Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Exchange rate wikipedia , lookup
Reserve currency wikipedia , lookup
Currency war wikipedia , lookup
Bank for International Settlements wikipedia , lookup
Fixed exchange-rate system wikipedia , lookup
Foreign-exchange reserves wikipedia , lookup
Bretton Woods system wikipedia , lookup
Viewpoint: Understanding the Great Depression Author(s): Barry Eichengreen Source: The Canadian Journal of Economics / Revue canadienne d'Economique, Vol. 37, No. 1 (Feb., 2004), pp. 1-27 Published by: Blackwell Publishing on behalf of the Canadian Economics Association Stable URL: http://www.jstor.org/stable/3696096 Accessed: 28/08/2009 17:03 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=black. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected]. Canadian Economics Association and Blackwell Publishing are collaborating with JSTOR to digitize, preserve and extend access to The Canadian Journal of Economics / Revue canadienne d'Economique. http://www.jstor.org Viewpoint:Understandingthe Great Depression Barry Eichengreen University of California, Berkeley If there is a feature of modern scholarship on the Great Depression that distinguishes it from its antecedents, it is the tendency of recent contributions to frame that event as a global phenomenon (see, e.g., Eichengreen 1992; Johnson 1998; Bernanke 2000; and James 2001). It is now commonplace to view the Depression from an internationalperspectiveand to test causal hypotheses by comparing the experiences of countries that endured more and less severe cyclical downturns. This is in contrast with the earlier generation of scholarship, notably the influential work of Friedman and Schwartz (1963) and Temin (1976), which viewed the Depression as centred in the United States and focused on shocks and policies in that country. Specific hypotheses associated with the global perspectivecontinue to be disputed, but there seems little question that the internationalview serves as a point of departure for a growing portion of recent scholarship. Whether this new research has produced a scholarly consensus is another matter. In this paper I ask whether it is possible to arrive at such a consensus by synthesizing the old and new literatures.I describe a perspective that views the Great Depression as a global phenomenon, emphasizinglinkages among countries and the destabilizinginfluence of the internationalmonetary and financial system, but that at the same time acknowledgesthe uniquely importantcharacterof events and policies in the United States. I therebysuggest that the opposition between the 'old' and 'new' (or 'U.S.-centric'and 'global') views of the Great Depression may ultimately prove artificial.The task for macroeconomic and historical scholarship going forward is to reconcilethese previouslycompeting strands of thought. This is a revisionof the 2002 MackintoshLecture,deliveredat Queen'sUniversity.I am gratefulto MugeAdaletfor able researchassistanceand to RobertGordon,ChrisMeissner, Kris Mitchener,and two anonymousrefereesof thisjournalfor commentsand assortedforms of help. Email:[email protected] Canadian Journal of Economics / Revue canadienne d'Economique, Vol. 37, No. 1 February / fevrier 2004. Printed in Canada / Imprime au Canada 0008-4085/ 04 / 1-27 / ? CanadianEconomicsAssociation 2 B. Eichengreen 1. The onset of the GreatDepression The most difficultchallengefor any analysisof money and output, and hence for the monetaryinterpretationof the onset of the Great Depression,is the identificationproblem.Tobin (1965)famouslymade this point in his admiring review of Friedman and Schwartz's Monetary History of the United States (1963).Temin(1976)elaboratedit when he observedthat a negativeshock to output (his favouredcandidatebeing an autonomousfall in consumptionin 1929)could in principlehave led to the fall in money, ratherthan a negative shock to money causingthe fall in output. From this point of view, it is the depthand breadthof historicaldetail that renderFriedmanand Schwartz'saccount so compelling.In their Monetary Historythey used historicaldetail to make the case that the monetaryshock was autonomousand the fall in outputwas induced- that causalityeffectively ran frommoneyto output.They documentedthe Fed's concernwith excessive financialspeculation.They showed how this led the centralbank to raise the discountraterepeatedly,especiallyfollowingthe deathof BenjaminStrongand the ascent of Adolph Miller,a diehardbelieverin the dangersof speculative excesses.The implicationwas that moneywas not tightenedin responseto the evolutionof output.Rather,it was tightenedfor unrelatedfinancialreasons. Friedmanand Schwartznext arguedthat the untimelydeath of Strongand disputesamongregionalreservebanksand betweenthe reservebanksand the Federal Reserve Board preventedpolicy from respondingin its customary stabilizingfashion to the decline in industrialproduction.An inexperienced FederalReserveBoard had neverencounteredan analogouspeacetimerecession. Misinterpreting the real-billsdoctrine,predisposedto a perverseand selffulfillingliquidationalistview of the businesscycle, and preoccupiedby the belief that monetaryease might reignitespeculation,it saw inaction as the appropriatepolicy response (DeLong 1990; Calomirisand Wheelock 1998; Meltzer2002). Thus, monetarypolicy was slow to loosen as output declined. theoperaThis,then,wasevidenceof an autonomousmonetaryshock,reflecting the in turnencourages tionof a uniquesetof historical circumstances. Thisargument beliefthatmonetaryfactorsplayedan importantrolein theonsetof theDepression. But there are problemswith this U.S.-centredmonetaryinterpretationof the slump. While the tighteningof monetarypolicy through the summerof 1929 was not particularlydramatic,the subsequentdecline in activity was.1 1 Ritschl and Woitek (2000) document this in a forecasting framework. Their time series analysis suggests that the effects of the Fed's discount policy remained 'far too small to explain the collapse in output after 1929' (11). To put these changes in perspective, the Fed has typically (since 1951) raised the discount rate by 425 basis points between the trough and the peak of the business cycle (as dated by the NBER). By comparison, the tightening in 1928-9 was 350 basis points, hardly out of the ordinary. (This comparison assumes, of course, that the 1926-7 slowdown was a business cycle trough, a characterization that would not be universally accepted.) The implication in the text only follows, of course, if one agrees that the 'direct action' part of the Fed's initiative ultimately had little effect (which is my view). Understandingthe Great Depression 3 Recall that the Fed attempted to use 'direct action' to deny Wall Street credit for use in speculative activities. Rather than simply relying on the classic device of a higher discount rate to reduce the volume of credit for use in financing brokers' loans (making it more expensive for banks to obtain credit from the Fed and thereby forcing them to pass through that cost to their customers), it applied moral suasion to the banks. It hesitated to make more active use of the