What Ended the Great Depression

8/3/2019 What Ended the Great Depression

1/28

THE JOURNAL OF ECONOMIC HISTORYVOLUME 52 DECEMBER1992 NUMBER 4

What Ended the Great Depression?CHRISTINA D. ROMER

This paperexamines the role of aggregate-demandtimulus n endingthe GreatDepression. Plausible estimates of the effects of fiscal and monetary changesindicatethatnearlyall the observedrecovery of the U.S. economy priorto 1942was due to monetary expansion. A huge gold inflow in the mid- and late 1930sswelled the money stock and stimulated he economy by lowering real interestrates and encouragingnvestmentspendingand purchasesof durablegoods. Thatmonetarydevelopmentswere crucialto the recovery impliesthat self-correctionplayedlittle role in the growthof real output between 1933 and 1942.

Between 1933 and 1937real GNP in the United States grew at anaverage rate of over 8 percentper year; between 1938and,1941itgrew over 10 percent per year. These rates of growth are spectacular,even for an economy pulling out of a severe depression. Yet therecovery from the collapse of 1929 to 1933 has received little of theattention that economists have lavished on the Great Depression.Perhapsbecause the cataclysmof the early 1930swas so severe, modemeconomists have focused on the causes of the downturnand of theturningpoint in 1933. Once the end of the precipitousdecline in outputhas been explained, there has been a tendency to let the story drop.1

The eventual returnto full employment s simply characterizedas slowand incompleteuntil the outbreakof WorldWarII.In this articleI examinein detail the source of the recovery from theGreatDepression. I arguethat the rapidrates of growthof real outputin the mid- and late 1930swere largelydue to conventionalaggregate-demand stimulus, primarilyin the form of monetary expansion. Mycalculations suggest that in the absence of these stimuli the economyThe Journal of Economic History, Vol. 52, No. 4 (Dec. 1992). ? The Economic History

Association.All rights reserved.ISSN 0022-0507.The author s Associate Professorof Economics,Universityof California,Berkeley,Berkeley,CA 94720.MichaelBernstein, BarryEichengreen,RobertGordon,RichardGrossman,FredericMishkin,David Romer,Peter Temin,ThomasWeiss,David Wilcox,and two anonymousreferees providedextremelyhelpfulcommentsand suggestions.The researchwas supportedby theNationalScienceFoundationand the AlfredP. Sloan Foundation.1 Temin and Wigmore, "End of One Big Deflation," for example, provided a convincingexplanationor the turningpointin 1933but did not analyzethe process of recoveryafter 1934.Anotable exception to this usual pattern is Bernstein, Great Depression, which analyzed theimportanceof structural hangesthroughouthe recovery period.

757


2/28

758 Romerwould have remaineddepressed far longer and far more deeply than itactuallydid. This in turn suggests that any self-correctingresponse ofthe U.S. economy to low outputwas weak or nonexistentin the 1930s.The possibilitythat aggregate-demandtimulus was the source of therecovery from the Depression has been considered and discounted bymany studies. E. Cary Brown, for example, used a conventionalKeynesian multipliermodel and the concept of discretionary govern-ment spendingto arguethat fiscal policy was unimportant.His often-cited conclusion was that "fiscal policy ... seems to have been anunsuccessful recovery device in the 'thirties-not because it did notwork, but because it was not tried."2 Milton Friedman and AnnaSchwartzstressed that FederalReservepolicy was not the source of therecovery either: "In the period under consideration [1933-1941], theFederal Reserve System made essentially no attempt to alter thequantity of high-poweredmoney."3While they were clearly aware thatother developments led to a rise in the money supply during themid-1930s,FriedmanandSchwartzappear o have been moreinterestedin the role that FederalReserve inaction playedin causingandprolong-ing the GreatDepressionthanthey were in quantifying he importanceof monetary expansion in generatingrecovery.The emphasis that these early studies placed on policy inaction andineffectiveness may have led the authors of more recent studies toassume that conventional aggregate-demand timulus could not haveinfluencedthe recovery from the GreatDepression. Ben Bernanke andMartin Parkinson, for example, analyzed the apparent reversion ofemployment toward its trend level in the 1930sand were struckby thestrengthof the recovery. They believed, however, that "the New Dealis better characterized as having 'cleared the way' for a naturalrecovery ... rather than as being the engine of recovery itself."4 As aresult, they argued that the trendreversion of the interwareconomy isevidence of a strong self-corrective force. J. Bradford De Long andLawrence Summerssoundeda similar heme: "the substantialdegree ofmean reversion by 1941 is evidence that shocks to outputare transito-ry." The only aggregate-demandtimulusthatthey thought mighthavecontributed o the recovery was World WarII, andthey concluded that"it is hardto attributeany of the pre-1942catch-upof the economy tothe war,"5Despite this conventional wisdom, there is cause to believe thataggregate-demanddevelopments, particularlymonetarychanges, wereimportantin fostering the recovery from the Great Depression. Thatcause is the simple but often neglected fact that the money supply

2 Brown, "Fiscal Policy," pp. 863-66.3Friedman and Schwartz,MonetaryHistory, p. 511.4 BernankeandParkinson,"Unemployment, nflation,and Wages,"p. 212.5 De Long and Summers,"How Does MacroeconomicPolicy?" p. 467.


3/28

Ending of the Great Depression 759(measuredas MI) grew at an averagerate of nearly 10 percentper yearbetween 1933and 1937, and at an even higherrate in the early 1940s.Such largeandpersistentrates of moneygrowthwere unprecedented nU.S. economic history. The simulationsI present in this paper usingpolicy multipliersbased on the experiences of 1921and 1938, as well asmultipliersderived from macroeconometricmodels, suggest that thesemonetarychanges were cruciallyimportant o the recovery. Accordingto my calculations, real GNP would have been approximately 25percentlowerin 1937and nearly50 percentlowerin 1942 hanit actuallywas if the money supplyhad continuedto grow at its historicalaveragerate. Similar simulationsfor fiscal policy suggest that changes in thegovernmentbudgetsurplusplayedlittlerole in generating he recovery.In addition to estimating the effects of the tremendous monetaryexpansionduringthe mid- and late 1930s,I also examine the source ofthis expansion and the transmissionmechanismthat operatedbetweenthe monetarychangesand the realeconomy. The increasein the moneysupply was primarilydue to a gold inflow, which was in turn due todevaluation n 1933andto capitalflight romEuropebecause of politicalinstability after 1934. My estimates of the ex ante real interest ratesuggest that, coincident with this gold inflow, real interest rates fellprecipitouslyin 1933and remained ow or negativethroughoutmost ofthe second half of the 1930s.These low real interest rates are closelycorrelatedwith a strongrebound n interest-sensitivespending.Thus, itis plausible that expansionarymonetary developments were workingthrougha conventional interest-rate ransmissionmechanism.

THE STRENGTHOF THE RECOVERYMy concern in this articlewith finding he source of the highrates ofreal growth duringthe recovery from the GreatDepression may seemstrangeto those accustomed to thinkingof that recovery as slow. Theconventionalwisdom is that the U.S. economy remaineddepressedforall of the 1930s and only returned to full employment following theoutbreak of WorldWar II. The reconciliationof these two seeminglydisparateviews lies in the fact thatthe declinesin realoutput n the early1930s, and again in 1938, were so large that it took many years ofunprecedentedgrowth to undo them and returnreal output to normallevels.For most of my analysis I examined annualestimates of real GNPfrom the U.S. Bureau of Economic Analysis.6 Because this seriesbeginsat 1929,1 extendedit backward ntime, whennecessary, withmyrevised version of the Kendrick-KuznetsGNP series.7 The percentagechanges in real GNP shown in Figure 1 clearly demonstrate both the6 U.S. Bureauof EconomicAnalysis,National Income and Product Accounts, table 1.2, p. 6.7 Romer, "WorldWarI," table 5, p. 104.


