the combined effect on the rate o f inflation o f the average growth rates in velocity and real output have been virtually of fsetting — their net impact on the rate o f inflation was essentially zero, as shown in the third tier o f chart 2. Therefore, the rate of inflation since 1954 must be closely related to the rate o f growth in the money stock. The close long-run relationship between the growth in M1B and the rise in the GNP implicit price deflator in both periods is evident from a comparison o f their mean growth rates in table l.23 During 23For e v id e n ce that M1B is the preferable monetary aggregate to use in assessing the m on ey-price link, see Keith M. Carlson, “ T he Lag From M on ey to P rices,” this R eview (O ctob er 1980) pp. 3-10; Keith M. Carlson and Scott E. H ein , “ M onetary Aggregates as M onetary Indicators,” this R ev iew (N ovem b er 1980), pp. 12-21; R. W . H afer, “ S electin g a M onetary Indicator. A T e s t o f th e N e w M o n e ta ry A g g r e g a te s ,” this R e v ie w (February 1981), pp. 12-18; and R. W. H afer, “ M uch A d o A bout M 2 ,” this R eview (O ctob er 1981), pp. 13-18. the 1954-66 period, M1B growth averaged 2.46 percent per year and prices rose, on average, 2.19 percent per year; since 1967, growth in MIBhas averaged 6.52 percent per year and prices have risen, on average, at an annual rate o f 6.43 percent. There is no significant statistical difference between the long-run growth in M1B and that in prices in either period. The small differences that w e observe b e tween inflation and money growth in both periods can be attributed simply to random error.24 The close relationship betw een inflation and money growth is important because, unlike the other variables cited above, the long-term rate o f growth in the stock o f money is a direct consequence of monetary policy actions. Although the money stock may vary randomly over periods up to a month or 24T he t-statistics for the h ypothesis that the rate o f inflation equals the rate o f m on ey grow th are - 0 .7 3 for 1954-66 and —0.19 for 1967-81. 13 FEDERAL RESERVE BANK OF ST. LOUIS NOVEMBER 1981 two, these short-run fluctuations have virtually no discernible impact on inflation. As the data in table 1 point out, the crucial variable for explaining the long-run behavior o f prices is the behavior o f the policy-determined, long-term rate o f growth in money. money and their increased variability. These results indicate that the increase in the average rate o f inflation and, to a greatextent, the rise in its volatility can be directly associated with similar movements in the growth o f M1B. The significant rise in the average rate o f growth in the money stock (M IB) and its volatility are shown in tables 1 and 5. The substantial increase in the long-term average growth rate in money between the two periods essentially accounts for the entire change in the rate o f inflation. The average annual growth rate in M1B increased by 4.06 percent between the two periods; the average rate of inflation rose 4.24 percent per year. There is no signif icant statistical difference between the average increase in the rate o f inflation and the rise in M1B growth since 1967.25 Conclusion Also, as shown in table 5, money growth has becom e more volatile since 1967. Its standard devia tion has increased by 19 percent over the more recent period. Further, its mean absolute quarter-toquarter change has risen 53 percent; its standard error (from a time-trend regression) has increased 23 percent. Chart 3 provides evidence o f the close relationship between the average rise in the growth of prices and 25T he t-statistic for the (null) hypothesis that the rise in the rate o f inflation since 1967 actually equals 4.06 percent (the rise in m on ey grow th) is 0.49. Digitized for 14 FRASER Over the past 15 years, nominal interest rates have been higher and more volatile, on average, than they were from 1954 to 1966. Judging from the volume o f public discussion, these increases are economically and politically disquieting. Besides satisfying a natural curiosity about why such drastic changes in financial markets have occurred, one major reason for studying the factors responsible is to determine the extent to which policy actions can influence interest rate movements. The evidence in this article suggests that higher and more variable money growth since 1967 have been primarily responsible for the longer-term rise and increased variability in interest rates. Alterna tive explanations o f the movements in interest rates — that they are due, in part, to higher and more variable real interest rates or to significant changes in the behavior o f velocity or real output growth — are not generally supported by the longer-run evi dence cited here. A corollary to the analysis is that, barring fortuitous (but so far unobserved) changes in velocity and output growth, only an extended period o f lower and less variable money growth is likely to generate lower and more stable interest rates. Trends in Federal Spending: 1955-86 K E IT H M . C A R L S O N T HE Reagan administration has embarked on an ambitious program to slow the growth of federal spending, a program that is part o f an overall econom ic plan to reduce inflation and promote sustainable e con om ic growth. The purpose o f slowing the growth o f federal outlays in the overall program is to shift resources from the public to the private sector. As o f July 15, 1981, the administration had pro posed a reduction in the growth o f federal outlays over the next five years to a 6.2 percent annual rate, down from an estimated 12.5 percent annual rate from 1976 to 1981.1 The planned slowing in federal spending is especially pronounced in the early years o f the projection period. Outlays are projected to grow at only a 4.7 percent rate from 1981 to 1984, followed by an 8.6 percent rate from 1984 to 1986. The spending plan is targeted to reduce federal outlays to 18.6 percent o f GNP in 1986 from an estimated 23.0 percent in 1981. A considerable amount o f budget discussion is couched in terms of expenditure “ cuts.” For the most part, however, these spending plans are not cuts at all, but reductions in spending from what they would otherwise be. Thus, any attempt to assess budget developm ents and/or the administration program must come to grips with that elusive esti mate o f what outlays “ would otherwise be.” The Congressional Budget Office (CBO) has prepared a set o f such estimates, which it calls “ baseline pro 1All references to years in this article are to fiscal years, unless oth erw ise indicated. jections.” 2 The administration’ s spending plan will be presented in light o f these baseline projections. Any assessment o f current developm ents and future trends, moreover, requires a sense o f his torical perspective.3 Thus, this article reviews the course of federal spending over the last 25 years, focusing on the growth o f federal outlays relative to the size o f the econom y as measured by GNP. Trends in the composition o f federal spending by major program category also are summarized. PAST TRENDS IN FEDERAL SPENDING: 1955-80 To obtain a sense of historical perspective on federal spending, trends are examined for the period 1955 through 1980. Using 1955 as a starting point removes most o f the influence o f World War II and the Korean War, periods o f extensive defense spend ing, yet the period still includes the Cold War o f the 1950s and the Vietnam War. Excluding all defense buildups is undesirable since political and inter national conditions will always impinge to some degree on decision-making processes relating to federal spending. C o n g r e s s io n a l B u d get O ffice , B aselin e B u d get P ro jectio n s: Fiscal Years 1982-1986 (July 1981). 