conventional instruments. Essentially, this is another way of saying that the Fed did not tighten particularly aggressively.2Thus, the first problem, to put it in the form of a question, is: How could this modest monetary tightening produce such a sharp contraction in the United States? If the Fed's modest monetary tightening was not obviously sufficient to explain the unusually sharp downturn in U.S. economic activity, then it was even less sufficient to explain the unusually sharp downturn in the rest of the world. The Institut fur Konjunkturforschung's index of world industrial production fell by 10% between 1929 and 1930, while the League of Nations' index of mining and manufacturing fell by 12%.3This is the second problem: it is not clear how modest monetary tightening in one country could have been responsible for a depression that engulfed the entire world. The third problem is related: a number of other countries - Canada, Germany, Poland, Argentina, Brazil, and Australia among them - turned down before the United States.4 The model of international transmission in traditional accounts - inspired by the IS-LM model with low capital mobility that was the workhorse for multi-country macroeconomic thought experiments in the 1960s and 1970s - is that the Fed's tightening led to a deceleration in growth in the United States, which in turn led the United States to import less, transmitting the slowdown to other countries. Clearly, the actual sequence of events is wrong for this story.5 The gold standard - and the fact that capital mobility was anything but low in the 1920s - helps to resolve all three paradoxes. Because capital was mobile and exchange rates were fixed, higher interest rates in the United States meant higher interest rates in the rest of the world as a result of interest arbitrage. If market interest rates rose in the United States, they also had to rise in other countries, as capital flowed toward the country or at least flowed out at a slower rate. If foreign central banks failed to follow the Fed in raising discount 2 Note that the sameproblemarisesin the immediatepost-crashperiod.The Fed loosened significantlyfollowingthe stock marketcrash,whichmakesit all the more surprisingthat the economycontractedas rapidlyas it did - althoughone can certainlyargue,with benefitof hindsight,that it did not loosen far or fast enough. 3 By comparison,the largestfall since 1971in the IMF's indexof worldrealGDP was 1.5%in 1982. 4 In Germany,for example,recentnationalproductestimatesshow that the peakwas reachedin 1928.Constructionpeakedin the summerof 1927,and the productionof consumergoods peakedin February1928.See Ritschl(1997, 1999). 5 U.S. merchandiseimportspeakin the firsthalf of 1929,not in the firsthalf of 1928.Whilethere is anotherupwardspikein U.S. importsin October1929,this appearsto be a seasonaleffect. 4 B. Eichengreen rates, they would suffer reserve losses, jeopardizing the stability of their currencies. Because Europe and Latin America had relatively weak balances of payments, reflecting the strengthened competitive position in international markets for manufacturing exports acquired by the United States during and after World War I and the weakness of agricultural and primary product prices in the 1920s, their central banks could ill afford reserve losses. Because their commitment to the gold standard was uncertain, reflecting the expansion of the franchise and the inability of populist governments to subordinate other goals of policy to the overarching imperative of exchange rate stabilization, the tendency for capital flows to reverse direction and now flow not from but to the United States raised doubts about the stability of currencies elsewhere.6 Central banks thus had to respond by raising rates even more sharply to calm skittish investors. Consequently, the restrictive turn in monetary policy in the United States provoked an even more restrictive turn in monetary policy elsewhere. The deceleration in the rate of money growth in 1928 was even faster in Latin America and Europe than in the United States. The same was again true in 1929. The only region where this was not the case was East Asia, not surprisingly, since Australia had already begun to tinker with its gold standard and Japan was not on the gold standard at all, except for a few months in 1930-1. Now consider again our three paradoxes. That the shock was a global monetary contraction, not just a U.S. monetary contraction, helps to explain why the U.S. economy turned down so sharply. Exports were the first component of aggregate demand to begin falling, indicative of the impact of the contraction in the rest of the world on the United States. That foreign central banks had to respond sharply helps to explain why foreign cycles peaked even earlier. The one factor that insulated U.S. economic activity from the effects of higher Federal Reserve discount rates, namely, the reversal of U.S. capital exports, further explains why other countries began to contract before the United States. It is revealing that countries whose business cycle peaks preceded that of the U.S. had been capital importers in the 1920s. And a global monetary shock works better than a U.S. monetary shock in explaining a global recession. Of course, other factors besides Federal Reserve policy set the stage for the Great Depression. Even if one focuses on rising interest rates and tightening monetary conditions as precipitating factors, interest rates in other countries could have risen for unrelated reasons; the tightening of monetary policy elsewhere, it could be argued, did not simply result from the international transmission by the gold standard of tighter money in the United States. There are at least some national cases where this objection has validity. For example, the 6 Obstfeld and Taylor (2002) invoke this connection between gold standard credibility and popular politics and provide supportive data for the interwar years. Flandreau, Le Cacheaux, and Zumer (1998) challenge the applicability of the thesis to the period before 1913, but that is a different question. Understandingthe Great Depression 5 operation of additional factors was clearly evident in France.7 The Poincare stabilization raised French real interest rates in the second half of 1926, as inflation came down but nominal interest rates remained stuck in positive territory. Higher real interest rates made France a more attractive place for portfolio capital; hence, funds flowed to Paris from other financial centres. But high real interest rates also made consumption and investment less attractive, strengthening the French current account. Lower nominal rates, for their part, stimulated the demand for money, which could be met only by running a payments surplus and attracting gold from the rest of the world, given the reluctance of the Bank of France to expand domestic credit.8 For its own reasons, then, France became an independent source of monetary deflation in the second half of the 1920s. Germany is another other important country for which this case can be made (Balderston 1993; Ritschl 1999). The president of the Reichsbank, Hjalmar Horace Greeley Schacht, became convinced of the dangers of a speculative bubble in the spring of 1927, even before similar fears infected the members of the Federal Reserve Board, leading the German central bank to tighten credit (Voth 2002).9 Schacht too resorted to a policy of direct pressure, meeting with the directors of the big Berlin banks to express his concerns. Tighter money and credit succeeded in pushing down the German stock market, as Schacht intended, but they also pushed down the German economy, which was presumably not his intention.10 Here too, then, we see policies of monetary tightening implemented for reasons largely unrelated to the decisions of the Federal Reserve. The other objection to the gold standard story is that it was not monetary policy but other shocks and imbalances that caused the slump. This returns us to the debate between Friedman and Schwartz and Temin; indeed, it returns us 7 As emphasizedin Eichengreen(1992)and Johnson(1998). 