4/28

760 Romer20 1I|15 -I05-

0

-5--10-IS119Z7 1931 1935 1939

FIGURE 1PERCENTAGECHANGESIN REAL GROSSNATIONALPRODUCT, 1927-1942

Sources:The datafor 1929-1942 re from he U.S. Bureauof EconomicAnalysis,National Incomeand ProductAccounts, table 1.2, p. 6. The data for 1927-1928are fromRomer,"WorldWarI,"table5, p. 104.

severity of the collapse of real outputbetween 1929and 1933and thestrengthof the subsequentrecovery. Between 1929and 1933,real GNPdeclined 35 percent;between 1933and 1937, it rose 33 percent. In 1938the economy suffered another5 percent decrease in real GNP, but thiswas followed by an even more spectacular increase of 49 percentbetween 1938and 1942.By almostany standard, he growth of realGNPin the four-year periods before and after 1938was spectacular.It is certainly the case, however, that despite this rapid growth,outputremainedsubstantiallybelow normaluntil about 1942.A simpleway to estimate trendoutputfor the 1930sis to extrapolatethe averageannualgrowth rate of real GNP between 1923and 1927 forwardfrom1927.The years 1923through1927 were chosen for estimatingnormalgrowth because they are the four most normalyears of the 1920s;thisperiod excludes the recession and recovery of the early 1920s and theboom in 1928 and 1929. This was also a period of price stability,suggestingthatoutputwas neitherabnormallyhighnorabnormallyow.The resultingfigurefor normalannualreal GNP growthis 3.15 percent.Figure2 shows the log value of actual real GNP and trend GNP basedon this definitionof normalgrowth. The graph shows that GNP wasabout 38 percentbelow its trend level in 1935and26 percentbelow it in1937.Only in 1942did GNP return to trend.The behavior of unemployment duringthe recovery from the GreatDepression is roughly consistent with the behavior of real GNP.Although many scholars have rightly emphasized that the unemploy-ment rate was still nearly 10percent as late as 1941,it had fallen quite


5/28

Ending of the Great Depression 7617.0r''''''''l'll

6.8 -

E 6.6 --c

62-

1919 1923 1927 1931 1935 1939FIGuRE 2

ACTUAL AND TREND REALGROSSNATIONALPRODUCT,1919-1942Note: TrendGNP, whichis shown by the dashed line, is calculatedby extrapolatinghe growthrateof real GNP between 1923and 1927 orward rom 1927. Therefore,this series does not startuntil 1927.Source:The sourcefor realGNP is the same as in Figure1.

rapidlyfromits highof 25 percentin 1933.8It declined, for example, bymore than three percentage points in both 1934 and 1936. That fullemploymentwas not reachedagain until 1942 s consistent with the factthat real output remained significantlybelow trenduntil that year.THE EFFECTS OF AGGREGATE-DEMAND STIMULUS IN THE

RECOVERYTo examinewhetheraggregate-demandtimuluscan explainthe highrates of realgrowthduring he recoveryphase of the GreatDepression,I performed an illustrative calculation. Consider decomposing thedeviation of output growth from normal into the effect of laggeddeviations of monetaryandfiscalchangesfrom normalandthe effect ofall other factors that mightinfluencereal growth, so that

output change, = 83m(monetaryhange),_1 + Pf(fiscal change),_1 + Et (1)where f,,mand f3P re the multipliers or monetaryand fiscal policy and(, is a residual term that includes such things as supply shocks andchanges in animalspirits.This residualtermalso includesany tendencythatthe economy mighthave to right tself followinga recession. Usingannualdata, this decompositionis most likely to hold with a one-year

The unemploymentstatistics are from Lebergott, Manpower,table A-3, p. 512. Darby,"Three-and-a-HalfMillion,"argued hat the returnof unemploymento its full employmentevelwas significantlymore rapid if one counts workerson public worksjobs as employed. Margo,"InterwarUnemployment,"concluded romananalysisof the 1940censusdata thatat least someof Darby'scorrectionwas warranted.


6/28

762 Romerlag between policy changesandoutputchanges because policy changesdo not immediatelyaffect real output.

Withinthis framework, f one measuresPm,38f, outputdeviations,andpolicy changes, it is possible to calculatewhatthe residualtermmust bein any given year. Since these yearlyresidual ermsreflect all the factorsaffecting growth other than policy, they show how fast the economywould have grown (relativeto normal)hadmonetaryandfiscal changesnot occurred.A comparisonof the actual pathof real output with whatoutputwouldhave been in the absenceof policy changesprovidesa wayof quantifying he importanceof policy.To apply this decomposition to the recovery phase of the GreatDepression, I used as the measure of outputchangethe deviation of thegrowth rateof real GNP fromits average annualgrowthrateduringtheyears 1923through 1927. For the monetarypolicy variableI used thedeviation of the annual (December to December) growth rate of Mlfrom its normal growth rate, where normal is again defined as theaverageannualgrowth rate between 1923and 1927.9The averageannualgrowth rate of Ml over this period was 2.88 percent. For the fiscalpolicy variable I used the annualchange in the ratio of the real federalsurplusto real GNP.'0 This measureof fiscal policy assumes that thenormalchange in the real federal surplus is zero.'1

9 The dataon MI are fromFriedman ndSchwartz,MonetaryHistory, tableA-i, column7, pp.704-34.An alternativemeasureof monetarypolicythat mightbe considered s the deviationof realmoney growthfromnormal.However, changesin nominalmoney are what shift the aggregate-demandfunction; changes in real money result from the interactionof aggregate-demand ndaggregate-supplymovements. Since the purpose of this paper is to isolate the effects ofaggregate-demandtimulus, t is appropriateo use a measureof monetarypolicythatonly reflectschangesin demand.10 The surplusdata are from the U.S. Department f the Treasury,StatisticalAppendix, able 2,pp. 4-11, and are based on the administrative udget.Because these data are for fiscal years, Iconverted them to a calendar-yearbasis by averaging he observations or a given year and thesubsequent year. The datawere deflatedusingthe implicitprice deflator or GNP. The deflatorseries andthe realGNP series for 1929to 1942are from the U.S. Bureauof EconomicAnalysis,NationalIncome and ProductAccounts;data for 1919to 1928are fromRomer,"WorldWarI."I used the administrative udgetdatainsteadof the NIPA surplusdata becausethey areavailableon a consistent basis fortheentire nterwar ra. Whilethe two surplusseries differsubstantiallynsomeyears, the gross movements n the series aregenerallysimilar.I dividedthe surplusby GNPto scale the variablerelativeto the economy." In place of the actualsurplus-to-GNPatio,the full-employmenturplus-to-GNPatio couldbe used. I did not use this variablebecauseit treats a decline in revenues caused by a declineinincome as normalrather hanas anactivistpolicy.Thisis inappropriateor theprewarand nterwareras,whenraising axes in recessions was usuallypreferredo letting hebudgetslipseriously ntodeficit.However, the differencesbetween thefull-employmenturplusand the actualsurpluswereso small even in the worst years of the Depression that the two measuresyield similarresults.Anotherpossiblemeasureof fiscalpolicy is theweightedsurplus,which takesinto accountthefactthata surpluscaused by changes n taxes andtransferswillhavea differentmpactthan a surpluscaused by a change in governmentpurchases.Blinder and Solow, "AnalyticalFoundations,"showed that the practicaleffects of such weightingare typically small and sensitive to modelspecificationand the time horizon considered.


7/28

Ending of the Great Depression 763Estimates of the Policy Multipliers

Deriving the policy multipliers to use in the decomposition is a farmore difficult ask than measuring he deviationof monetaryand fiscalpolicy from normal. One way of deriving the multipliers is to takeestimatesfrom a largepostwarmacroeconomicmodel. Another strategyis to simply posit reasonablevalues for these multipliers.In my laterdiscussionof robustness,I show the resultsof bothof these approaches.However, an alternative procedure that is more in the spirit of theexercise is to use historicalevidence to identifycertainyears when theresidualtermin equation1was smalland when the changesin monetaryand fiscal policy were independentof movementsin realoutput.If therewere two such episodes, one can simply infer estimates of gm and offromthe decompositionitself.12The recessions of 1921 and 1938 are arguably two such crucialepisodes. In both cases there were largemovementsin real outputthathave been almost universally ascribed to monetary and fiscal policydecisions. Friedmanand Schwartz, for example, stated that "in bothcases, the subsequentdecline in the money stock was associated with asevere economic decline."'3 This emphasison monetaryfactorsin 1921and 1938 was echoed by W. Arthur Lewis and by Kenneth Roose.'4Otherauthorsassigned a much more importantrole to fiscal policy asthe source of these two interwar downturns. Alvin Hansen, ArthurSmithies, Leonard Ayres, and Robert A. Gordon all attributedtherecession of 1938to the decline in governmentspending.'5 Gordonalsoarguedthat the decline in governmentspendingafterWorld WarI andthe increase in the discountrate were the two factors that helped to tipa vulnerable economy into a severe recession in 1920.16Furthermore,most alternativeexplanationsthathave been advancedforthese two recessions areeasily disproved; here is little evidencethatother factors (the e, in equation 1) were important n determiningthebehaviorof real outputin 1921and 1938.For example, one explanationfor the downturnin 1938 is that increases in wages due to increasedunionization decreased outputand investment;in short, that there wasan adverse supplyshock in 1937. 7 An adverse supply shock, however,should have been accompanied by rising prices. This did not occur:between 1937and 1938producer prices fell 9.4 percent. On the otherhand, the policy hypotheses that stress a fall in aggregate demandare

12 This methodof derivingroughestimatesof the effectsof policyis an exampleof the narrativeapproachdescribed n Romerand Romer, "Does MonetaryPolicy Matter?"13 Friedmanand Schwartz,MonetaryHistory,p. 678.14 Lewis, Economic Survey,pp. 19-20;and Roose, Economicsof Recession, p. 239.15 Hansen, Full Recovery; Smithies, "American Economy"; Ayres, TurningPoints; andGordon,EconomicInstability.16 Gordon,EconomicInstability,p. 20.17 See, for example, Roose, Economics of Recession, p. 239.