3T he pu rpose o f the article is to describe, rather than analyze, federal spen din g trends. F o r a discu ssion o f the theoretical basis for various governm ental activities, see Richard A. M usgrave and P eggy B. M usgrave, Public F inance in T heory and Practice, 2nd e d ., (M c G r a w -H ill B o o k C o m p a n y , 1 976). S ee a lso Sam Peltzman, “ T he G row th o f G overn m en t,” The Journal o f Law and E con om ics (O ctob er 1980), pp. 209-88. 15 FEDERAL RESERVE BANK OF ST. LOUIS To review trends in federal spending, it is useful to categorize federal outlays. The Office o f Manage ment and Budget develops the budget each year in two fundamental ways: by agency and by function.4 These categorizations are important for the budget planning process, but for this article, a smaller number o f categories is preferable. The categor ization chosen is by major program, a categorization used by the CBO. Explanation o f Major Program Categories Categorizing federal outlays by major program essentially divides government activities into de fense and non-defense spending. The latter category is further subdivided according to the form that this spending takes. Table 1 summarizes the major program categories used as a basis for assessing federal spending trends. National defense consists mainly o f the military activities o f the Department o f Defense. The de fense category, how ever, also includes benefit payments for retired military personnel and D e partment o f Energy programs diverted toward national defense. NOVEMBER 1981 Table 1 Federal Spending by Major Program Categories (billions of dollars) Program National defense Non-defense Benefit payments fo r individuals Social security Medicare and Medicaid Unem ploym ent com pensation Public assistance and related program s Federal employee retirem ent and disability Food and nutrition assistance Veterans com pensation, pensions and readjustm ent benefits Other CBO estim ate fo r 1981 $159.6 500.2 316.1 138.3 56.7 22.7 18.8 18.1 16.3 14.9 30.3 Other grants to state and local governments 56.9 Net interest Other federal operations 66.1 61.1 Total $659.8 In the non-defense category, the largest com ponent consists o f benefit payments to individuals. 4T h e bu d get breakdow ns used b y the O ffice o f M anagem ent and Budget are as follow s: By A g en cy L egislative and ju d icia l branches Funds appropriated to the president Agriculture D efen se — Military D efen se — C ivil Education E nergy Health and Human Services H ousing and Urban D evelop m en t Interior Justice Labor State Transportation Treasury Environm ental Protection A gen cy National Aeronautics and Space Administration O ffice o f person nel m anagem ent Veterans administration O th er agen cies A llow an ces U ndistributed offsetting receipts Digitized for16 FRASER By Function National defen se International affairs G eneral scien ce, space and tech n ology EnergyNatural resources and environm ent Agriculture C om m erce and h ousing credit Transportation C om m unity and regional d ev elop m en t E ducation, training, em p loym en t and social services Health In com e security Veterans benefits and services Adm inistration o f ju stice General governm ent G eneral pu rpose fiscal assistance Interest A llow an ces U ndistributed offsetting receipts These programs include direct payments from the federal governm ent to individuals (e.g., social security and federal retirement pay) and indirect payments through state and local governments (e.g., public assistance and child nutrition). Some pro grams provide cash payments for recipients to use at their discretion, w hile other programs provide specific services (e.g., Medicare and Medicaid). Grants to state and local governments, other than benefit payments, include general revenue sharing, the Comprehensive Employment and Training Act, education, community development, highway con struction, etc. Net interest is the interest paid on that portion o f the federal debt held by the public. This is a net figure because it excludes interest paid to gov ernment trust funds that hold government securities, while including interest payments from federal agencies and the public on borrowing from the government. Other federal operations include farm price sup ports, domestic energy programs, foreign aid, gen eral science research, space technology, etc. In addition, general expenses required to run the government are included in this category. FEDERAL RESERVE BANK OF ST. LOUIS NOVEMBER 1981 C h a rt l Federal B u dg et O u t l a y s as a Percent of G N P 11 1955 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 1986 |J_ E stim ates fo r 1981-1986 a re a lte rn a tiv e p ro je c tio n s : d a s h e d lines are CBO b a s e lin e p ro je c tio n s a n d s o lid lines are a d m in is tra tio n p ro je c tio n s fro m M id -S e ssion Review o f the 1982 B u d g e t (July 1981). [2 O th e r f e d e r a l o p e ra tio n s , plus g ra n ts to state and lo c a l g o v e rn m e n ts o th e r th a n b e n e fit paym ents fo r in d iv id u a ls . Outlays Relative to GNP A very useful way to summarize trends in federal spending is to compare them with the growth o f GNP, a comparison that indicates the degree o f government intervention in the economy. Chart 1 summarizes the historical record by showing total federal outlays and the major program categories as a percent o f GNP. 17 FEDERAL RESERVE BANK OF ST. LOUIS The trend of total budget outlays relative to GNP has been unmistakably upward during the 1955-80 period, rising from 17.9 percent in 1955 to 22.6 percent in 1980. This trend has not been smooth, however, in that outlays relative to GNP have surged in relatively short periods. Though the percentage tends to subside after the surge, it returns to levels higher than prevailed before the surge. The surges during the 1955-80 period seem to be associated with (1) the 1957-58 recession and the abbreviated recovery that followed, (2) the Vietnam War, and (3) the 1973-75 recession. But the reason outlays as a percent o f GNP does not return to pre-surge levels is unclear, except, perhaps, as a result o f the momen tum o f the government spending process. National defense spending relative to GNP de clined throughout 1955-80 except for the period o f the Vietnam War. In 1955, 10.4 percent o f the na tion’s GNP was directed toward defense outlays, a percentage that declined to 7.2 percent in 1965, before rising to 9.4 percent during the Vietnam War in 1968. Since then, the decline has been dramatic, with the ratio plummeting to 5 percent in 1978 and 1979. M eanw hile, benefit payments for individuals were only 3.7 percent o f GNP in 1955 with social security accounting for 31.4 percent o f the total. With a major