8 Whilethe Frenchcentralbank'srevisedstatutelimitedthe scopefor conventionalexpansionary open marketoperations,it still could haveexpandeddomesticcreditby unconventionalmeans suchas purchasingforeignexchangeon the open market.Preoccupiedwith inflationto the point of phobia,however,it hesitatedto do so. If this referenceto the needto use unconventionalmeansand to the centralbank'sneuroticpreoccupationwith inflationreminds readersof the Bankof Japanin the 1990s,my wordingwill have accomplishedits goal. 9 Schachtwasalsoconcernedaboutthe highlevelof foreignborrowing; he sawthe stockmarket boomin Berlinandexpansionin Germanyin generalas beingfuelledby unsustainable capital externalobligations,therebysetting inflows,whichaddedto the country'salreadyunsustainable theeconomyup for an evenmorepainfulfuturefall.Thedilemma,then,was the sameas that facinganycentralbankerconfrontinga problemof capitalinflows:higherinterestratesmight eventuallyprickthe stockmarketbubble,butin themeantimetheywouldonlyencouragefurther capitalinflows(as theydid,in the Germancase,untilthe Fedrespondedin 1928withfurther increasesin its discountrate).Schacht'spreferredsolution,whichis themedicinenowadays recommended for countriesexperiencing difficult-to-manage capitalinflows,wasfiscalrestraint, or Reich but thiswasnot somethinghe couldforceon free-spending municipalgovernments authoritiesfeelingpressurefroma varietyof specialinterestgroups. 10 This line of argumentalso addressesa problemnoted by Temin(1971),that the German economystartedturningdown even beforecapitaloutflowsfromthe UnitedStateswere curtailed. 6 B. Eichengreen to what was the debate in the 1930s.After experiencingthe 1990s,observers are perhapsless inclinedto find explanationsfor the downturnin an autonomous fall in consumptionspendingand more likely to considera role for the creditboom, the stock marketbubble,realestatespeculation,excessiveoptimism about the commercialpotential of information and communications technologies(in the 1920s, radio; in the 1990s, the Internet),and the naive belief that recessionshad become a thing of the past. The culpritcould have beentechnologicalchangecreatingstructuralimbalancesthat left the economy vulnerableto destabilizingshocks.Many old industrieswerefinanciallyfragile, and new ones like radio were not yet profitable.Thus, even a minor shock, whateverits source, could create financial stress. This is the story told by Bernstein(1987) and elaboratedby Szostak (1995). Alternatively,the culprit could have been overinvestmentin property,which left financialinstitutions exposed and bequeatheda legacy of inappropriatelysubdividedtracts that depressedproductivitywell into the 1930s(Field 1992). The problemwith these interpretationsis that they tend to be sector and countryspecific.Theyfocus on sectorssuchas radio,electricity,and residential constructionand on a particularcountry,typicallythe United States.It is not clear how sector-specificfactors could have produced such a pronounced economy-wideslump or how country-specificfactors could have produced such a sharp global downturn.And as an explanationfor global macroeconomic developments,they are weakened by the absence of a propagation mechanism. One way of attemptingto knit these stories together is by applying 'the Bank for InternationalSettlementsview' (BIS) of the 1990s (e.g., Borio, Fufine, and Lowe 2001; Borio and Lowe 2002; Vila 2002) to the 1920s.The BIS view is that pronouncedcredit booms set the stage for sharp economic downturns,often accompaniedby severefinancialdistress."Thereis, first, an upswing in activity, at least partially grounded in fundamentals.Whether becausethe exchangerate is pegged or for other reasons, such as a positive supplyshock,upwardpressureon wholesaleand retailpricesremainssubdued. Consequently,money and credit conditions remain accommodating.Ample supplies of liquidity animate financial markets, driving up asset prices. As lending by commercialbanks expands, increasinglyspeculativeinvestments are underwritten,and the quality of bank loans declines. Higher property and securitiespricesencourageinvestmentactivity,especiallyin interest-ratesensitiveactivitieslike construction.Here too, as the volume of investment activityrises, the averagequalityof projectsdeteriorates. 11 The BISviewhas at leasttwo significantprecursorsin the literatureon the interwarperiod:the Austrianinterpretation of the Depression,and the view that attributesthe Depressionto the stockmarketboomandcrash.TheAustrianviewtracesto the workof Mises(1924)andHayek (1925);see Robbins(1935)and Rothbarth(1975).Theroleof the GreatCrashis emphasizedby Galbraith(1972)and Kindleberger (1973). Understandingthe Great Depression 7 Duringthis upswing,the demandfor speculativeinvestmentsriseswith the supply, since, in the prevailingenvironmentof stable prices,nominalinterest ratesand thereforeyields on safe assets are too low to be attractive.In search of yield, investorsdabbleincreasinglyin risky investments.Their appetitefor speculativeinvestmentsis strongerstill to the extentthat these trendscoincide with the developmentof new technologies,networktechnologiesof promising but uncertaincommercialpotential in particular.Eventually,central banks begin tightening,the financialbubbleis pricked,and asset pricesdecline.The economy is then left with an overhangof ill-designed,non-viableinvestment projects,distressedbanks, and heavily indebtedhouseholdsand firms, all of which aggravatethe subsequentdownturn. This tale from the 1990shas obviousappealfor historiansof the 1920s.The 1920swas a decadeof cyclicalexpansion,reflectingrecoveryfromWorldWar I, new informationand communicationstechnologieslike radio, and new processeslike the productionof motor vehiclesusing assembly-linemethods. It was accompaniedby accommodatingmoney and credit conditionsin the United States, reflectingthe ample gold reservesaccumulatedby the country duringWorldWar I and financialinnovationsrangingfrom the development of the moderninvestmenttrust to consumercredit tied to purchasesof bigticketdurablegoods like the automobile.Ample suppliesof money and credit fueleda realestateboom in Floridain 1925,a Wall Streetboom in 1928-9,and