8/28

764 Romerconsistent with the observedfall in prices. The monetary explanation salso consistent with the fact that interestrates rose sharply n early 1937and interest-sensitive spendingsuch as constructionexpendituresplum-meted in late 1937.The mainalternativeexplanationadvanced for the recession of 1921is that the tremendous pent-up demand for consumer goods thatdeveloped duringand after WorldWar I was satisfiedby 1920 and firmsfaced a dramaticdecline in sales.'8 The problemwith this story is thatreal consumer expendituresrose 4.8 percentbetween 1919and 1920and6.2 percentbetween 1920 and 1921.19Any spending story also conflictswith the fact that interest rates rose substantially n 1920.One partialexplanationfor the behavior of real outputin 1921 that ishard to dismiss is the occurrence of a positive supply shock. In aprevious article I argued that the recovery of agriculturalproduction nEurope caused prices of agriculturalgoods in the United States toplummet n 1920.20This, in turn, stimulated he productionof industriesthat used agricultural ommodities as inputs. The presence of a favor-able supply shock in this episode implies that the E,in equation 1 for1921 could be positive. In the discussion of robustness that follows thesimple calculation of the multiplier,I show that even the inclusion of asubstantialpositive residual in 1921 does not change the qualitativeresults.The natureof the policy changesin the years precedingthe recessionsof 1921 and 1938 indicates that these changes were independent ofmovements in the realeconomy: the money supplyandthe governmentsurpluschangedin 1920and 1937because of active policy decisions, notbecause of endogenous responses of money growth or governmentspendingto a fall in real output. Most obviously, in 1920 t was the endof World War I that led to an enormous drop in real governmentspending.The magnitudeof this changecan be seen in the fact that thesurplus-to-GNPratio rose from -8.3 percent in 1919 to 0.5 percent in1920.Monetary policy changes in this episode were also quite pronouncedand largely independent. According to Friedman and Schwartz, theFederal Reserve in 1919became concernedabout the lingering nflationfrom World War I and the postwar boom.2' In response, the FederalReserve raised the discountratethree-quarters f a percentage point inDecember. The diariesand papersof members of the Boardof Gover-nors of the Federal Reserve System that Friedman and Schwartzanalyzed suggest that the FederalReserve did not understandthe lagswithwhichmonetarypolicy affected the economy. As aresult,when the

18 See, for example, Lewis, Economic Survey, p. 19.19 The consumptiondataare fromKendrick,ProductivityTrends, ableA-Ila, p. 294.20 Romer, "WorldWarI."21 Friedmanand Schwartz, MonetaryHistory, pp. 221-39.


9/28

Ending of the Great Depression 765economy failed to respondimmediately o the increasein interest rates,the Federal Reserve raised the discount rate another 1 1/4 percentagepoints in January 1920and an additionalpercentagepoint in June 1920.Because these large increases in interest rates appearto be mainlytheresult of Federal Reserve inexperience, they represent independentmonetary developmentsratherthanconscious responses to the currentstate of the real economy.In 1937the tighteningof fiscal policy was less dramatic,but still quitesevere. In 1936 a largebonus had been paid to veterans of World WarI. In 1937, not only was there no payment of this kind, but socialsecurity taxes also were collected for the first time. This increase inrevenues was clearlyunrelatedto developments n the real economy; itreflected a conscious decision to permanentlyraise taxes to finance apension system. The result of these two changes was that the surplus-to-GNP ratio rose from -4.4 percent in 1936to -2.2 percent in 1937.Monetary changes in 1937 were less straightforwardhan those in1920, but still largelyindependent.Friedmanand Schwartz viewed themainmonetary shock as the doublingof reserve requirements n threesteps between July 1936and May 1937.22The Federal Reserve raisedreserve requirementsbecause it was concernedabout the high level ofexcess reserves in 1936and wanted to turntheminto requiredreserves.Accordingto Friedmanand Schwartz,this actiongreatlydecreasedthemoney supply because banks wanted to hold excess reserves. As aresult, they decreased lendingso that reserves were still higherthanthenew required levels.23 Friedman and Schwartz viewed the resultingchange in the money supply as independent because the FederalReserve was not responding to the real economy: it inadvertentlycontractedthe money supplybecause it misunderstood he motivationof bankers.24The independence of policy movements in 1920 and 1937 and theabsence of additionalcauses of the recessions of 1921and 1938suggestthat these two episodes can be used to estimatemultipliers ormonetaryand fiscal policy. To do this calculation, I merely substituted therelevant data for 1921and 1938 nto equation1 andthen solved the twoequation system for 83f nd Pm Table 1 shows the calculation.

22 Ibid., pp. 543-45.23 The fact that nterestrates rose substantiallyn 1937adds credence o theview that ending ellbecausebanksrestricted oans and not because the demand or loans declined.24 In addition o the change in reserve requirements,he Treasury n 1936began sterilizing hegoldinflow. This resulted n a substantial lowing n the growthrate, thoughnot anactualdecline,of the stock of high-poweredmoney. This switch to sterilizationappears o be partof the samepolicy mistakethatled to the increase n reserve requirements.According o Chandler,America'sGreatestDepression, pp. 177-181,the Treasuryundertook he sterilizationat the behest of theFederal Reserve, which feared that an unsterilizedgold inflow would exacerbate the excessreserves problem.Chandler ited as evidence thatthe Treasurydid not meanto affect the moneysupply the fact thatthey were greatly concernedby the resultingrise in interest ratesin 1937.


10/28

766 RomerTABLE 1CALCULATION OF THE POLICY MULTIPLIERS

Substitutingdatainto equation1 andsettingEt equalto zero yields:1921: 0.0554 = Pm(-0.0424) + 8f(0.0878)1938: -0.0772 = Pm(-0.0877) + 8y (0.0218)

Solvingtwo equations or two unknownsyields:P (-0.0554)(0.0218) - (0.0878)(-0.0772) = 823

( -0.0424)(0.0218)- (-0.0877)(0.0878)= - 0.0772 - 8m(-0.0877) = -0.233

0.0218Note: The intermediate alculationspresenteddifferslightlyfrom the finalmultipliersbecauseofrounding.Source: See the text.

Using this approach,the estimatedmultiplier or monetary policy is0.823 and the estimatedmultiplier or fiscal policy is -0.233. The signsof the two multipliers are what would be expected. f3f is negativebecause the fiscal policy variable is based on the federal surplus; anincreasein the fiscalpolicy measure s contractionary.Themagnitudeofthe monetary policy multiplier is quite reasonable. It implies that agrowthrate of MI that is one percentagepoint lower than normalresultsin real output growththat is 0.82 percentage points lower than normal.As I describein more detaillater,this result is consistent with the effectsof monetaryfactors found in largemacromodels.The magnitudeof thefiscal policy multiplier is quite small. It implies that a rise in thesurplus-to-GNPratio of one percentage point lowers the growthrateofrealoutputrelativeto normalby 0.23 percentagepoints. The reason forthis small multiplier s the fact that the deviation of real output growthfrom normal was slightly smaller in 1921 than in 1938, but the fiscalpolicy shock was nearly four times as large in 1920 as in 1937.Consequently,it wouldbe very difficult o attributemost of the declinesin outputin 1921and 1938 to fiscal policy.SimulationsArmed with these multipliers,it is possible to calculate the likelyeffects of monetary and fiscal developments duringthe mid- and late1930s. As I have set up the analysis, the multipliertimes the policymeasurelaggedone year shows the effect of policy on the deviationofoutput growthfrom normal in a given year. If one subtractsthis effectof unusualpolicy from the actualgrowthrate of real output, one is leftwithestimates of what the growthrate of outputwouldhave been under