surge in the 1967-76 period, benefit pay ments hit 10.6 percent o f GNP in 1980 as social security rose to 43.2 percent o f the total. Given the decline in national defense, the bulk o f the relative rise in total outlays over the 1955-80 period is attributable to the sharp increase in benefit pay ments to individuals. The remaining major program categories, though relatively small, show some trends. Net interest was virtually constant at 1.3 percent of GNP from 1955 to 1968. Since then, the trend has been upward as the government runs continuous deficits and interest rates keep rising. Net interest reached 2.0 percent o f GNP in 1980. Other grants to state and local governments, the smallest category in 1955, has been trending upward throughout the period (not shown separately in chart 1). Starting at 0.4 percent o f GNP in 1955, these grants rose to 2.2 percent in 1980. All other outlays relative to GNP, a residual com ponent o f the total, changed little on balance from 1955 to 1980. More recently, from 1975 to 1980, federal spend 18 NOVEMBER 1981 ing has grown only slightly faster than GNP. Total outlays averaged 22.6 percent o f GNP in 1980, with the decline in defense offset by an upward creep in non-defense categories. Program Categories Relative to Total Outlays Another way o f looking at federal spending is to examine the composition o f total outlays. Using the same program categories as before, chart 2 summa rizes the composition o f federal outlays for 1955-80. Chart 2 shows the changing composition o f budget outlays more dramatically than chart 1, though the same basic data are used in the construction of both. The sharply declining portion o f the budget for na tional defense is immediately evident. In 1955, 58.1 percent o f total outlays went to defense; by 1980, this proportion had dwindled to 23.4 percent, though the trend has been relatively stable since 1976. The rise in benefit payments for individuals as a percent o f total outlays, w hich has m ore than doubled since 1955, actually took place during two subperiods. First was the period from 1955 to 1961, which included two recessions with a weak recovery sandwiched in between. Second was the period from 1968 to 1976, a period o f expanding social programs that also included two recessions. Net interest held steady at about 1 percent o f total outlays from 1955 to 1974. Since then, the percent age has been rising slowly but steadily. Other grants to state and local governments ex hibited an upward trend from 1955 to 1973, but have since leveled off. All other outlays, on the other hand, have varied considerably, rising sharply from 1955 to 1965, declining until 1974, then stabilizing in the late 1970s. In summary, the composition o f the budget has undergone a substantial change over the last 25 years. While national defense used to be far and away the most important function o f the federal government, this category has yielded to social programs in the fonn o f benefits for individuals. The sum o f these two large program categories has de clined as a percent o f total outlays from 78.5 percent in 1955 to 70.2 percent in 1980. The slack has been taken up by an increasing proportion o f outlays channeled to state and local government and paid in interest to the public. FEDERAL RESERVE BANK OF ST. LOUIS NOVEMBER 1981 C h a rt 2 Federal Budget O u t l a y s as a Percent of Total O u t l a y s 11 |_L E stim ates fo r 1981-1986 a re a lte rn a tiv e p ro je c tio n s : d a s h e d lines a re CBO b a s e lin e p ro je c tio n s a n d s o lid lines a re a d m in is tra tio n p ro je c tio n s fro m M id-S e ssion Review o f the 1982 B u d g e t (July 1981). [2 O th e r fe d e r a l o p e ra tio n s , plus g ra n ts to state a n d lo c a l g o v e rn m e n ts o th e r th a n b e n e fit paym ents fo r in d iv id u a ls . 19 FEDERAL RESERVE BANK OF ST. LOUIS BASELINE PROJECTIONS OF FED ERAL SPENDING: 1981-86 An evaluation o f spending plans for the future requires a baseline for comparison, namely, a course that federal spending would follow were there no changes in spending policies. A set o f such baseline projections has been prepared by the CBO and is presented here. CBO's Economic Assumptions The projections assume that expenditures are based on spending policies and laws in effect as o f Decem ber 1980. With respect to so-called entitle ment programs (social security, Medicaid, veterans pensions, federal employee retirement, etc.), it is assumed that future spending will respond to eco nomic and demographic changes in the same way as in the past. Since the remaining portion o f federal outlays are discretionary, that is, they depend on annual appro priations, special assumptions are required. The general assumption is that programs in effect in Decem ber 1980 will continue into the future with increases in outlays a reflection o f the rate o f infla tion. Thus, inflation rates are a critical part o f the econom ic assumptions. In the case o f national de fense, CBO’s baseline projections also include an allowance for programmatic changes. Specifically, projections o f outlays are consistent with a defense force and investment program that is implied by congressional action through Decem ber 1980. The future course o f federal outlays thus depends on the underlying econ om ic assumptions. The CBO’ s econom ic assumptions are summarized in table 2. With about 30 percent o f federal spending directly indexed for inflation, the inflation as sumptions bear considerable weight. In addition, the costs o f many other programs are based on the assumption that congressional actions will provide the funding to keep the programs apace with in flation. There are also programs like unem ploy ment compensation and food stamps that depend on the assumptions made about unemployment. Net interest paid depends on interest rate projections as well as future deficits. In general, the CBO’ s economic assumptions are taken as given for the baseline budget projections. There is no allowance for the feedback o f fiscal actions to the economy. For example, CBO notes that the projected surplus in the baseline budget 20 NOVEMBER 1981 would be inconsistent with the underlying e co nomic assumptions. In other words, their econom ic assumptions do not represent the output o f a fullfledged econometric model. CBO's Baseline Projections Chart 1 summarizes the baseline projections as a percent of GNP for 1981-86. The momentum o f total outlays as well as high inflation projections in the near term indicate a slight rise in total outlays in 1981-82 relative to 1980. After 1982, total outlays decline relative to GNP. There are several reasons for this. First, the baseline projections do not provide for real growth in a number o f program areas, while the GNP projections include substantial increases in real GNP. Second, witli unemployment assumed to decline, programs tied to the unemployment rate grow more slowly. Similarly, net interest outlays decline, assuming declines in interest rates and in the deficit.5 The national defense component o f federal out lays is assumed to hold fairly constant under the baseline assumptions, im plying real growth in defense outlays. The major reason for an increase in real terms is an increase in the cost