a consumerdurablesspendingspreein the secondhalf of the 1920s.That these booms developedunderthe fixed exchangerates of the gold standardmeant that they generatedlittle inflationarypressureat home but also that their effects were transmittedto the rest of the world. The ready availabilityof moneyand creditproduceda visibledeteriorationin the qualityof projects- as in the Florida land boom of the mid-1920s- and a growing prevalenceof malfeasanceand graft, evident in the activitiesof CharlesPonzi in Florida, ClarenceHatry in London,and Ivar Kreugerin Stockholm.Thesechangesin turnheightenedthe concernof centralbanksaboutthe broadereffectsof assetprice inflation, leading them eventuallyto tighten. The not uncontroversial implicationis that the Fed should have preventedthe developmentof speculative excesses by maintaininga tighter policy stance, specificallybetween 1925 and 1927, despite the absence of wholesale and retail price inflation. Doing so might have obviated the need to tighten so sharply starting in 1928, by which time the financialposition of banks, firms, and households had weakenedsignificantly.It might have avoided precipitatinga downturn that banks,firms,and ultimatelythe economy,burdenedby excessesaccumulated in the RoaringTwenties,werein no position to withstand. Borio and Lowe (2002)developindicatorsof this creditboom phenomenon - constructedas a weightedaverageof the rate of growthof bankcredit,stock marketvaluations,and the investmentratio - and show that recessionsand crises are most likely to follow when these indicatorsrise relativeto trend. Togetherwith Kris Mitchener,I have constructedanalogousindicatorsfor the 8 B. Eichengreen 1920s.12 The idea of this composite is to capture not just the availability of credit to the private sector but also its transmission to asset prices and economic activity. The motivation is that the same increase in credit may have different effects depending on the structure of the economy that amplifies or muffles its impact. The composite indicator thus seeks to capture both the impulse and its amplification by measuring not only the growth of credit but also its impact on asset prices and aggregate demand. Figure 1 shows the unweighted average of the composite indicators for 16 countries for which the necessary data are available.13 As shown there, the composite indicator rises relative to trend after 1927, signalling turbulence ahead. Figure 2 shows that the larger the credit boom circa 1928 (i.e., the greater the rise in the actual value of the composite indicator, plotted on the horizontal axis), the larger the subsequent fall in per capita GDP (plotted on the vertical axis).14 We also see from the figure that the credit-boom - subsequent-depression relationship is above all a U.S. story. This is consistent with the older literature on the United States, emphasizing the downward shock to consumption and investment emanating from the collapse of the stock market and from overbuilding in the 1920s. But a number of other countries also experienced pronounced credit booms in the late 1920s - Canada, France, and Italy among them - and suffered the consequences subsequently. Credit booms had littered the nineteenth-century financial landscape, of course. Typically, a combination of lax domestic credit conditions and capital inflows fuelled a railway- or public-utility-related construction boom, investment in which was encouraged by government guarantees. Such stories feature prominently in Sprague's (1910) history of U.S. financial crises, and other countries, like Argentina in the 1880s, similarly displayed all the classic symptoms. Why, then, did none of these late-nineteenth-century episodes produce a slump as deep, long, and widespread as that of the 1930s? 12 While Borio and Lowe use measures of domestic credit to the private sector, comparable data are not available for the interwar years. We therefore use M2. Equity prices are measured as the ratio of a broad stock market index to the consumer price index, while investment is expressed as a share of GDP. The three ratios were then detrended by fitting a linear trend through 1930. The composite index is constructed by weighting the three ratios by their respective signal-to-noise ratios - that is, by the ratio of the share of subsequent crises successfully predicted by data through 1928 to the share of false positives, where the signalling threshold is set to maximize this ratio. When this is done separately for currency and banking crises, it yields slightly different composite indicators for the two cases, although the prevalence of twin crises in the 1930s dictates that the differences in the two variants are small. Unweighted averages of the three components of the composite yield essentially identical results in the forecasting exercises described below. 13 The countries are Argentina, Australia, Belgium, Canada, Denmark, Finland, France, Germany, Italy, Japan, the Netherlands, Norway, Spain, Sweden, the United Kingdom and the United States. 14 Multiple regressions, including controls for other country characteristics influencing their susceptibility to the international slump (e.g., the size of the current account deficit in 1928), do not weaken this relationship. The typical t-statistic on the value of the composite indicator in 1928 in such a multiple regression is consistent with statistical significance at the 95% but not the 99% confidence level. Understandingthe Great Depression 9 10- 5X a) 'ID c a) Q) 0 0- E 0 0 -5- -10I 1920 1925 1a CompositeIndicatorfor BankingCrises 1930 1935 date 10- x 0V 0- ._ o Q. E 0 ) -10- -20I I I I 920 1920 1925 1930 1935 1935 date 1b CompositeIndicatorforCurrencyCrises FIGURE 1 Composite Indicator for Currency Crises 10 B. Eichengreen 20- FRANCE y = 1.2676+ 0.8864 x t-stat:(0.31) (2.05) UNITED STATES CANADA AUENTIN ARGENTINA co CL 0CD 10- C: UNITED NORWAY oo3 Cu 0- 0) BELGIUM DENMARK -10- ~~I -5 JAPAN ~I i 0 I I I 1 15 1 10 5 1928 Composite Indicator 30- UNITED STATES Y= 3.6974 + 0.9696 x t-stat: (0.80) (1.96) CANADA FRANCE CO a) 20- ARGENTINAGERMANY a. (3 0) NETHERLAND AUSTRALIA o 10- SWEDEN ol, JAPAN FINLAND i\ SPAIN KINGDOM 0- ~I -5 ITALY BELGIUM NORWAY DENMARK I~ i 0 II 5 I I 10 1 15 1928 Composite Indicator FIGURE 2 The competing answers focus, predictably,on U.S. and foreign policies. Before 1913 Treasury operations were capable of some monetary fine tuning, but there was no central bank to behave in strongly destabilizingfashion. The prewar gold standard imposed automatic limits on the extent of credit fluctuations. In the Understandingthe Great Depression 11 1920s, however, an additional element of discretionwas introduced.The Fed kept interest rates too low for too long, allowing the credit boom to develop, before ratchetingthem up sharply.The key mistake in this view was at least as much that monetary policy was too loose before 1927 as that it was too tight later. The problem was not merely the gold standard, in other words; it was gold standard management. The limitation of this U.S.