11/28

Ending of the Great Depression 7677.0 E | 1 X 1 1 1 1 1

/'6.8 ~~~Actualeal GNP6.8-

E 6.6 GNPUnderNormal. 6.6 - FiscalPolicyg%6- 4

6.2 1933 1935 1937 1939 1941FIGURE 3

ACTUALOUTPUT AND OUTPUTUNDER NORMALFISCALPOLICY,1933-1942Note: The dashed ine shows the pathof the log-valueof real GNP under he assumption hat fiscalpolicywas at its normal evel throughouthe mid- andlate 1930s; he solid line shows the pathofactual real GNP.Sources:The calculationof outputundernormal iscalpolicyis described n the text. The sourcefor real GNP is the same as in Figure1.normal policy. Accumulating these growth rates of real output undernormalpolicy andthen adding hemto the level of outputin a base yearyields a series of the levels of outputunder normalpolicy.The difference between the path of actual output and the path ofoutputundernormalpolicy shows how muchslowerthe recoverywouldhave been in the absence of expansionarypolicy. In calculating he pathof real output under normalpolicy I used 1933as the base year. Thispath shows what output would have been under normal policy after1933, without taking into account the fact that the Depression wasprobably caused to a large extent by serious policy mistakes. Thisprocedureis appropriatebecause the purpose of this article is not toarguethatpolicy did not contribute o the downturnof the early 1930s,but rather that policy was central to the recovery in the mid- and late1930s. In calculatingthe effects of unusual policy, I did the analysisseparatelyfor monetaryand fiscal policy. In one experimentI askedwhat output would have been if fiscal policy had been normal butmonetary policy had followed its actual historicalpath. In a second, Iheld monetarypolicy to its normallevel and let fiscal policy follow itsactualpath.Figure3 shows the experiment or fiscal policy. Thegreat similarityofactual real GNP and GNP under normal fiscal policy indicates thatunusual fiscal policy contributedalmost nothingto the recovery fromthe Great Depression. Only in 1942 is there a noticeable differencebetween actual and hypothetical output, and even in this year thedifference is small.


12/28

768 Romer

C

L -2

1923 1927 1931 1935 1939FIGURE 4

CHANGESIN SURPLUS-TO-GROSS ATIONAL PRODUCTRATIO, 1923-1942Note: The changes are shown lagged one year because this is the formin which they enter mycalculation.Sources:The surplusdataare fromthe U.S. Departmentof the Treasury,StatisticalAppendix,table 2, pp. 4-11 The text describesadjustments hatI madeto the base data. ThesourceforrealGNP is the same as in Figure1.

The small estimated effect of fiscal policy stems in partfromthe factthat the multiplier based on 1921 and 1938 is small, but it is morefundamentallydue to the fact that the deviations of fiscal policy fromnormal were not large during he 1930s.This fact can be seen in Figure4, which shows the change in the surplus-to-GNPratio (lagged oneyear). The change in this ratioin the mid-1930swas typically less thanone percentagepointand was actually positivein some years, indicatingthat fiscal policy was sometimes contractionaryduringthe recovery.Even in 1941, the first year of a substantial wartime increase inspending,the surplus-to-GNPratioonly fell by six percentage points.Figure5 shows the experimentfor monetarypolicy.25 This time thepaths for actual GNP and GNP under normal monetary policy aretremendouslydifferent. The differencein the two paths indicates thathad the money growth ratebeen held to its usuallevel in the mid-1930s,real GNP in 1937 would have been nearly 25 percent lower than itactually was. By 1942the differencebetween GNP under normal andactual monetary policy grows to nearly 50 percent. These calculationssuggest that monetary developments were crucial to the recovery. Ifmoney growth had been held to its normallevel, the U.S. economy in

25 McCallum,"Could a MonetaryBase Rule?" also used a simulationapproach o analyze theeffects of monetary actorsin the 1930s.McCallum's ocus, however,was on whethera monetarybaserule could have prevented he GreatDepression,rather han onwhetheractualmoney growthfueled the recovery.


13/28

Ending of the Great Depression 7697.0 1 1 1 1 1 1 1 1 X

Actual Real GN6.8E 6.6-.9 GNPUnderNormal A,6.4 - MonetaryPolicy --

,/6.2 -1933 1935 1937 1939 1941FIGURE 5

ACTUALOUTPUTAND OUTPUTUNDER NORMALMONETARYPOLICY, 1933-1942Note: The dashed line shows the path of real GNP under the assumption hatthe money growthratewas held to its normalpre-Depression evel throughout he mid- and late 1930s; he solid lineshows the pathof actual realGNP.Sources:The calculationof output under normalmonetarypolicy is describedin the text. Thesourcefor real GNP is the same as in Figure 1.

1942would have been 50 percent below its pre-Depressiontrend path,ratherthan back to its normal evel.26The source of this largeestimatedeffect of monetarydevelopmentsisnot hard to find. As I point out in greater detail in the followingdiscussion, the monetarypolicy multiplier stimatedfrom 1921and 1938is not implausibly arge: it is roughlyof the magnitude ound in postwarmacromodels. The large estimated effects of monetary developmentsaredueto the extraordinarily igh rates of money growth n the mid- andlate 1930s.The monetarypolicy variable(lagged one year) is graphed nFigure6. As can be seen, the deviations of the money growth rate fromnormalwere enormousin the mid-and late 1930s. For most years thesedeviations were over 10 percent. It is not at all surprising, herefore, tofind thathadthis deviation from normalbeen heldat zero, the recoveryfrom the Depression would have been dramaticallyslower.Robustness

The results of these simulations are quite robust. Monetary policywas so expansionary duringthe recovery, and fiscal policy so non-expansionary, that changing the multipliers substantially would notmakemonetarypolicy unimportantand fiscalpolicy crucial. For exam-26 Oniecould start the simulations n 1929 to estimate the role of monetarydevelopmentsincausing heDepression.Whilethisprocedures notstrictlycorrect,because some of themonetarydevelopments n the early 1930s were clearlyendogenous, the results confirmthe conventionalwisdom:monetary orceshad ittle effectduring heonset of the GreatDepression n 1929and 1930,butwere the crucial cause of the deepeningof the Depression n 1931and 1932.


14/28

770 Romer16

84- 4C :? 0

-4-8-

-16l1923 1927 1931 1935 1939FIGURE 6

DEVIATIONSOF MONEY GROWTHRATE FROMNORMAL, 1923-1942Notes: The normalmoney growthrate is definedas the averagegrowthrate of MI between 1923and 1927.The deviationsare shownlaggedone yearbecause this is the formin whichthey entermy calculation.Source:The dataon MI are from FriedmanandSchwartz,MonetaryHistory,tableA-1, column7, pp. 704-34.

pie, assumingthattherewas a substantialpositive supply shock in 1921decreases the monetarypolicy multiplierand increases the fiscalpolicymultiplier.27Even with an extremechange,however, suchas cuttingthemonetary policy multiplierin half and quadrupling he fiscal policymultiplier,real GNP in 1942would have been roughly25 percentlowerthan it actuallywas had monetary policy been held to its normallevelduring the mid- and late 1930s. This result still suggests that theaggregate-demandstimulus of monetary policy was crucial to therecovery. In the case of fiscal policy, quadruplinghe multipliereads tothe conclusion that real GNP would have been 6 percent lower in 1942than it actually was had the change in the surplus-to-GNPratio beenheld to zero. This increases the apparentrole of fiscal policy, but notdramatically.Anotherway to evaluate the robustness of the calculationsis to usepolicy multipliers derived from the estimation of a postwar macro-model. The Massachusetts Institute of Technology-Universityof Penn-sylvania-Social Science Research Council (MPS) model is the main

27 The assumption hat e, in equation1 is large and positive can be included n the calculationshown in Table1by simply subtracting he residual romthe change n output n 1921.This reflectsthe fact that in the absence of the supply shock, the effect of the monetaryand fiscalcontractionwouldhave been larger.An increase n the effectivecontraction f GNP in 1921woulddecrease theestimate of fjam and increase the estimate of f3. For example, if e, in 1921 were 0.0554, then thechange n real GNP less the supplyshock wouldbe -0.1108, double he actualchange n real GNP.Redoing he calculationwith this changeresults n a monetarypolicy multiplier f 0.644and a fiscalpolicy multiplier f -0.951.