o f strategic forces, namely, the inclusion o f funding for the MX missile and a new manned bomber. Consequently, based on programmatic changes im plied by congressional action through D ecem ber 1980, defense outlays would be 5.2 percent o f GNP in 1986, little changed from 5.3 percent in 1980. Benefit payments for individuals are projected to rise in 1981-82 relative to GNP, but fall slowly from 1983-86. The proportion stays high relative to GNP because so many o f the benefit programs are indexed to inflation. The most important o f these are social security benefits, railroad retirement benefits, sup plemental security income, veterans’ pensions, and civil service retirement benefits. Other programs, like food stamps and child nutrition, unemployment compensation, and Medicare and Medicaid, are in directly tied to inflation. But in addition to keeping pace with inflation, benefit payments for individuals rise faster than the price level because the number o f 5It shou ld b e noted that the C B O ’ s baseline estimate o f net interest d oes not reflect fully the surplus/deficit estim ates im p lied b y their full set of baseline estim ates of reven ues and expenditures. For purposes of their b a selin e spen din g projections, the bu d get was assum ed to b e b a lanced b egin n in g in 1984, and the sub stantial surpluses estim ated after 1984 w ere not assum ed to be a p p lied to reductions in the p u b lic debt. FEDERAL RESERVE BANK OF ST. LOUIS NOVEMBER 1981 Table 2 CBO Economic Assumptions (calendar years) Estimate Actual Gross national product C urrent dollars Am ount (billions of dollars) Percent change, year to year Constant (1972) dollars Am ount (billion s of dollars) Percent change, year to year Deflator (percent change, year to year) Unemploym ent rate Interest rate (91-day Treasury bills) 1980 1981 1982 1983 1984 1985 1986 $2,626 8.8% $2,941 12.0% $3,323 13.0% $3,734 12.4% $4,135 10.8% $4,541 9.8% $4,963 9.3% $1,481 -0 .2 % 9.0% $1,511 2.0% 9.7% $1,572 4.1% 8.6% $1,651 5.0% 7.0% $1,725 4.5% 6.0% $1,797 4.2% 5.4% $1,872 4.2% 4.9% 7.1% 7.5% 7.2% 6.6% 6.4% 5.9% 5.6% 11.4% 13.5% 10.5% 9.4% 8.2% 7.0% 6.0% SOURCE: Congressional Budget Office, Baseline Budget P rojections: Fiscal Years 1982-1986 (July 1981) retirees and disabled persons grows. Baseline projections indicate benefit payments would be 10 percent o f GNP in 1986 compared with 10.6 percent in 1980. Other grants to state and local governments are projected to decline to 1.5 percent o f GNP in 1986 compared with 2.2 percent in 1980. This would be a decline in real terms and is accounted for in part by statutory ceilings on social service grants and gen eral revenue sharing. Furthermore, community de velopment grants are projected to rise little under existing law. Baseline projections for net interest indicate a substantial decline relative to GNP, from 2.0 percent in 1980 to 1.2 percent in 1986. This projection is a direct result o f the econom ic assumption that in terest rates will decline throughout the period and that the budget will be balanced by 1984. The catch-all category o f “ all other” is also pro jected to decline — from 2.4 percent o f GNP in 1980 to 1.7 percent in 1986. Implicit in the projection is that “ all other” outlays will keep pace with inflation though they will show no real growth. The implications o f the baseline projections for the composition o f federal outlays are summarized in chart 2. Benefit payments are slated to stay high, rising som ew hat relative to the total. Benefit payments, 46.8 percent o f total outlays in 1980, are projected to rise to 51.0 percent based on existing law. The future pattern o f national defense relative to total outlays also is one o f steady increase. National defense, which was 23.4 percent o f total outlays in 1980, is projected to rise to 26.7 percent in 1986 assuming the expansion o f strategic forces. With the sum o f benefit payments and national defense rising from 70.2 percent of total outlays in 1980 to 77.7 percent in 1986, there is a substantially lesser proportion o f the total going toward the remaining three categories. These categories share the decline about equally, dropping about 2 or 3 percentage points from 1980 to 1986. ADMINISTRATION PROJECTIONS OF FEDERAL SPENDING: 1981-86 The new administration announced a program o f spending cuts shortly after taking office early this year. These proposals were first presented in March and subsequently revised in the mid-session review o f the budget in July.6 Controversy surrounding these proposals probably will continue as the ad ministration works with Congress in review ing these proposals. For the purposes o f this article, the estimates o f the administration as o f July 15, 1981, are used because they are the most current set o f officially published estimates. The final results will differ from those presented here, but the July pro posals are representative o f the administration’ s plan. ®Office o f M anagem ent and Budget, Fiscal Y ear 1982 Budget R evisions (M arch 1981) and M id-Session R eview o f the 1982 B u dget (July 15, 1981). 21 NOVEMBER 1981 FEDERAL RESERVE BANK OF ST. LOUIS Table 3 Administration Economic Assumptions (calendar years) Actual Gross national product Current dollars Am ount (billions of dollars) Percent change, year to year Constant (1972) dollars Am ount (billions of dollars) Percent change, year to year Deflator (percent change, year to year) Unem ploym ent rate Interest rate (91-day Treasury bills) Estimate 1980 1981 1982 1983 1984 1985 1986 $2,626 8.8% $2,951 12.4% $3,296 11.7% $3,700 12.3% $4,097 10.8% $4,500 9.8% $4,918 9.3% $1,481 -0 .2 % 9.0% $1,519 2.6% 9.6% $1,570 3.4% 8.0% $1,648 5.0% 7.0% $1,721 4.5% 6.0% $1,794 4.2% 5.4% $1,870 4.2% 4.9% 7.1% 7.5% 7.3% 6.6% 6.2% 5.8% 5.5% 11.4% 13.6% 10,5% 7.5% 6.8% 6.0% 5.5% SOURCE: O ffice of Management and Budget, Mid-Session Review of the 1982 Budget (July 1981) Economic Assutnptions Administration Spending Projections As alluded to above, any spending proposals depend on the particular econom ic assumptions that are made. Table 3 summarizes the administration’ s econom ic assumptions as presented in the mid session review of the budget on July 15. These assumptions are similar to CBO’ s in that they are called assumptions, not projections. The adminis tration goes slightly further, however, saying that these assumptions are projections o f the trend o f econom ic performance expected under the adminis tration’ s new policy. So, even though these assump tions do not appear to be based on a consistently estimated econometric model, they do allow for the feedback o f economic policies to the econom ic as sumptions, albeit in an informal way. In general, the administration plans to reduce total outlays relative to GNP from 22.6 percent in 1980 to 18.6 percent in 1986. In comparison, the CBO pro jects total outlays to be 19.6 percent o f GNP in 1986. In 1986 dollars, this difference amounts to $55 billion (see table 4); in other words, the adminis tration is proposing that total outlays be reduced by $55 billion from what they would otherwise be, based on existing law. A comparison o f table 3 with the CBO ’s assump tions in table 2 indicates that the general contours o f the econom ic projections are