JAMES L. PIERCE Staff, Board of Governorsof the Federal Reserve System Interest Rates and Their Prospect in the Recovery MUCH CONCERN has been expressedin the financialpress and by otherob- serversabout the prospectsfor interestrates in 1975.The particularfear is that the sharpdeclines in short-terminterest rates in 1974will be followed by sharpincreases in both short- and long-termrates in 1975 under the pressureof the massivefederal deficits expected in 1975 and 1976. This paperaddresses two questions:First, has the movementin short- andlong-term interest rates since mid-1974 been unusualin light of the slow growthof money and the collapsein economicactivity? Second, will the large volumeof deficitfinancing, induced in part by the tax cuts, lend a strongupward push on interestrates in 1975?These can be two aspectsof the samequestion, because an unexpectedand fundamentalshift in the re- lationshipsamong interest rates, income, and moneymay cloud the impli- cationsof the deficitfor interestrates. The RecentBehavior of InterestRates andMoney Demand As figure1 demonstrates,short-term rates peaked in the summerof 1974, andhave since fallen steeply until quite recently. Long-term rates have also Note: The views expressedin this paper are my own and do not necessarilyagree withthose of the Board of Governors.I want to thank membersof the Brookings panel for their many constructivecomments on an earlierversion of this paper. 89 90 Brookings Papers on Economic Activity, 1:1975 Figure 1. Selected Interest Rates, Monthly Averages, January 1974-March 1975 Percent 12 90-119-day commercial paper 10 Aaa utility bondsa % /~~~~~~~~~~~~~ 7 ' 3-month Treasury billsb 6 J F M A M J J A S 0 N D J F M 1974 1975 Sources: Federal Reserve Bulletin, vol. 61 (April 1975), pp. A 27, A 28; and two preceding issues. a. Recently offered series. b. Market yield. fallen from their peaks, but much more gradually; since November they have moved without a clear trend. These moveiiients were accompanied by an appreciable slowing in the growth of nominal income during 1974, which, however, remained positive until the first quarter of 1975. Real in- come fell throughout 1974, moving into a severe decline in the last quarter of 1974 and the first quarter of 1975. While the money stock (MI) grew fairly rapidly in the first half of 1974, after June its growth first slowed and then behaved erratically; on balance, it averaged only 1.4 percent (annual rate) from June through January, before speeding up during February and March. In light of these patterns, was the large decline in short-term interest rates and the much smaller decline in long-term rates to be expected? Was the erratic pattern of growth in Ml predictable? The answer to the first question is a qualified "yes" and to the second question a qualified "no." These answers arise from an examination of the residuals from regression James L. Pierce 91 equationsin the SMP(Social Science Research Council-M.I.T.-University of Pennsylvania)model that describethe behaviorof money demandand interestrates. If these equationsexhibit no unusualbehavior in theirpre- diction errors,one would infer that the behaviorof money and interest ratesduring the recessionwas to be expected.If, on the otherhand, the pre- dictionerrors are unusually large, one wouldconclude that recentfinancial developmentswere unexpected. The single-periodprediction errors from the money-demand(currency plus demanddeposit) equations in the SMP model for 1973 and 1974 are shownin figure2. Theequations used to formthe predictionsof M1 wereas follows: (1) ln MC = 0.22 ln PCE - 0.005 ln RTB + 0.88 ln MC- Standarderror = 0.003; sample period: 1955:4-1971:4. (2) ln -0.28ln MD1-0.06ln RTB-0.12ln RSD y y + 0.08 In RDIS -0.34 In N-0.05' RDIS ~ N Standarderror = 0.0068; sample period: 1955:2-1972:4. where MC = currencyheld by the nonbankpublic PCE = personal consumption expenditures RTB = interestrate on Treasurybills MD = demanddeposits held by the nonbankpublic Y= GNP RSD = interest rate on savings deposits RDIS = discountrate y/N = GNP per capitain 1958 dollars. Actual values of the exogenous variablesin these equations-real and nominalGNP and short-terminterest rates-were used to formthe predic- tions. As figure2 clearlyreveals, the equationsseriously overstated M1 growthin the secondhalf of 1974.1What is more, the errorswere several timesgreater than those obtainedin earlierperiods.2 1. All the equation predictionsin this paper are single-periodpredictions that have set the rho termequal to zero. As a result,the equationsexhibit serially correlated errors. When actual predictionsare made, there is less concern in distinguishingthe role of structuralvariables from informationon the errorstructure and an estimatedvalue of rho is used. 2. As will be shown below, the predictionof the demand for demand deposits was responsiblefor the large errors;the currencypredictions were quite accurate. 