15/28

Ending of the Great Depression 771forecastingmodelcurrentlyused by the Federal Reserve Board. In thismodel, the short-runmultiplier or monetarypolicy is 1.2, slightly largerthan the multiplier derived from the 1921 and 1938 episodes; themultiplier or fiscal policy is -2.13, roughlyten times largerthan thatderived from the 1921and 1938episodes.28Using the multipliers rom the MPS model in place of those derivedfrom my calculation increases the apparent importance of monetarypolicy-real GNP in 1942 would have been roughly 70 percent lowerthan it actually was had monetary policy been held to its normalcourse-and increases the role for fiscal policy-real GNP in 1942would have been 14percent lower thanit actually was had fiscal policybeen held to its normal evel. Essentiallyall of this effect of fiscalpolicy,however, comes fromthe last year of the simulation;real GNP in 1941would have been only 1percent lower thanit actually was if fiscal policyhad been held to its normal evel. Thus,using policy multipliersderivedfrom a much differentprocedure thanI used in my illustrativecalcula-tionleads to the same conclusion thatmonetarypolicy was crucial to therecovery from the Great Depression and fiscal policy was of littleimportance.29

One characteristicof most multipliersderived fromlargemacromod-els is that the effects of aggregate-demandpolicy on the level of realoutputare forced to become zero in the long run. This is certainlythecase in the MPS model in which the long-runbehavior of the economyis assumed to follow the predictionsof a Solow growthmodel. In mysimulations,bothwithmy own multipliersand with those from the MPSmodel, I only considered the short-runmultipliersand did not requirethat the positive effects of an expansionaryaggregate-demand hock onthe level of real output be eventually undone. I did this because the28 These multipliers re reportednthe U.S. Boardof Governorsof the FederalReserve System,"Structureand Uses of the MPS QuarterlyEconometricModel," tables 1 and 2. The monetarypolicy shock used in the MPS simulation s a permanentncrease in the level of MI of 1 percentover the projectedbaseline. Thisis equivalent o the shockI considered n my simulations,whichis a one-time deviation in the growth rate of MI from its normalgrowth rate. I used the MPSmultiplier erived romthefull-model esponse(case 3 of table 2). The fiscal shock usedinthe MPSsimulation s a permanentncrease n the purchasesof the federalgovernmentby 1percentof realGNP over the baseline projection.This differs rom the shock I considered,which is a change inthe surplus-to-GNP atio, because tax revenues will rise in responseto the induced increase inGNP. To make the MPS multiplierconsistent with my measureof fiscal policy, I assumedthe

marginal ax rate to be 0.3 and then calculated the change in the surplus-to-GNPratio thatcorresponded o a 1 percentincreasein federal purchases.The MPS multiplier hat I adjusted nthis way is basedon the full-modelresponse, with MI fixed (case 4 of table 1).29 Weinstein,"Some Macroeconomic mpacts," performeda similarcalculation or monetarypolicy using multipliersderived rom the Hickman-Cohenmodeland found a large potentialeffectof the monetaryexpansion n 1934and 1935.However, he emphasized hat theNationalIndustrialRecovery Act acted as a negativesupply shock and counteracted he monetaryexpansion. Whilethe NIRA may indeed have stuntedthe recovery somewhat, it does not follow from this thatmonetarypolicy was unimportanto the recovery. In the absenceof the monetaryexpansion,thesupplyshockcould have led to continueddecline rather hanto therapidgrowthof realoutput hatactually occurred.


16/28

772 Romerconstraintthat the long-runeffects of policy are zero is simply imposeda priori n most models;availableevidence indicates thatthe realeffectsof policy shifts are in fact highlypersistent.30Provided that we do not assume that the positive effects of expan-sionary policy are quickly reversed (that is, within a year or two),allowingfor negativefeedback effects from a policy stimulus would notsubstantially diminish the role of policy in generating the high realgrowth rates observed in the mid- and late 1930s. This is true for tworeasons: in the first few years of the expansionthere would have beenno negativefeedbackeffectsfrompreviouspolicy expansions, and therewere progressivelylargermonetarygrowth rates toward the end of therecovery. Furthermore, here is no supportfor the view that the effectsof policy shifts are counteracted rapidly. In the MPS model, forexample, the effects of both fiscal and monetaryshocks do not start tobe counteracted substantiallyuntil twelve quartersafter the shocks.Thus, even under the assumptionthatpolicy does not matter n the longrun, we would still find that policy was importantfor the eight to tenyears that encompassed the recovery phase of the GreatDepression.

THE SOURCEOF THE MONETARYEXPANSIONThat economic developments would have been very different in themid- and late 1930shad money growthbeen held to its normal level isevident from the calculations above. But to go furtherand argue thataggregate-demand timulus actually caused the recovery, it must beshown that the rapid rates of monetary growth were due to policyactionsandhistoricalaccidents, and were not the result of higheroutputbringing orth money creation. This is easy to do.The main way that the money supply might grow endogenously isthrough demand-inducedchanges in the money multiplier. If, in re-sponse to a boom, banks raise the deposit-to-reserveratio and custom-ers accept a higher deposit-to-currencyratio, a given supply of high-powered money can support a larger stock of MI. Neither of thesechanges, however, occurred duringthe recovery from the Great De-pression. The deposit-to-reserveratio fell steadily in the mid- and late1930s, from 8.86 in January 1933 to 4.67 in December 1942. Thedeposit-to-currencyratio rose initially in the recovery as the bankingsystem regainedcredibility,but remained airlyconstant from 1935until1941,and then fell sharplyin late 1941 and 1942.31Since the behavior of both these ratios suggests that the moneymultiplier fell during the recovery from the Great Depression, theobservedrise in MI must have been due to even larger ncreases in thestock of high-poweredmoney duringthis period. This increase in the

30See, for example, Romerand Romer,"Does MonetaryPolicyMatter?"31 The data are from Friedmanand Schwartz,MonetaryHistory, table B-3, pp. 799-808.


17/28

Ending of the Great Depression 773stock of high-poweredmoney was also not endogenous. There is noevidence that the Federal Reserve increased the stock of high-poweredmoney to accommodate the higher transactions demand for moneycaused by increasedoutput. Instead, the Federal Reserve maintainedapolicy of cautionthroughout he recovery and even stoppedincreasingFederal Reserve credit to meet seasonal demandsin the mid- and late1930s.32The source of the hugeincreasesin the U.S. money supplyduring herecovery was a tremendousgold inflow that began in 1933. Friedmanand Schwartzstated thatthe "rapidrate [of growthof the money stock]in the three successive years from June 1933 to June 1936 . . . was aconsequence of the gold inflow producedby the revaluationof goldplusthe flightof capitalto the United States. It was in no way a consequenceof the contemporaneousbusiness expansion."33The monetary goldstock nearly doubled between December 1933and July 1934and thenincreased at an average annual rate of nearly 15 percent betweenDecember 1934and December 1941. Arthur Bloomfieldagrees withFriedmanand Schwartzthat "the devaluationof the dollar, for techni-cal reasons, was . . . the direct cause of much of the heavy net goldimports of $758 million in February-March,1934."35 Thus, the initialgold inflowwas the result of an active policy decision on the partof theRoosevelt administration.Both these studies, however, attributed most of the continuingincreases in the U.S. monetarygold stock throughout he later 1930stopolitical developments in Europe. Bloomfield pointed out that thecontinued gold inflow was caused primarily by huge net imports offoreign capital into the United States; the United States ran persistentandlargecapitalaccount surpluses n the mid-and late 1930s.36He thenarguedthat "probablythe most importantsingle cause of the massivemovement of funds to the United States in 1934-39 as a whole was therapiddeterioration n the internationalpolitical situation.The growingthreat of a European war created fears of seizure or destruction ofwealth by the enemy, impositionof exchange restrictions, oppressivewar taxation. . . . Huge volumes of funds were consequently trans-ferredin panic to the United States from WesternEuropeancountrieslikely to be involved in such a conflict."37Friedmanand Schwartzweremore succinct when they concluded: "Munichand the outbreakof warin Europewere the mainfactors determining he U.S. money stock in

32 Ibid., pp. 511-14.33 Ibid., p. 544.34 The data are from Chandler,America's GreatestDepression, p. 162.3' Bloomfield,CapitalImports,p. 142.36 According o Bloomfield,CapitalImports, p. 269, the United Statesalso ran a smallcurrentaccount surplus n every year except 1936.37 Ibid., pp. 24-25.


18/28

774 Romerthose years [1938-1941],as Hitlerand the gold miners had been in 1934to 1936. ,38

Finally, the Roosevelt Administration'sdecisions to devalue and notto sterilize the gold inflow were clearly not endogenous. Barrie Wig-more showed thatRoosevelt spoke favorablyof devaluation n January1933.39Since this was many months before recovery commenced,Roosevelt could not have been respondingto real growth. Indeed, G.Griffith ohnson'sanalysisof the Roosevelt administration's old policysuggested that, if anything,the Treasurywas tryingto counteract theDepressionthrougheasy money, rather hantryingto accommodatetherecovery.40Johnsonand Wigmorealso showed that Roosevelt's desireto encouragea gold inflow was not based on a conventional view of themonetarytransmissionmechanism,but ratheron the view that devalu-ation would directly raise prices and reflationwould directly stimulaterecovery.4'The fact that the continuing gold inflow of the mid-1930swas notsterilizedappearsto be partlythe result of technicalproblemswith thesterilizationprocess. TheGoldReserve Act of 1934set upa stabilizationfund and made explicit the role of the Treasury in intervening in theforeignexchange market.However, because the stabilization und wasendowed only with gold, it was technically able only to counteract agold outflow, not a gold inflow.42As a result, sterilizationwould haverequiredan active decision to change the new operatingprocedures.Such a decision was not made because Roosevelt believed that anunsterilizedgold inflowwould stimulatethe economy throughreflation.The devaluation and the absence of sterilizationthus appearto havebeen the result of active policy decisions and a lack of understandingabout the process of exchange marketintervention. To the degree thatactive policy was involved, it was clearly aimed at encouragingrecov-ery, not simplyat responding o a recoverythatwas alreadyunderway.Combinedwith the fact thatpoliticalinstabilitycaused much of the goldinflow in the late 1930s, these findings ndicate that the increase in themoney supply in the recovery phase of the Great Depression was notendogenous. Since the simulationresults showed that the large devia-tions of money growth rates from normal account for much of therecovery of real outputbetween 1933and 1937and between 1938and1942, it is possible to conclude that independent monetary develop-ments account for the bulk of the recovery from the GreatDepressionin the United States.