similar. The course of GNP and its distribution between prices and output is essentially the same for the administration as for the CBO. For example, the difference in the average annual rate o f increase o f nominal GNP over the 1980-86 period is only 0.1 percentage point — 11.1 percent for the administration compared with 11.2 percent for the CBO. The principal difference in the two sets o f econom ic assumptions is with respect to interest rates; the administration assumes Treasury bills will decline to 5.5 percent in 1986 compared with 6.0 percent for the CBO. 22 The major departure o f the administration’s plan from the baseline projections is that it plans a sharp increase in national defense spending even as a percent of GNP. The administration’ s plan calls for defense spending to be 6.7 percent o f GNP in 1986 compared with 5.2 percent for the baseline projec tions. This would be the largest proportion o f GNP diverted to defense since 1972. With total outlays projected to decline relative to GNP and national defense projected to increase, non-defense must decline sharply as a percent o f GNP. The largest component o f non-defense outlays — benefit payments for individuals — is projected at 8.5 percent o f GNP in 1986 compared with a baseline figure o f 10.0 percent. In 1986 dollars, this is a re duction o f $73.1 billion, or 15 percent. Included in the administration’ s reduction in benefit payments is the cutback in social security benefits that generated considerable controversy last spring and subse quently was withdrawn pending the reeommenda- FEDERAL RESERVE BANK OF ST. LOUIS NOVEMBER 1981 Table 4 Federal Spending Projections: Administration vs. Baseline (billions of dollars) 1986 Projected Annual rate of change: 1981-86 Actual 1980 Baseline A dm inistration1 $135.9 $254.0 $325.7 Non-defense Benefit payments for individuals 443.8 271.2 695.9 484.0 569.4 410.9 7.8 10.1 4.2 7.2 Other grants to state and local governm ents 57.2 71.3 42.9 3.7 -4 .7 National defense Baseline 11.0% A dm inistratior 15.7% Net interest 52.5 59.1 60.6 2.0 2.4 Other federal operations 62.9 81.5 55.1 4.4 -2 .2 579.6 949.9 895.1 8.6 7.5 Total 'In th e Mid-Session Review o f the 1982 Budget, total outlays included $42.7 billion of "co ntin gen cie s and additional savings to be proposed.' For purposes of this table, this $42.7 b illio n was distributed proportionally among all of the m ajor program categories. tions o f a new commission.7 Although social security benefits as a percent o f total benefit payments to persons would rise slightly, such benefits as a per cent of GNP would drop from 4.5 percent in 1980 to 4.1 percent in 1986. O f the remaining program categories, only net interest is close to the baseline projection for 1986. As a percent o f GNP, net interest is projected at 1.3 percent o f GNP, essentially the same as the 1.2 per cent baseline projection.8 The other two remaining categories are sharply below the baseline projec tions. All other grants to state and local governments are projected by the administration at 0.9 percent of GNP compared with 1.5 percent for the baseline projection, a cutback o f $28 billion in 1986 dollars, or 7T h e set o f proposals a ffectin g social security in clu d e d the follow in g : (1) R ed u cin g the w elfare-orien ted elem ents that d u plicate other program s; (2) Relating disability benefits m ore closely to a p erson ’ s work history and m edical con ditions; (3) R edu cin g the opportunity for “ w in dfall” benefits; (4) Shifting the effective date for paying the automatic cost-oflivin g increases from July to O ctober; (5) E n couraging workers to stay on the jo b , at least until the traditional retirem ent age o f 65; (6) L ow erin g future replacem en t rates — the initial benefit as com pared with recent preretirem ent earnings. 8E ven though the adm inistration assumes that interest rates w ill b e lo w e r b y 1986, its estimate o f net interest exceed s the baseline projection b eca u se its deficit estimates are larger (or surpluses are smaller) during the p rojection period. T h e reason for this is that the b a selin e reven u e projections are based on tax laws as of D e ce m b e r 1980, w hereas the adm inistration’ s plan in cludes the tax cut. 40 percent, which would bring this category back to 1963 levels. Similarly with the “ all other” category, the administration’s plan calls for all other outlays to be scaled back to 1.1 percent o f GNP compared with 1.7 percent for the baseline projection. This would be a reduction o f $26.4 billion in 1986 dollars, or 32 percent, and would bring this category lower than ever recorded in the 1955-80 period. The scope o f the administration’ s program is dramatized even more when one examines the pro gram categories relative to total outlays. In order to reduce total outlays by $55 billion in 1986 relative to baseline projections and at the same time increase national defense by $72 billion, the administration must reduce non-defense outlays by $126 billion. Furthermore, with an administration net interest projection o f $2 billion greater than the baseline estimate, the remaining categories must be reduced by $128 billion. Given the baseline projection of $637 billion for non-defense excluding net interest, a $128 billion cutback translates into a 20 percent re duction from what they would otherwise be. As a percent o f total outlays, the national defense projection amounts to 36.4 percent compared with 26.7 percent for the baseline projection. To achieve such a goal would give defense an importance in the budget not realized since 1970. The scaling back o f benefit payments would re duce such payments as a percent o f the total to 45.9 percent compared with 50.9 percent for the baseline. 23 FEDERAL RESERVE BANK OF ST. LOUIS NOVEMBER 1981 On this basis, this would be the lowest share for this category o f spending since 1974. non-defense spending, particularly benefit pay ments for individuals. With net interest being projected at 6.8 percent o f total outlays compared with 6.2 percent for the base line, the other reductions come from the two re maining catch-all categories. All other grants to state and local governments would be reduced to 4.8 percent o f the budget compared with 7.5 percent for the baseline. This would be the smallest proportion o f the budget for this category since 1963. In planning for the next five or six years, the administration has set forth proposals designed to alter past trends. One goal is to reduce the overall growth o f federal outlays relative to GNP. According to the administration’s plan, outlays will be 18.6 percent o f GNP in 1986, down from 22.6 percent in 1980, reducing the relative size o f government to its lowest level since 1966. Existing law indicates that outlays are scheduled to be reduced relative to GNP anyway, but to 19.6 percent rather than 18.6 percent. The final “ all other” category, according to ad ministration projections, would be reduced to 6.2 percent o f total outlays compared with an 8.6 percent baseline projection. This would be lower than any other year in the 1955-80 period. SUMMARY AN D CONCLUSIONS The administration has embarked on a budget program designed to reduce the size o f government in the U.S. economy. At the