-centric story, once again, is that it does not explain why other countries responded in like fashion. It can explain the U.S. credit boom but not the global credit boom, unless we are prepared to assume that low interest rates in a country accounting for only a quarter of global output and 15% of world exports can explain financial developments throughout the world. Thus, foreign policies must also have been to blame insofar as other central banks similarly allowed interest rates to remain too low for too long. Countries like Britain were already suffering double-digit unemployment, and raising interest rates would have only added to the problems of a politically challenged Conservative government. Part of what the governor of the Bank of England, Montagu Norman, meant when he described the Bank as continuously 'under the harrow' was the political pressure he felt not to raise interest rates even when doing so was dictated by gold standard considerations.15 Germany, for its part, had a moral hazard problem: tighter credit that strengthened the balance of payments undermined the argument that the country was incapable of meeting its reparations obligations. The problem, in this view, was not so much misguided policy in the United States as the politicization of policy in other countries. Central banks' ill-conceived efforts to cooperate may have contributed to this situation. Benjamin Strong understood that his foreign counterparts, in particular Norman, were under pressure not to raise interest rates. If they failed to act, this might raise questions about the credibility of Britain's commitment to the gold standard. This was the motivation for the secret meeting of central bankers on Long Island in May 1927 - the 1920s analogue of 'the committee to save the world.' Rather than forcing the Bank of England to raise its discount rate, as dictated by the 'rules of the game,' the Fed agreed to lower its rate as a result of that meeting It was at this point, it is sometimes 15 For those who recall textbook accounts in which the instability of the 1920s was a result of misaligned exchange rates - and specifically of Winston Churchill's decision to restore sterling's pre-war parity in 1925 - this is where these factors enter the story. Sterling's overvaluation compounded the problem of British unemployment and balance-of-payments weakness. More generally, the failure of central banks to force prices back down to pre-war levels before restoring convertibility at the pre-war rate of exchange forced the interwar system to operate on a narrower gold basis - the real value of existing gold stocks would have been higher had price levels been lower - leaving less margin for error. The resulting global gold shortage continues to be emphasized by authors writing in the tradition of Cassel (1936); see, for example, Johnson (1998) and Mundell (2000). 12 B. Eichengreen alleged,that the creditboom and the Wall Streetbubblebegan to get out of hand. This bringsus to the role of centralbankcooperation.Flandreau(1997)and Moure(2002)have criticizedthe argument(associatedwith Eichengreen1992) that the inadequacyof centralbank cooperationin the 1920scontributedto the collapseof the interwargold standardand ultimatelyto the severityof the Great Depression.They suggest that such cooperationwas never especially prevalentbefore 1913,makingit hardto arguethat its failurein the 1920scan explainthe inferiorperformanceof the gold-exchangestandardreconstructed afterWorldWarI, and that to the extentthat cooperationactuallyoccurredin the 1920sit was partof the problem,not part of the solution.Not surprisingly, I remainunconvinced.For one thing, the criticsexaggeratethe role of internationalfactorsin explainingthe Fed's interestrate cut in 1927.The Fed was respondingto the slowdown in industrialproduction(as Henry Ford shut down his assemblylines to retool for the Model A) and to distressin the agriculturalmidwestdue to depressedfarm prices.Therewas a large element of unilateralismand self-interest,in otherwords,in the Fed's 1927interestrate cut, as opposedto internationalmotives.This is anotherreminderof the need to distinguishbetweenU.S. and internationalfactors. In addition,when analysingthe pre-1913gold standard,my emphasiswas on the importanceof 'regime-preserving cooperation.'The argumentwas not that centralbankshad cooperatedcontinuouslywhensettinginterestrates,but that the leadingcentral banks and governmentshad supportedone another with emergencyassistancein timesof crisis.This preventedthe collapseof any one currencyfrom jeopardizingthe entire system. It preventedpressureson particularcurrenciesfromprovokingsystemiccrisesof the sort that eruptedin the 1930s.It was this regime-preserving cooperationthat was missingin 1931, when the Bankof Francehesitatedto assistits Austrianand Germancounterpartsand the BIS was preventedfrom doing so for a time by the disputeover reparations,Germany'sprogram of building pocket battle ships, and the proposal for a Austro-Germancustoms union in contraventionof the VersaillesTreaty.Thereis no contradictionbetweenthe argumentthat the absence of regime-preserving cooperationwas part of the problemin the 1930s, but that misguidedcooperationof a differentsort was equallyproblematicin the 1920s. 2. The downwardspiral As output and asset pricesbegan to fall, banks experiencedfinancialdistress. Since borrowers'obligationswere fixed in nominalterms,the fall in incomes led to a rise in non-performingloans. Growingconcernsover the stabilityof the gold standardthen led investors to substitutegold for financial assets where doing so was permittedand central banks to liquidate their foreign Understandingthe Great Depression 13 exchange reserves to avoid suffering capital losses in the event that reserve currencies were devalued. Under these circumstances, the gold standard became an engine of deflation. Money supplies were a multiple of the international reserve backing of central banks and governments. This backing contracted as central banks, fearful for the stability of the gold standard, liquidated their foreign exchange reserves. Central banks whose obligations were reserve currencies, suffering corresponding reserve losses, raised their discount rates to stem capital outflows.16 Higher interest rates heightened the distress experienced by commercial banks with maturity mismatches on their balance sheets. They consequently sought to rebuild their liquidity and reserves by calling in loans. Depositors, with fears about the liquidity of the banks, fled into currency. Bernanke and Mihov (2000) decompose this process by writing the money stock (M1) as follows: M1 = (M1/BASE)*(BASE/INTRES)*(INTRES/GOLD)*GOLD, where BASE is currency in circulation plus the reserves of commercial banks, INTRES is the country's international reserves (gold plus foreign exchange), and GOLD is the value of gold reserves denominated in domestic currency. We can take the second term as fixed in the short run by the backing rules under which the gold standard operated and the fourth term as slow to change, given the lagged response of the gold-mining industry. Thus, most of the action in this product should have come from the first and third terms. The first term, the money multiplier, will fall as