38 Friedman and Schwartz, Monetary History, p. 545.39 Wigmore,"Was the BankHolidayof 1933?"p. 743.4 Johnson, Treasury and Monetary Policy, pp. 9-28.41 Johnson,Treasury ndMonetaryPolicy,pp. 14-16;andWigmore,"Was the Bank Holidayof1933?"p. 743.42 Johnson, Treasury and Monetary Policy, pp. 92-114.


19/28

Ending of the Great Depression 775THE TRANSMISSIONMECHANISM

The argument that monetary developments were the source of therecovery can be made more plausible by identifying the transmissionmechanism.It is generallyassumedthat the usual way anincreasein themoney supply stimulatesthe economy is througha decline in interestrates. An increase in the money stock lowers nominal interest rates;with fixed or increasing expected inflation,this decline in nominal ratesimplies a decline in real interest rates. A fall in real interest ratesstimulates purchases of plant and equipment and durable consumergoods by loweringthe cost of borrowingand by reducing the opportu-nity cost of spending.For this mechanismto have been operating n the mid- and late 1930s,the rapid money growth could not have been immediately and fullyoffsetby increases in wages andprices. If wages andprices increased asrapidlyor morerapidlythanthe money supply,realbalanceswould nothave increased and there would have been no pressure on nominalinterestrates. The realmoney supplydid in fact rise at a very rapidrateduringthe second half of the 1930s:MI deflatedby the wholesale priceindex increased by 27 percent between December 1933and December1936and by 56 percentbetween December 1937andDecember 1942.43This suggeststhatpricesandwages did not fully adjustto the rapidratesof money growth. The fact that nominal interest rates fell duringtherecovery is consistent with this increase in realbalances. The commer-cial paper rate, for example, fell froman averagevalue of 2.73 in 1932to 0.75 in 1936.44For the interest-rate ransmissionmechanismto have been operatingin the mid-andlate 1930s,it would also have to have been the case thatthe rapid money growth rates generated expectations of inflation. By1933 nominal interest rates were already so low that there was littlescope for a monetaryexpansionto lower nominal rates further. There-fore, the main way that the monetary expansion could stimulate theeconomy was by generatingexpectationsof inflationand thus causingareductionin real interest rates. Such expectations of inflationare notinconsistent with the existence of the wage andprice inertia. Indeed, avery plausible explanation s that the rapid money growthrates did notimmediatelyincrease wages and prices by an equivalent amount be-cause of internal abormarkets, governmentregulations,or managerial

43 To calculatereal money I subtracted he logarithm f the producerpriceindex(PPI)fromthelogarithmof Ml. The data on the PPI are from the U.S. Bureau of Labor Statistics, HistoricalData. Because Ml is only availableseasonally adjusted, I also seasonally adjustedthe PPI byregressing t on monthlydummyvariablesanda trend.44The commercialpaperrate data are fromthe U.S. Boardof Governorsof the FederalReserveSystem, Bankingand MonetaryStatistics,1943,pp. 448-51,and 1976,p. 674.Theycover four-tosix-monthprime commercialpaperandare not seasonally adjusted.


20/28

776 Romer40

NominalRate20-

00.-20-

-40- Ex PostRealRate%99 931 F3 35 s37 G9 94

FIGURE 7NOMINAL AND EX POST REAL COMMERCIAL PAPER RATES, 1929-1942

Note: The data are quarterlyobservations.Sources: The commercialpaperrate data are from the U.S. Board of Governorsof the FederalReserve System, Banking and Monetary Statistics, 1943, pp. 448-51, and 1976, p. 674. Thecalculationof the ex post realrate is described n the text.

inertia.45However, consumers andinvestors realized thatprices wouldhave to rise eventually and therefore expected inflation over thenot-too-distanthorizon.Regressionestimatesof the ex ante realinterestrate suggestthatthiscondition is met in the recovery phaseof the GreatDepression.FredericMishkin showed using the Fisher identity that the difference betweenthe ex ante real rate that we want to know and the ex post real rate thatwe observe is unanticipated inflation.46 Under the assumption ofrationalexpectations, the expectation of unanticipated nflationusinginformationavailableat the time the forecast is made is zero. Therefore,if one regresses the ex post real rateon currentandlaggedinformation,the fitted values provide estimates of the ex ante real rate.To apply this procedure I first calculated ex post real rates bysubtractingthe change in the producerprice index over the followingquarter at an annualrate)from the four-to-six month commercialpaperrate.47These ex post real rates, along with the nominal commercialpaperrate, are shownin Figure7. I thenregressedthe ex post realrateson the current value and four quarterlylags of the monetary policyvariable described in the multipliercalculations (but disaggregatedtoquarterly values), the percentage change in industrial production,inflation, and the level of the nominal commercial paper rate. Toaccount for possible seasonal variationI also included a constant term

" O'Brien,"A BehavioralExplanation,"providedone suchexplanationorwagerigidityduringthe 1930s.4 Mishkin, "The Real InterestRate."47 In this calculationneither series was seasonally adjusted.


21/28

Ending of the Great Depression 777TABLE 2REGRESSIONUSED TO ESTIMATEEX ANTE REAL INTERESTRATES

ExplanatoryVariable Coefficient T-StatisticMonetaryPolicyVariableLag0 0.044 0.29Lag 1 -0.463 -3.02Lag2 0.182 1.09Lag3 -0.196 -1.20Lag4 0.352 2.30Nominal CommercialPaper RateLag0 0.834 0.25Lag 1 0.191 0.04Lag 2 1.181 0.22Lag 3 0.954 0.18Lag 4 -1.079 -0.32InflationRateLag0 -0.396 -2.54Lag 1 0.129 0.81Lag 2 -0.014 -0.09Lag 3 0.111 0.72Lag 4 -0.031 -0.21Change n IndustrialProductionLag0 -0.026 -0.47Lag 1 0.045 0.78Lag 2 -0.120 -2.00Lag 3 0.012 0.22Lag 4 -0.036 -0.67QuarterlyDummyVariablesQuarter2 1.497 0.27Quarter3 -6.961 -1.76Quarter4 5.271 0.97Constant -1.804 -0.44

Notes: The dependentvariable s the quarterly x post real interest rate. The sample periodusedin the estimation s 1923:1 o 1942:2.The R2 of the regression s .52.Source: See the text.

and three quarterly dummy variables. I ran this regression over thesample period 1923:1to 1942:2.48The results are shown in Table 2. The explanatory variables Iincluded in the regression explain a substantial fraction of the totalvariation n the ex post realinterestrate: the R2 of the regressionis .52.Of the individualexplanatoryvariables, the one of most interest is themonetary policy variable. If the conventional transmissionmechanismwas operating, the monetary policy variable should be negativelycorrelatedwith the ex post real rate. As can be seen, this is clearly thecase: the firstlag of the monetary policy variableenters the regressionwith a coefficient of -0.463 and has a t-statistic of -3.02.

48The monetarypolicy variablewas disaggregated y converting he quarterlygrowthratesofMl during he recoveryto annualratesand thensubtracting ff the averageannualgrowthrateofMI in the mid-1920s.The industrialproduction eries is fromthe U.S. Boardof Governorsof theFederal Reserve System,IndustrialProduction, able A.11, p. 303.


22/28

778 Romer40I

30-

Q so2010

020-

- 929 131 1933 135 1937 139 194FIGURE 8

EX ANTE REAL COMMERCIAL APERRATES, 1929-1942Note: The data are quarterlyobservations.Source:The regressionused to estimateex ante real rates is given in Table2 anddescribed n thetext.