same time, they propose to alter greatly the composition o f the budget. Over the last 25 years, federal outlays grew more rapidly than the nation’ s GNP. At the same time, the com position o f these outlays shifted toward greater 24 Slowing the momentum o f government spending is an ambitious goal, but the way in which the ad ministration proposes to achieve it makes the task all the more difficult. By announcing a goal o f acceler ating defense spending while slowing the growth o f total outlays, the administration implies that sharp cutbacks are required for non-defense spending rel ative to baseline projections. The scaling back o f non-defense programs other than net interest amounts to 20 percent compared with baseline pro jections. Belative to baseline projections, this non defense cutback is distributed as a 15 percent reduction in benefit payments for individuals, 40 percent for other grants to state and local govern ments, and 32 percent for the all other category. The Voluntary Automobile Import Agreement with Japan — More Protectionism C L IF T O N B. L U T T R E L L A HE signing o f “ voluntary” agreements to reduce imports has made considerable headway in recent years. The recent accord to limit automobile imports from Japan is an example o f such an agreement. Ship ments o f Japanese automobiles to the United States in the first year following the agreement (April 1981 through March 1982) are to be held to 1,680,000 cars, compared with 1,820,000 in 1980 — an 8 percent re duction.1 The rationale given for this agreement is similar to that traditionally offered to support protectionist policies.2 For example, economist Marina v. N. Whit man argued that the agreement was necessary to help the auto industry adjust to sharply changed cir cumstances and consumer preferences. In her view, the U.S. automobile industry is similar to an “ infant industry,” one that needs time and massive invest ment to adjust fully to new circumstances.3 C h r is to p h e r C onte and Urban C. L eh ner, “ Car Im port Lim it Eases U.S. - Japan Trade Rift; D om estic Makers Gain L eew ay to Boost Prices,” The W all S treet Journal, M ay 4, 1981. 2For exam ple, see Charles P. K in dleb erger and Peter H. Lindert in In tern ation a l E con om ics, 6th ed. (Richard D. Irw in, Inc., 1978), pp. 130-47. 3W illiam H. Kester, “ E con om ist O utlines Auto W o e s ,” St. Louis P ost-D ispa tch , A pril 8, 1981. T h e infant industry argum ent is ty pica lly u sed to ju stify tem porary tariffs or other protection measures that cu t dow n on im ports from m od em manufacturers w h ile the infant dom estic industry learns h ow to p rod u ce at low en ough costs to co m p ete w ithout the h elp o f protection. “ Voluntary” Import Controls — Who Gains? Who Loses? The purpose o f “ voluntary” trade agreements b e comes clear when one analyzes the recent agreement with Japan. The agreement was made after months o f discussion over the rising volume of Japanese auto mobile sales in the United States. As a consequence o f these rising imports, U.S. legislation had been proposed to limit such imports to 1.6 million ve hicles per year for three years. The proposed legis lation was more stringent than the provisions o f the so-called voluntary agreement. One government spokesman was reported to have demanded “ that the Japanese restrict car sales (to the United States) to between 1.4 and 1.6 million for more than one year.” 4 It is likely that the Japanese participated in the agreement to avoid the imposition o f even more stringent protectionist measures by the United States.5 Some groups in the U.S. auto workers and m obile manufacturers, ment. Presumably the United States, specifically stockholders o f U.S. auto benefited from the agree agreement will lead to an 4Secretary o f State A lexan der H aig as reported in H obart Row an, “ T h e Japanese C ar C harade,” W ash in gton P ost, M ay 7, 1981. 5T h e agreem ent was not voluntary based on the usual sense o f the w ord. T h e action d id not p ro ce e d from the free ch o ice of each party in the absen ce o f co ercion or legal obligation. 25 FEDERAL RESERVE BANK OF ST. LOUIS increase in sales o f U.S.-manufactured cars. H ow ever, the current Japanese voluntary import control represents simply a protectionist action. As such, its impact on national and consumer well-being is no less harmful than that from higher tariffs, import quotas, or other devices designed to curtail foreign competition in the domestic automobile market. NOVEMBER 1981 Table 1 Imported Cars as a Percent of Total United States Passenger Car Retail Sales (thousands) Total sales Thirty Years o f Expanding Trade The major impetus for the protection of American industries from foreign competition has been rising imports in a few industries. These represent, in part, the consequences o f reduced tariffs and other moves toward free international trade that began in the 1950s. These moves followed a period o f highly pro tective tariffs authorized in the Smoot-Hawley Tariff Act o f 1930. With the tariff reductions, trade with other nations began to increase. Exports o f U.S. mer chandise grew, risingannually ata 6.8 percent rate in the 1950s and at an 8 percent rate in the 1960s. Imports o f merchandise rose annually at a 5.0 percent rate during the 1950s and at a 10.4 percent rate during the 1960s. These imports generated increased com petition for some U.S.-produced goods, such as shoes, clothing and steel products. In the 1970s, the volume o f U.S. international trade spurted and other industries began to experi ence competition from imports. During this decade, imports and exports grew at 20 percent annual rates. What Happened to Automobile Imports? Automobiles were a major factor in the accelera tion o f import growth. In the 1970s, the U.S. auto mobile industry began to experience greater com petition from imports, just as the shoe, clothing and steel industries had in the previous decade. In the automobile case, sharply higher gasoline prices, escalating wage rates and mandatory environmental and safety regulations increased the cost o f Ameri can-manufactured automobiles relative to foreignproduced cars. These factors contributed to sales reductions and increased unemployment in the U.S. automobile industry. The almost doubling o f real gasoline prices in 1979 and 1980 led to both a sizable reduction in de mand for larger automobiles manufactured in the United States and a sharp increase in demand for the smaller cars produced by foreign manufacturers. 