the worsening prospects of financial institutions lead depositors to flee into cash and banks to call in their loans. The third term, the international reserve multiplier (total international reserves relative to gold), will fall as central banks liquidate their foreign currency assets. How important were these components in producing the observed decline in money stocks? The INTRES/GOLD ratio fell from 1.27 at the end of 1929 to 1.25 at the end of 1930 and 1.12 at the end of 1931 in 26 countries (the 24 countries considered in Nurkse 1944, appendix II, plus the United States and 16 As the Fed did, for example, in September-October 1931 following Britain's abandonment of the gold standard - see below. Say that the National Bank of Belgium sought to liquidate its holdings of British treasury bills and to replace them with gold in order to avoid suffering capital losses on the former. It would sell those bills on the British market for sterling and present that sterling to the Bank of England for conversion into gold. The Bank of England, experiencing gold losses, would then have to raise interest rates to limit the deterioration of its position. The National Bank of Belgium, to minimize its own reserve losses, would presumably respond in kind. The net effect would be a decline in reserve backing (since foreign exchange reserves had declined but the amount of gold in the international system was fixed) and higher interest rates all around - two indications of deflationary pressure. In fact, the Belgian case is an interesting one. The British authorities lobbied the National Bank to hold onto its sterling, as a result of which the Belgian authorities suffered extensive capital losses when sterling was ultimately devalued. 14 B. Eichengreen the United Kingdom). Other things being equal, this implies a 12% fall in global money supplies centred in 1931. If we exclude the United States and the United Kingdom, the principal reserve-currency countries (the ratio being fixed at unity in reserve-currencycountries that issued rather than held foreign exchange reserves), then INTRES/GOLD falls from 1.60 at the end of 1929 to 1.53 at the end of 1930 and 1.23 at the end of 1931. In other words, for countries other than the reserve centres, the corresponding fall in money supplies was 23%. While the INTRES/GOLD ratio fell across the board (except in the United States and the United Kingdom for the aforementioned institutional reasons), the direction of the change in M1/BASE varied across countries. Bernanke (1995) analyses data for six countries, of which the largest fall was in the United States (25%), but in two of which (the United Kingdom and Sweden) the ratio instead rose between the end of 1929 and the end of 1931.17 The weighted average for Bernanke's six countries (weighted by 1929 money supplies expressed in U.S. dollars) declined by 21%. The weighted average, excluding the U.S., was 11%. I undertook a broader comparison for a total of 31 countries; in this larger sample, the weighted average decline, excluding the United States (30 countries), was 10%.18 Evidently, then, the liquidation of foreign exchange reserves played a more important role than the liquidation of bank deposits in the monetary contraction outside the United States. The United States was unusual in that the percentage fall in the M1/BASE ratio was larger than the percentage fall in the INTRES/GOLD ratio.19 (Recall that as the issuer of a reserve currency, the Fed, like the Bank of England, held no foreign-exchange reserves.) It is tempting to say that for the rest of the world the collapse of the gold-exchange standard mattered more than the collapse of the banking system, while the opposite was true for the United States. It is not surprising, from this point of view, that the literature on the global depression focuses on the gold standard, while that on the United States concentrates on the banking crisis. One should be cautious here, since banking and currency crises were related. Currency crises in other countries were one factor forcing the Fed to ratchet up interest rates in the final months of 1931 (both devaluation and higher interest 17 Of course,theseweretwo of the countriesthat went off the gold standardtowardsthe end of 1931.But the othertwo countriesconsidered,Polandand Belgium,did not experiencedeclines in the M1/BASEratioapproachingthat of the UnitedStates. 18 The 30 countrieswereBelgium,Bolivia,Brazil,Bulgaria,Canada,Chile,Colombia, Czechoslovakia,Denmark,Estonia,Ecuador,Finland,France,Germany,Hungary,Italy, Japan,Latvia,Lithuania,Mexico,Netherlands,Norway,Poland,Portugal,Romania,Spain, Sweden,Switzerland,the UnitedKingdom,and Uruguay.Again,I took the weightedaverage by convertingM1 and the base into dollarsat marketexchangerates. 19 In the United Kingdom, in contrast, the M1/BASE ratio rose slightly between the end of 1929 and the end of 1930and fell slightlybetweenthe end of 1930and the end of 1931,hence remainingessentiallyunchangedover the period.This stabilityreflectsthe absenceof serious banking-sector problemsin Britain,an issue to whichI returnbelow. Understandingthe Great Depression 15 rates abroad increased the perceived likelihood that the dollar might have to be devalued, forcing the Fed to raise interest rates to stem capital outflows). Those higher interest rates surely contributed to the difficulties of the U.S. banking sector. In other countries, the banking crisis worsened the liquidity crisis. All of the banking-crisis episodes identified by Bernanke and James (1991) begin before the corresponding currency crises in the same countries. Note that the same pattern - banking crises first, currency crises second - is typical of twin crises today (Kaminsky and Reinhart 1998). When deposits hemorrhage out of the banking system, governments and central banks find themselves between a rock and a hard place. In the 1930s providing large amounts of liquidity to the banking system almost certainly would have violated gold-standard statutes; doing so would have raised questions about whether the authorities attached priority to the maintenance of the exchange rate peg relative to other economic and social goals. But not doing so might have allowed the banking system to come crashing down, with even more disruptive effects on economic activity. Except in the United States, where they remained strangely unperturbed, the authorities generally took steps to prevent the collapse of banking systems. But adding liquidity to the financial system undermined confidence in the stability of the currency, provoking capital flight and encouraging the continued conversion of foreign exchange reserves into gold, thereby accelerating the fall of the gold-reserve multiplier. Again, it follows that the main engine of deflation was the banking crisis in the United States and the currency crisis in other countries. A conventional critique of the literature on twin crises is that timing does not prove causality.20That a banking crisis becomes evident before a currency crisis does not prove that the former caused the latter. It could be that anticipations of a subsequent currency crisis led savers to liquidate domesticcurrency-denominated deposits and flee into foreign exchange in order to avoid capital losses due to the eventual devaluation. This of course was the dynamic evident in Argentina in the final months of 2001, although deposit withdrawals were not enough to provoke a full-blown banking crisis prior to the actual devaluation. Ferguson and Temin (2001) show that Germany was running chronic budget deficits in the first half of 1931 and reason that expectations that these would be monetized, leading to a currency crisis, provoked the deposit withdrawals of May and June.21Wigmore (1984) argues that anticipations of Roosevelt's decision to take the dollar off gold led to the run on U.S. banks in the interregnum between FDR's election and assumption 20 Recall that this is the same critique of timing evidence on money and output alluded to in the preceding discussion of Friedman and Schwartz's (1963) examination of monetary factors in the onset of the Depression. 