The fitted values of the regression,which providean estimate of theex ante real rate, are graphed n Figure 8. These estimates suggest thatex ante real rates droppedprecipitouslyat the start of the monetaryexpansion in 1933 and remainedlow or negative for the rest of thedecade (except for the rise duringthe monetary contraction of 1937/38).49Indeed, the drop in real rates between the contractionaryandexpansionaryphases of the GreatDepressionis remarkable: x ante realrates fell fromvalues often over 15percentin the early 1930sto valuestypicallybetween -5 and - 10 percentin the mid-1930sand early 1940s.While one cannotbe surethat actualex ante real ratesdroppedthe sameamount as these estimates or that the drop was caused by monetarydevelopments, the regressionresults certainly suggest that the expan-sionary monetarydevelopmentsof the mid- and late 1930sdid have asubstantial impact on real interest rates.50 Thus, this aspect of theconventionalmonetarytransmissionmechanismappearsto have beenoperating n the recovery phase of the GreatDepression.For expansionary monetary developments to have stimulated the

49The estimates are strikingly obustto variations n the specificationof the regression.I triedmany variants of the basic regression, such as excluding contemporaneousvalues of theexplanatoryvariables,extending he sample periodto include 1921,and leaving out the seasonaldummyvariables.None of these changes noticeablyalteredthe estimatesof the ex antereal rate.so Some of the inflation n 1933and 1934couldhave been due to the NIRA, which encouragedcollusionaimedat raisingprices,rather hanto monetarypolicy. However,the NIRA was declaredunconstitutionaln 1935and its policieswere ones thatwould tend to cause a one-time umpin theprice level rather han continued nflation.Thus, thoughsome of the initial all in real interestratescould have beendue to the NIRA, the continuednegativereal ratesinthe mid-and ate 1930smusthave been due to other causes.


23/28

Ending of the Great Depression 7790.4 I I I II I I I I 20

04 6 I I I 1 1 1 1 1 1 1 1 -20

FixedInvestmenta60.2 ' A ~~~~~~~~~~12

o ~~~~~~~~~~40' /~~~~~~~~~.C ~ ~~~~~~~~~ C-)

c-0.4'/economyn the mid- and late 193Real nterestRate n y

--0.6 I I 1I I I -121930 1932 1934 1936 1938 1940FIGURE 9

REAL FIXED INVESTMENTAND EX ANTE REALRATES, 1930-1941Sources:Dataon realfixed investmentare from the U.S. Bureauof EconomicAnalysis,NatiernalIncomeand ProductAccounts,table 1.2, p. 6. The estimationof ex ante real ratesis described nthe text.

economy in the mid- and late 1930s,real interestrates not only had tofall, butinvestmentandothertypes of interest-sensitivespendinghadtorespond positively to this drop. Figure9 shows the annualpercentagechanges in real total fixed investmentand Figure 10 shows the changes0.3 w l l l l l l l l l l 207 c Consumer xpenditures| 0.20\ Arg on DurableGoods 16

F1 V24-

I-

o CR.c-o.1 I/I~~~~~~~~1% -4o-0.2 -A

aRel Intereste --0.3 I I -I I I I -' -121930 1932 1934 1936 1938 1940FIGURE 10

REAL CONSUMEREXPENDITURESON DURABLEGOODSAND EX ANTE REALRATES, 1930-1941Sources: Data on real consumerexpenditureson durablegoods are from the U.S. BureauofEconomicAnalysis,NationalIncome andProductAccounts,table 1.2, p. 6. The estimationof exante real rates is described n the text.


24/28

780 RomerTABLE 3CORRELATION BETWEEN SPENDING AND REAL INTEREST RATES, 1934-1941

Percentage ChangePercentage Change in Real Consumerin Real Fixed Expenditures onInvestment Durable GoodsEx Ante Real RateLag 0 -0.687 -0.746Lag 1 -0.292 -0.238Lag 2 -0.052 -0.030Sources: The sources are the same as for Figures 9 and 10.

in real consumer expenditureson durablegoods.51In both figurestheannualaveragesof the estimates of the ex anterealinterestratearealsoshown. These graphs suggest that there was a very strong negativerelationshipbetween real interest rates and the percentage change inspending n the mid-and late 1930s. Fixed investment and the consump-tion of durablegoods both turnedupwardsoon after the plunge in realratesin 1933. Over the next fouryears, realrates remainednegativeandspending grew rapidly. In 1938the recovery was interrupted,as realrates turnedsubstantiallypositive and spendingfell sharply.Starting n1939real rates fell again, and the rapidgrowth of spending resumed.The relationshipbetween spendingand interest rates can be quanti-fied by computing the correlationsbetween the percentage change infixed investment or consumerspendingon durablesandthe level of theex ante real rate. Table 3 shows these correlationsestimated over theperiod 1934 to 1941. The table shows that there is a strong negativecontemporaneous correlation between interest rates and the growthrates of investmentandconsumerspendingon durablegoods during herecovery phase of the Great Depression. There is also a moderatelystrongnegative correlationbetween the percentagechange in spendingand interest rates laggedone year.A negative relationship also exists between quarterlydata on con-structioncontracts and real interest rates. The contracts data show thefloor space of new buildings or which contractswere drawnup duringthe quarter.52One might reasonably expect the volume of such con-tracts to respond quickly to movements in interest rates because theyinvolved plannedratherthanactualexpenditures.And indeed, over theperiod 1933:2to 1942:2 the contemporaneouscorrelation between the

" These data are from the U.S. Bureauof EconomicAnalysis,NationalIncome and ProductAccounts, table 1.2, p. 6.52 The Dodgeconstructioncontractseries forresidential,commercial,andindustrial tructuresis available n LipseyandPreston,SourceBook, seriesA8, p. 73;seriesA17, pp. 95-96; andseriesA19, pp. 100-101.I used the versionthat shows the floor space of each type of buildingwithoutseasonaladjustment.Thedatafor27stateswas splicedontodatafor 37 states in 1925.I seasonallyadjustedthe series by regressing the logarithmof contracts on a trend, a constant, and threequarterlydummyvariables.


25/28

Ending of the Great Depression 781percentage changein constructioncontracts and the ex ante real rate is-0.4. The low interest rates of the mid-1930s and the early 1940scorrespond to periods of rapid increase in construction contracts.These correlations cannot prove that the fall in interest rates causedthe surgein investment, durablegoods expenditures, and construction.They do, however, suggest that there is no obvious evidence that theconventional transmission mechanism for monetary developmentsfailed to operate duringthe mid- and late 1930s. One piece of evidencethat suggests a more causal link between the fall in interest rates and therecovery is the lag in the reboundof consumer expenditureson servicescomparedwith those on durables.Expenditureson durables increasedbetween 1933and 1934,but real consumerexpenditureson services didnot turn around until 1935. This suggests that it was not a surge ofoptimismthatwas pullingupalltypes of consumerexpenditures n 1934,butrather some force, such as a fall in interestrates, that was operatingprimarilyon durablegoods.54

CONCLUSIONSMonetary developmentswere a crucial source of the recovery of theU.S. economy from the Great Depression. Fiscal policy, in contrast,contributedalmostnothingto the recovery before 1942. The very rapidgrowthof the money supply beginning n 1933appearsto have loweredreal interest rates and stimulated nvestment spending ust as a conven-tional model of the transmissionmechanismwould predict.The moneysupply grew rapidly in the mid- and late 1930s because of a hugeunsterilizedgold inflow to the United States. Although the later gold

inflowwas mainlydue to political developmentsin Europe, the largestinflow occurred mmediately ollowingthe revaluationof goldmandatedby the Roosevelt administrationn 1934.Thus, the gold inflowwas duepartlyto historical accident andpartlyto policy. The decision to let thegold inflow swell the U.S. money supplywas also, at least in part, anindependent policy choice. The Roosevelt administration hose not tosterilize the gold inflow because it hoped that an increase in themonetary gold stock would stimulate the depressed economy.53 For this calculation,I seasonallyadjusted he ex ante real interestrateseries by regressing ton a constant and three quarterlydummyvariables.54 The conventionalmonetary ransmissionmechanismneed not have been the only way thatexpansionary monetary developmentsstimulatedreal growth duringthe mid- and late 1930s.Recent studies, such as Bernanke,"NonmonetaryEffects," have emphasized hat debt-deflationcould have been an important ourceof weakness in the banking ector,and that banking ailurescould havehurtrealoutputby reducing he amountof credit ntermediation. f this was indeed thecase, then the inflationgeneratedby the tremendous ncrease n the money supply startingn 1933could have had a beneficialeffect on the financial ystem. By reducing he real value of outstandingdebts, the inflationmay have strengthenedhe solvency of banksand businessesand hastened herecoveryof the financial ystem.