26 Percent im ported Percent im ported from Japan1 1980 8,979 26.7% 22.2% 1979 10,671 21.9 15.2 1978 11,312 17.7 13.8 1977 11,185 18.6 12.0 1976 10,110 14.8 11.2 1975 8,640 18.4 8.1 1974 8,867 15.9 8.9 1973 11,439 15.4 5.5 1972 10,950 14.8 6.4 1971 10,250 15.3 6.9 1970 8,405 15.4 4.5 1969 9,583 11.7 3.7 1968 9,656 10.7 1.8 1967 8,337 7.2 0.8 'D ata are im ports of assembled cars and are not pre cisely consistent with sales data included in first colum n. SOURCE: MVMA M otor Vehicle Facts and Figures 1981, pp. 13 and 71. Consequently, the sales o f foreign-built cars, espe cially those made in Japan, accelerated (table 1). Total automobile sales leveled o ff in the early 1970s, and the percent imported held fairly stable until 1979, when it jumped sharply, rising from 17.7 percent in 1978 to 21.8 percent in 1979 to 26.7 per cent in 1980. Most o f the increase in imports came from Japan. The Japanese penetration o f the Ameri can market had been rising steadily since 1970, reaching 15.2 percent o f total sales in 1979. These imports then spurted in 1980 to 22.2 percent o f total sales. Contributing to the higher cost o f U.S. automobiles have been the more liberal wage settlements o f the automobile manufacturers since 1970. Prior to 1970, hourly earnings o f production and non-supervisory workers in the manufacture o f motor vehicles and car bodies averaged 30 to 32 percent more than in all manufacturing industries combined. By 1975, how ever, the autom obile w orkers’ hourly earnings exceeded the earnings o f all workers in manufac- FEDERAL RESERVE BANK OF ST. LOUIS NOVEMBER 1981 Table 2 Average Wage Levels of Production Workers in Automobile Manufacturing and All Manufacturing value o f production. They will all have more real goods for consumption and investment. This occurs because trade serves to allocate production to lowercost manufacturers, and final goods to higher-valued uses.7 .. .Rising Imports Lead to Rising Exports,... Average hourly earnings A ll m anufacturing A utom obile m anufacturing Percent earnings of autom obile to all manufacturin g workers 148% 1980 $7.27 $10.78 1975 4.83 6.82 1970 3.35 4.42 132 1965 2.61 3.45 132 1960 2.26 2.91 129 141 SOURCE: U.S. Department of Labor, Em ploym ent and Earnings, United States, 1909-1978, and March 1981. turing by 41 percent. By 1980 this differential had risen to 48 percent (table 2).6 Increased Imports Not Harmful to Economy, . . . Because o f the employment consequences o f ris ing automobile imports, the claim is often made that the U.S. automobile market must not be opened w ide to foreign automobile manufacturers. This, how ever, is a short-sighted view o f the impact that such imports have on the U.S. economy. There are two general conclusions that follow from an un derstanding o f the econom ic consequences o f trad ing among nations. First, foreign trade (imports and exports) in the longer run are neutral with respect to total employment; that is, employment gains in some industries will offset employment losses in other in dustries. Second, all nations participating in trade will experience gains arising from an increase in the 6It co u ld b e argued that such factors as greater productivity or m ore overtim e work can explain the m ore rapid w age growth in autom obile m anufacturing. Increased productiv ity o f auto m o b ile workers (w hich rose at a 1 percent faster rate than pro ductivity in all manufacturing) cou ld accou nt for part o f this rise. H o w e v e r, a utom obile workers and all m anufacturing workers w ork ed essentially the same n um ber o f hours per w eek in 1970 as in 1980. H e n ce, the faster grow th in hourly earnings o f auto m ob ile w orkers does not appear to have arisen from lon ger work w eek s in a u tom obile manufacturing. Changes in U.S. imports o f goods and services are closely associated with changes in exports. Nations sell goods and services to other nations because they wish to import goods or purchase capital assets from them. Either directly or indirectly, U.S. imports o f Japanese automobiles will create income abroad that will be spent on U.S. goods and services or U.S. fi nancial assets. Imports do not cause general unem ployment; they create job opportunities in some industries as part o f the very process by which they reduce others.8 Export industries will increase total sales and employment; industries facing increasing imports (such as automobile manufacturing) will realize reduced sales and employment.9 . . . Hence, Employment Gains Offset Employment Losses Offsetting the observed employment losses in automobile manufacturing are the sizable gains in sales and employment in a number o f other indus tries resulting from the gain in purchasing power abroad and the rising exports. Major employment gains have occurred since 1964-65 in a number o f in dustries as a result o f rising exports. Among those industries with major employment gains from exports are machinery, transport equipment (including auto mobiles), chemicals, and farm products. Exports have risen in these industries both absolutely and relative to domestic production. Exports o f machinery and transport equipment, for example, rose from an an nual average o f $12.5 billion in 1964-65 to $37.5 bil lion (constant dollars) in 1979-80. As a percent o f domestic production, such exports rose from 7.3 per cent in 1964-65 to 15.8 percent in 1979-80 (table 3). ’ F or a discu ssion o f the gains from trade, see Charles P. K indleberger and Peter H. Lindert, In tern ational E co n o m ics (Richard D . Irw in Inc., 1978) chapter 3 and Arm en A. Alchian and W illiam R. A llen , U n iversity E con om ics, 3rd ed. (W ordsw orth P u blishin g C om pany Inc., 1972), chapters 35-37. 8G eoffrey E. W o o d and D ouglas R. M u dd, “ T he R ecen t U.S. Trade D eficit - N o Cause for Panic,” this R eview (April 1978). 9See Clifton B. Luttrell, “ Im ports and Jobs - T h e O b served and the U n observ ed,” this R eview (June 1978). 27 NOVEMBER 1981 FEDERAL RESERVE BANK OF ST. LOUIS Table 3 Gains in Exports for Selected U.S. Industries Exports annual average (1972 prices) 1964-65 (b illion dollars) M achinery and transport equipm ent Chemicals and allied products A griculture $12.5 1979-80 Percent of dom estic production 7.3% (billion dollars) Percent of dom estic production $37.5 15.8% 3.1 6.8 9.2 11.9 8.0 14.1 18.6 25.3 SOURCE: U.S. Departm ent of Com m erce: S ta tistica l A bstra ct o f the U nited States, 1970, 91st ed., pp. 779-80; Survey o f Current Business, March 1981, pp. 12, S-19; Current Ind ustrial Reports, 1958-77, pp. 8 ,1 1 ,1 4 , 24; Current Ind ustrial Reports, 1972-80, pp. 28,31, 34, 43; Econom ic Report o f the President, January 1981, pp. 337,341; U.S. De partm ent of A gricultu re, A g ricu ltu ra l O utlook, January-February, 1981, p. 25. Table 4 Trend in Exports of Selected Major Farm Products 1964-65 Average Wheat Rice (rough equivalent) Corn Soybeans Cotton 1979-80 Average P rod uctio n1 Exports1 Percent exported P rod uctio n1 Exports1 Percent exported 1,314 75 3,784 733 15.0 796 47 629 228 3.5 60.1% 62.7 16.6 29.5 23.3 2,252 139 7,293 2,042 12.8 1,450 90 2,283 817 7.6 64.4% 65.0 31.3 40.0 59.4 ’ M illion bushels of wheat, corn and soybeans, m illion cwt. of rice and m illion bales of cotton. SOURCE: U.S. Department of A gricultu re, A g ricu ltu ra l O utlook, May 1981, and A g ricu Itu ra l S ta tis tic s , 1968. Exports in the agricultural sector have achieved even greater gains relative to production than in the machinery and transport equipment sector. Exports o f farm products rose from an average $8.0 billion per year in 1964-65 to $18.6 billion in 1979-80. As a percent o f domestic production, such exports rose from 14.1 percent in 1964-65 to 25.3 percent in 1979-80. The impact o f trade on five selected categories o f farm products is shown in table 4. In the case o f two groups, wheat and rice, almost two-thirds o f the crop is exported. Furthermore, major gains in exports since 1964-65 have occurred both in real terms