21 The authors argue that Bruening's references to a reparations moratorium worked in the same direction. Of course, there is a sense in which the reparations tangle and German budget deficits were two sides of the same coin; strengthening the budget would have weakened the argument that Germany could not pay. 16 B. Eichengreen of office. Thejudiciousconclusion,then as now, is that the causalitybetween bankingand currencycriseswas bidirectional. If expectationsof depreciationcould destabilizea bankingsystem,then the fact of depreciationcould be broadlystabilizing.Going off the gold standard freed up monetarypolicy. It allowed the central bank to cut interestrates, relieving banking sector distress. Grossman (1994) contrasts a number of explanationsfor why some countriessucceededin avertingbankingcrises in the 1930s.The singlemost importantfactor, he finds, was whethera country had abandonedthe gold standard,allowingthe centralbank to engage more freelyin lender-of-last-resort operationsand to jump-startthe recovery.This seems to have been the case in capital-importingand capital-exportingeconomies and industrialand developingcountriesalike. This situationis in contrastto the experienceof the 1990s, when capitalimportingemergingmarketssaw theirfinancialsystemsand economiesdestabilized by depreciation.Why didn't depreciationin the 1930s have similarly destabilizingeffects?One can think of severalpotentialexplanations. * Foreign deposits were widespread but foreign-currency-denominated deposits were less. This was at least one favourablelegacy of the gold standard,insofar as a history of fixed rates made it seem less imperative for foreigndepositorsto avoid currencyrisk. * Foreign-currency denominatedloansto the banks'domesticclientsweresimilarlyless prevalent.Hence,depreciationdid not destroythe balancesheetsof loans. the corporatesectorand aggravatethe problemof non-performing * Not a few countrieswith foreignobligationsrespondedto exchangemarket pressuresby slappingon currencyand exchangecontrols - in effect, they respondedlike Malaysiain 1998.This allowedthe authoritiesto injectcredit into the bankingsystemwithout precipitatingthe uncontrolledcollapse of the currency.Decouplingfrom internationalfinancialmarketsmade more sensein the 1930sthan the 1990sbecausethe marketfor new foreignissues was effectivelyshut down. * Finally, a number of countries substitutedclear and coherent monetary policy operatingstrategiesfor the gold-standardanchor. Swedenadopted a priceleveltargetnot dissimilarfromthe inflationtargetingregimethat has becomea popularapproachto anchoringfloatingexchangeratesin recent years. Britainadopted a dirty float consistentwith an exchangerate that remainedstable over the intermediaterun. The Commonwealthand many of Britain'stradingpartnersadopteda policy of followingthe pound, albeit more loosely than before,formingwhat came to be known as the Sterling Area.Thus,fearsthat abandoningthe gold standardwouldunleashanother round of uncontrolledinflation were quickly dispatched.This helped to reviveconfidencein bankingsystems. All this assumesthat the gold standardwas a bindingconstrainton stabilizing interventionand that its abandonmentwas a preconditionfor the adoption Understandingthe Great Depression 17 of reflationary policies. Any attempt to reduce interest rates or to engage in expansionary open market operations unilaterally would have precipitated capital outflows and reserve losses that endangered the exchange rate peg. Here was where the absence of international cooperation was an issue. A unilateral interest rate reduction would cause capital to flow towards other, higher interest rate markets, endangering the exchange rate of the initiating country, but there was no reason why simultaneous reductions by several countries would weaken any one currency relative to another. The United States was the one country arguably not prevented from taking unilateral monetary action. Controversy here centres on the open market purchases of April-August 1932 and whether they were abandoned as a result of fears that continued security purchases would lead to further reserve losses, to the point where the dollar would be attacked.22 Bordo, Choudhri, and Schwartz (1999) simulate a calibrated price-specie flow model and conclude that further increases in domestic credit, undertaken in either October 1930 February 1931 or September 1931 - January 1932, would not have exhausted the Fed's gold reserves. Their analysis assumes a linear relationship between domestic credit creation and reserve losses of a sort that will be familiar to readers from the monetary approach to the balance of payments. The authors raise but rule out the possibility of a speculative attack, which, in balance-ofpayments crisis models like Krugman (1979), produces a sharp non-linearity when reserves fall to a lower threshold. In effect, they rule out a priori the central problem of concern to those who see fears for the stability of the dollar as constraining the Fed. Hsieh and Romer (2001) argue that there was little perception, in either the markets or the corridors of the Fed, that a speculative attack was looming. They support this conclusion with Einzig's (1937) data on forward foreign exchange rates; they show that the forward discount on the dollar against the French franc implied a maximum probability of 11% of a major U.S. devaluation in the summer of 1932.23They show that the forward discount displayed no significant correlation with the magnitude of open market operations and that the behaviour of forward rates implies that expectations of a major dollar devaluation actually dropped during the latter part of the Fed's expansionary program. Sceptics will want more evidence on the volume of activity in the forward market and on how Einzig's forward quotations were constructed. 22 Thereis less disputeover the two otherperiodswherethe gold standardallegedlyinhibitedthe pursuitof moreexpansionarypolicies.The firsttime was in 1931,followingBritain's devaluation,whenthe Fed had to hold gold to backnot just its own liabilitiesbut also governmentsecuritiesin its portfolio;thiswas the problemof freegold eliminatedby the GlassSteagallAct of February1932.No one to my knowledgehas disputedthat the Fed neededto raiseratesin the wake of Britain'sdevaluationin orderto maintainconfidencein the dollar. The secondtimewas in January-February 1933,whenexcessreservesfell to zero, and thereis no questionthat the gold standardbound. 