92 Brookings Papers on Economic Activity, 1:1975 Figure2. Actualand Predicted M1, Quarterly,1973 and 1974 Billions of dollars 300 290 Predicted - 280280 - Actual 270 260 - 250 I II III IV I II III lV 1973 1974 Sources: Actual-SMP data bank; predicted-SMP model, using money-demand equations (1) and (2) given in the text. Mi = currency plus demand deposits. Theerrors in the money-demandrelationships give someidea of whether the declinein interestrates in 1974was smallerthan would have been ex- pectedon the basis of past relationships.The equationsimply that-given the actualvalues of other exogenousvariables-Treasury bill rates of 8.8 percentand 8.3 percentin the third and fourth quarterswould have pro- duceda 6 percentincrease in M1at an annualrate. Actualbill ratesin the two quarterswere 8.2 and 7.4 percent,respectively, which should have yieldeda greatermoney demand;yet actualM1 growthwas only 2.0 and 2.2 percentin those quarters.3Thus, according to the money-demandequa- tions,actual bill rateswere sufficiently low to producean M1growth in ex- cess of 6 percent. Treatingthe errorsin the equationsas reductionsin the interceptof the money-demandequations-that is, treatingthem as if the money-demand equation"shifted" downward-makes it possibleto calculatethe interest 3. The quarterlyM1 figures used in the equations are averages of the two months surroundingthe end of the quarter;for example,the figurefor the fourth quarteris the averageof Decemberand January. James L. Pierce 93 ratesrequired to achievea 6 percentM1 growth. To do this, the intercepts of the currencyand demand-depositequations in the third and fourth quarterswere adjusted by each equation'serror in each quarter.With the adjustedequations, the Treasurybill ratesrequired to obtain6 percentM1 growthin the thirdand fourthquarters were 6.2 percentand 4.1 percent, respectively. Thelarge errors in predictingmoney demand are sufficient but not neces- saryto explainthe slowgrowth in M1.The demandfor moneyprovides pre- dictionsof themoney stock given short-term interest rates. With the interest ratesthat prevailedin the third and fourth quarters,the money-demand functionpredicted M1 growth higher than the actual. However,money growthneed not havebeen as slow as it was.If bankreserves had been sup- plied at a rate consistentwith more rapid growth in M1, interestrates would have fallen sufficientlyto equate money demand with the more rapidlygrowing supply and the money stock would have expandedmore rapidly.Thus, the issueis not that the declinein interestrates was so large butrather that it was not largeenough to spurmore rapid expansion in M1. Giventhe actualdecline in short-terminterest rates, the behaviorof long- termrates was not surprising.The model'sequation for the interestrate on newlyissued corporate bonds, taken as a measureof long-termrates, is 18 18 APCONQ+3~C (3) RNI = 0.76 + , b,RCP_j + E C PCON1 + 033 aRCP bo = 0.18, bi = 0.94; co = 4.83, i = 31.18, Standarderror = 0.17; sample period: 1954:4-1971:2. where RNI = interestrate on newlyissued Aaa utilitybonds RCP = commercial paper rate PCON = consumptionprices (percent change measured at annualrate) aRCP= a measureof the standarddeviation of RCP overthe previous eight quarters. The equationdid overpredictthe rate on new issuesin late 1973and 1974 by a substantialamount, the maximumerror being 1.4 percentagepoints in the first quarterof 1974. But the patternof predictionscaptured the up- surgein long-termrates in 1973and 1974and the moderatedecline in late 1974.The errorsmade duringthe periodappear to be explainable.In the equationa distributedlag on the percentagechange in consumptionprices 94 Brookings Papers on Economic Activity, 1:1975 servesas a proxy for the expectedrate of inflation.The behaviorof this pricevariable was not an appropriatemeasure of expectedinflation because of the rapidrise in consumptionprices relative to other pricesduring the period.This variableprobably accounts for much of the overpredictionof the long-termrate in 1973and 1974. MONEY DEMAND While the declinesin interestrates may not be surprisinglylarge, the sizableerrors in predictingmoney demand in the secondhalf of 1974remain to be explained.The searchfor explanationsis frustratingand perhaps fruitless,particularly because the errorsmay have resultedfrom largeran- dom disturbances.An effortto "explain"the errorsis alwaysan exercisein ex post theorizing,which is often difficultto distinguishfrom pure ra- tionalization.If the ex post argumentsappear to be compelling,it is crucial to test the role of thesefactors not only in the currentsituation but also at othertimes when they