26/28

782 RomerThat monetary developments were very important, whereas fiscalpolicy was of little consequence even as late as 1942, suggests an

interestingtwist on the usualview thatWorldWarII caused, or at leastaccelerated, the recovery from the Great Depression. Since the econ-omy was essentially back to its trend level before the fiscal stimulusstarted in earnest, it would be difficultto argue that the changes ingovernment spending caused by the war were a major factor in therecovery. However, Bloomfield'sand Friedmanand Schwartz's analy-ses suggested that the U.S. money supply rose dramaticallyafter warwas declared in Europe because capitalflight rom countries involved inthe conflict swelled the U.S. gold inflow.In this way, the warmay haveaided the recoveryafter1938by causingthe U.S. money supplyto growrapidly. Thus, WorldWarII may indeed have helped to end the GreatDepression in the United States, but its expansionarybenefits workedinitially through monetary developments rather than through fiscalpolicy.The finding hat monetarydevelopmentswere crucial to the recoveryconfirms or complementsa numberof analyses of the end of the GreatDepression. Most obviously, it supportsFriedmanand Schwartz's viewthat monetary developments were very importantduringthe 1930s. Itsuggests, however, that Friedman and Schwartz's emphasis on theinactionof the Federal Reserve after1933 s somewhat misplaced.Whatmattered is that the money supply grew rapidly;the fact that this risewas orchestratedby the Treasuryratherthan the Federal Reserve is ofsecondary importance.The finding hat fiscalpolicy contributed ittle tothe recoveryechoes Brown'sfinding hat fiscalpolicy was not obviouslyexpansionary duringthe mid-1930s.

My analysis also supports studies that emphasizethe devaluation of1933/34as the engineof recovery. Peter Temin andWigmoreargued hatthe devaluationsignalledthe end of a deflationarymonetaryregimeandthat this change in regimewas crucial to improvingexpectations.55Inthis explanation t was the changein expectationsthatbroughtabout theturning point in the spring of 1933. My work bolsters Temin andWigmore's conclusion by showing that the deflationary regime wasindeed replaced by a very inflationary monetary policy. This mayexplain why the regimeshiftwas viewed as credible. More importantly,it can explain why the initialrecovery was followed by continued rapidexpansion. Withoutactual inflationand actual declines in real interestrates, the recovery stimulatedby a changein expectationswouldalmostsurelyhave been short-lived.In the same way, this articlealso bolstersthe argumentof Barry Eichengreen and Jeffrey Sachs that devaluation

55 Temin and Wigmore, "End of One Big Deflation."The importanceof devaluation s alsodiscussed in Temin, Lessons from the Great Depression.


27/28

Ending of the Great Depression 783can stimulaterecovery by allowingexpansionarymonetary policy."6Itshows that in the case of the United States, devaluation was indeedfollowed by salutary increases in the money supply.On the otherhand, my findingsappearto disputestudies that suggestthat the recovery from the Great Depression was due to the self-corrective powers of the U.S. economy in the 1930s. I find thataggregate-demand timuluswas the main source of the recovery fromthe Great Depression. Thus, the Great Depression does not provideevidence that large shocks are rapidly undone by the forces of meanreversion. Rather, it suggests that large falls in aggregate demand aresometimes followed by large rises, the combinationof which leaves theeconomy back on trend.

56 Eichengreenand Sachs, "ExchangeRates."

REFERENCESAyres, Leonard P., Turning Points in Business Cycles (New York, 1939).Bernanke, Ben S., "NonmonetaryEffects of the FinancialCrisis in the Propagation fthe Great Depression," American Economic Review, 73 (June 1983), pp. 257-76.Bernanke,Ben S., and MartinS. Parkinson,"Unemployment,Inflation,andWages inthe AmericanDepression: Are There Lessons for Europe?" AmericanEconomicReview, 79 (May 1989),pp. 210-14.Bernstein, MichaelA., TheGreat Depression(New York, 1987).Blinder, Alan S., andRobertM. Solow, "AnalyticalFoundationsof Fiscal Policy," inThe Economics of Public Finance, Brookings Institution Studies in GovernmentFinance(Washington,DC, 1974),pp. 3-115.Bloomfield, Arthur I., Capital Imports and the American Balance of Payments, 1934-39(Chicago, 1950).Brown, E. Cary, "Fiscal Policy in the 'Thirties:A Reappraisal,"AmericanEconomicReview, 46 (Dec. 1956), pp. 857-79.Chandler, Lester V., America's Greatest Depression, 1929-1941 (New York, 1970).Darby, Michael, "Three-and-a-HalfMillionU.S. EmployeesHaveBeen Mislaid:Or,anExplanation f Unemployment1934-1941,"Journalof PoliticalEconomy,84(Feb.1976), pp. 1-16.De Long,J. Bradford,andLawrenceH. Summers,"How Does MacroeconomicPolicyAffect Output?" Brookings Papers on Economic Activity (1988:2), pp. 433-80.Eichengreen,Barry, and Jeffrey Sachs, "Exchange Rates and EconomicRecovery inthe 1930s," this JOURNAL, 5 (Dec. 1985), pp. 925-46.Friedman, Milton, and Anna J. Schwartz, A Monetary History of the United States,1867-1960(Princeton,1963).Gordon, Robert Aaron, Economic Instability and Growth: The American Record (NewYork, 1974).Hansen, Alvin, Full Recovery or Stagnation? (New York, 1938).Johnson, G. Griffith, The Treasury and Monetary Policy, 1933-1938 (Cambridge, MA,1939).Kendrick, John W., Productivity Trends in the United States (Princeton, 1961).


28/28

784 RomerLebergott, Stanley, Manpower in Economic Growth: The Record Since 1800 (NewYork, 1964).Lewis, W. Arthur, Economic Survey, 1919-1939 (London, 1949).Lipsey, Robert E., and Doris Preston, Source Book of Statistics Related to Construc-tion (New York, 1966).Margo,Robert,"InterwarUnemployment nthe United States:Evidencefromthe 1940Census Sample," in BarryEichengreenand T. J. Hatton, eds., InterwarUnem-ployment in International Perspective (Dordrecht, 1988), pp. 325-52.McCallum,Bennett T., "Could A MonetaryBase Rule Have Prevented the GreatDepression?" Journal of Monetary Economics, 26 (Aug. 1990), pp. 3-26.Mishkin, Frederic, "The Real Interest Rate: An EmpiricalInvestigation,"in KarlBrunner and Alan Meltzer, eds., The Costs and Consequences of Inflation,Carnegie-RochesterConference Series on Public Policy, vol. 15 (Amsterdam,

1981),pp. 151-200.O'Brien, Anthony Patrick, "A BehavioralExplanationfor Nominal Wage RigidityDuring the Great Depression," Quarterly Journal of Economics, 104 (Nov. 1989),pp. 719-35.Romer, ChristinaD., "World WarI and the PostwarDepression: A ReinterpretationBased on Alternative Estimates of GNP," Journal of Monetary Economics, 22(July 1988), pp. 91-115.Romer, ChristinaD., and David H. Romer, "Does MonetaryPolicy Matter?A NewTest in the Spirit of Friedmanand Schwartz," NBERMacroeconomicsAnnual,4(1989), pp. 121-70.Roose, Kenneth D., The Economics of Recession and Revival (New Haven, 1954).Smithies, Arthur, "The AmericanEconomy in the Thirties," American EconomicReview, 36 (May 1946),pp. 11-27.Temin, Peter, Lessons from the Great Depression (Cambridge, MA, 1989).Temin, Peter, and BarrieWigmore,"The End of One Big Deflation,"Explorations nEconomicHistory,27 (Oct. 1990), pp. 483-502.U.S. Board of Governors of the Federal Reserve System, Banking and MonetaryStatistics (Washington,DC, 1943and 1976).U.S. Board of Governors of the Federal Reserve System, Industrial Production(Washington,DC, 1986).U.S. Board of Governorsof the Federal Reserve System, "Structureand Uses of theMPS QuarterlyEconometric Model of the United States," Federal ReserveBulletin, 73 (Feb. 1987),pp. 93-109.U.S. Bureau of Economic Analysis, The National Income and Product Accounts,1929-1982(Washington,DC, 1986).U.S. Bureau of Labor Statistics, Historical Data on the Producer Price Index(Microfiche,Washington,DC, 1986).U.S. Department of the Treasury, Statistical Appendix to the Annual Report of theSecretary of the Treasury on the State of the Finances (Washington, DC, 1979).Weinstein,Michael,"SomeMacroeconomic mpactsof the NationalIndustrialRecov-

ery Act, 1933-1935," in Karl Brunner, ed., The Great Depression Revisited(Boston, 1981),pp. 262-81.Wigmore,BarrieA., "Was the BankHolidayof 1933Causedby a Runon the Dollar?"this JOURNAL, 47 (Sept. 1987), pp. 739-55.

Date post:	07-Apr-2018
Category:	Documents
Upload:	raheelnadeem
View:	218 times
Download:	0 times

What Ended the Great Depression

Documents