and as a 28 percent o f production for corn, soybeans and cotton. Estimated employment gains attributed to export increases in the three selected industries with rapid increases in exports are shown in table 5. More than 130,000 workers in agriculture alone were required to produce the increased quantity of farm products exported in 1979-80, and over one million were required to produce that portion o f farm production which was exported (table 6). Export gains in machinery and transport equipment accounted for over 650,000 employees. Altogether, rising exports in these industrial groups — machinery and trans port equipm ent, chem icals, and agriculture — required almost one million additional workers. FEDERAL RESERVE BANK OF ST. LOUIS NOVEMBER 1981 Table 5 Number of Employees Attributed to Rising Exports in Selected Industries (thousands) Employees 1964-65 Employees 1979-80 Required fo r exports' Total Required fo r exports' Increase in employees required fo r exports Total Machinery and transport equipm ent Chem icals and allied products A griculture 5,043.9 368.2 6,598.0 1,042.5 674.3 893.2 60.7 1,112.0 132.3 71.6 5,860.0 826.3 3,782.0 956.8 130.5 2,131.6 876.4 TOTAL 1,255.2 'Based on percent of dom estic production exported (see table 1). SOURCE: Table 3 and U.S. Department of Commerce, Survey o f Current Business, March, 1981, p. S-12; E conom ic Report o f the President, January 1981, p. 339; U.S. Departm ent of Labor, Handbook o f Labor S tatistics, December 1980, p. 152. Virtually Everyone Loses From Import Restrictions Just as rising imports were a major factor in the expanding market for farm products, so a reduction in foreign imports will contribute to a reduction in exports o f U.S.-produced goods. For example, in 1980-81, exports to Japan alone accounted for 7 percent o f all U.S. coarse grain production (com, grain sorghum, barley and oats), 10 percent o f soy bean production, 5 percent o f wheat production and more than 12 percent o f cotton production.10 A reduction in Japanese earnings on automobile ex ports will reduce their demand for these products.11 O f course, in U.S. exports to Japan, the loss will not equal exactly the dollar amount o f the reduction in 10U.S. D epartm ent o f Agriculture, F oreign A gricultu ral Trade (M ay-June 1981), p. 10 a nd A gricu ltu ral O u tlook (June 1981), pp . 34-35. 1'A reduction in im ports o f autom obiles from Japan tends to red u ce the supply o f a u tom obiles on the Am erican market (shifts the su p ply cu rve to the left). H en ce, the price o f auto m ob iles w ill b e h igh er and a utom obile manufacturers in W est G erm any and other nations w ill ten d to export m ore cars to the U nited States. Thus, the d eclin e in dollar earnings b y the Japa n ese w ill be partially offset b y an increase in dollar earnings in other nations, and part o f this increase w ill b e u sed to purchase U.S. farm products. Such bilateral com parisons overstate the decline in autom obile im ports from all nations com bined, and, thus, overstate the foreign earnings losses and farm exp ort losses resulting from the restrictions. Table 6 Employment for Farm Commodity Export Sales Farm com m odity sales Employment on farms T o ta l' Percent exported T otal2 For exports2 1980 140.3 29.4% 4,152 1,220.7 1979 131.5 26.4 4,375 1,155.0 1978 1977 112.5 4,343 1,133.5 95.8 26.1 25.4 4,152 1,054.6 1976 94.8 24.7 4,374 1,080.4 1975 88.2 25.2 4,342 1,094.2 1974 92.4 24.2 4,389 1,062.1 897.8 1973 87.1 20.7 4,337 1972 61.2 15.5 4,373 677.8 1971 52.9 14.7 4,436 652.1 1970 50.5 14.7 4,523 664.9 1969 48.2 12.7 4,596 583.7 'B illio n s of dollars, th o u s a n d s of workers. SOURCE: U.S. Departm ent of A griculture, A g ricu ltu ra l Sta tistics 1980, p. 460; Econom ic Report o f the Presi dent, pp. 289, 296; Econom ic Indicators, March 1981, p. 11; FATUS, February 1979 and JanuaryFebruary 1981. 29 FEDERAL RESERVE BANK OF ST. LOUIS Japanese automobile sales in the United States. Japan will be able to purchase U.S. products with her earnings from exports to Western Europe or other nations. However, to put the potential losses in perspective, from 1969 to 1979, U.S. exports to Japan totaled about 80 percent o f U.S. imports from Japan. Moreover, Japanese imports from the United States increased at approximately the same rate as exports to the United States. Thus, all industries with net exports to the Japanese — especially the farm sector — will suffer losses. The greatest losses from protectionism, however, are not those employment and export losses experi enced by specific industries. The greatest losses occur in the reduction in real goods available to both nations for consumption and investment. With trade restrictions there will be fewer auto mobiles available for consumers in the United States, and consumers will pay a higher price for each car purchased. Similarly, there will be a smaller quantity o f farm products available for the Japanese, and food prices will be higher there. While food and farm commodity prices in the United States will be slightly lower as a result o f the restrictions, consumers would prefer the larger number o f automobiles and smaller quan tity o f farm commodities. Otherwise, the prior trade pattern would not have been profitable. The gains from trading occur because the Japanese have a comparative advantage in the production o f smaller automobiles relative to the United States, while we have a comparative advantage in the pro duction o f other goods, such as farm products. With each nation specializing in the production o f those goods in which it has a comparative advantage and ex changing these goods with other nations, all nations will benefit. Hence, the real gains from trade are the 30 NOVEMBER 1981 greater output and wealth that occur through greater specialization and exchange. These gains will not be fully realized if protectionist policies are adopted. SUMMARY The Japanese “ voluntary” autom obile import agreement is not only involuntary, but represents another form o f protectionism. Like all such mea sures, it is predicated on specious logic and faulty economics. Although the major impetus for this agreement has been the observed decline in employment in the U.S. automobile industry, evidence suggests that trade among nations has no impact on total domestic employment over the long run. Rising employment in industries with rising exports will offset em ploy ment losses in those industries that experience increased competition from imports. Reduced em ployment has occurred in some U.S. industries due to increased imports, but the decline has been offset by employment increases in other industries such as agriculture, machinery and transport equipment, and chemicals, where sharp increases in exports were realized. The greatest loss from such agreements, however, is the reduced wealth and w ell-being o f the pop ulation at large. These losses occur because trade is productive. Each nation gains by specializing in the production o f those goods in which it has a compara tive advantage and by exchanging these for other goods produced at lower cost elsewhere. This results in more wealth for all nations. Protectionist policies reduce these gains, and, consequently, reduce the wealth o f U.S. citizens as well.