23 Theyobtainbasicallythe sameresultwhetheror not theyadjustthe observedforwarddiscount for meanreversion,in the mannerof the modem exchangeratetargetzone literature. 18 B. Eichengreen There is also the question of whether the Fed aborted its program for fear of destabilizing the dollar even before a larger discount was allowed to emerge, and whether expectations of this action were what caused the discount to narrow in July. Hsieh and Romer's qualitative discussion speaks to this question; they focus on the Harrison Papers, which do not suggest significant concern within the Fed about the danger to the dollar. But Sumner (1997) provides qualitative evidence that points in the opposite direction. This controversy is one more sign that the United States was different. It is another indication that the gold standard mattered less while banking crises mattered more for the United States in this, the second stage of the Great Contraction. 3. The recovery This perspective implicating monetary blunders and constraints in the onset of the slump suggests a role for corrective monetary action in the recovery. Just as the gold standard prevented central banks from unilaterally pursuing expansionary monetary policies, abandoning the gold standard could have allowed them to impart an expansionary monetary impulse. Fiscal policy played a negligible role in recovery, even in countries like the United States, the United Kingdom, and Sweden, where the Keynesian revolution received the most intellectual play.24 Monetary policy did the hard lifting. In some cases, like the United States, the monetary authorities simply cut interest rates and allowed the exchange rate to decline; they accommodated the increase in the demand for money and credit by passively discounting or leaving capital inflows unsterilized. The U.S. case probably overstates the typical degree of monetary stimulus from devaluation, since the country was also on the receiving end of capital flight from Europe once war clouds began to darken. On the other hand, devaluation probably helped less insofar as the United States was a large open economy (for smaller economies, there would have been no offsetting negative impact on the rest of the world and hence on export demand). In other cases, the authorities actively expanded domestic credit but did so cautiously, given memories of high inflation in the 1920s, the last time the gold standard had been abeyance. This reluctance to expand more aggressively, while understandable in the circumstances, is one reason why the recovery was not more robust. But wasn't monetary policy rendered impotent by the liquidity trap? In the 1930s nominal interest rates fell to low levels, especially the United States. With rates so low, it hardly made sense for banks to lend, as opposed to 24 See Romer (1992) on the United States and Jonung (1981) on Sweden. The same was also true of emergingmarkets (see Della Paolera and Taylor 1998). The main exceptions were Germany and Japan, in which recoverywas stimulatedby public expenditureon rearmament.Towards the end of the 1930s,as militaryconflict loomed, stimulus from rearmamentbecame general(see, e.g., Thomas 1983 on Britain).But this occurredafter the period of principalconcern here. Understandingthe Great Depression 19 holding excess reserves. Expansionary open market operations that created additional Federal Reserve obligations to the private sector might simply go into additional bank reserves and, given where the banks held their reserves, back into the Fed. Buying bonds 'would simply increase the reserves of the banking system by the amount of government bonds which were purchased with currency. The currency would go out.. .but [it] would immediately go into the banks and from the banks into the federal Reserve Banks' (U.S. Congress 1935, quoted in Wilcox (1984), 1). In the 1930s, as in the 1990s, whether the supply or the demand for bank loans was the binding constraint mattered greatly for effectiveness of policy. If the problem was less that no one was willing to lend, given the low-interest-rate environment, than that no one was willing to borrow, given deflationary expectations and demoralized business conditions, then a monetary shock could transform the situation. A sharp change in the exchange rate was an obvious way of transforming expectations of future prices and policies.25 Allowing the exchange rate to depreciate by, say, 25% (the magnitude of a typical devaluation) implied a substantial rise in import prices and ultimately in the domestic price level. In the short run, before prices had completed their adjustment, the change in the exchange rate also had the effect of shifting demand from foreign to domestic goods. If the authorities wished to prevent the currency from rebounding too strongly from its initial drop, they could undertake unsterilized intervention (Britain's approach following the establishment of the Exchange Equalisation Fund), engage in expansionary open market operations, or re-peg the currency to gold at its new lower level (as the United States did in January 1934), allowing the now higher demand for money and credit to be met by capital and gold inflows.26 In this case, the fall in the exchange rate was an unequivocal signal that future prices would be higher than current prices; by transforming deflationary expectations, it pushed down the real interest rate and stimulated borrowing for investment. Temin and Wigmore (1984) show how devaluation of the dollar led to a sharp rise in activity in capital-goods-producing industries in a matter of months, just as if such a transformation of expectations had taken place. The literature on Britain's devaluation and recovery points in a similar direction; it 25 Similarly,this is the conclusionof the modem literatureon the liquiditytrap.Evenif the interest-ratechannelfor monetarypolicyis immobilizedby the lowerboundon the nominal rate,a reflationarystrategyof pushingdownthe exchangerateandcommittingto a futurepath for the foreignexchangevalueof the currencycan still help an open economyto escapethe liquiditytrap(Svensson2000;McCallum2001). 26 Svensson(2000)observesthat peggingthe exchangeratein an environmentof deflationary expectations(whenit is, if anything,expectedto appreciate)may requirearbitrarilylarge foreignexchangemarketinterventions.Peggingto gold thus may requirearbitrarilylarge purchasesof gold. In practice,it was throughdaily purchasesof gold at progressivelyhigher prices(setby FDR and his 'kitchencabinet'overthe president'sbreakfast)thatthe dollar'srate of exchangewas drivendown to lowerlevelsafterApril 1933. 20 B. Eichengreen emphasizesthe recoveryof residentialconstruction(the housing boom), and what is residentialconstructionif not an investmentactivity?Nominal interest rates declined still furtherstartingin 1932, but this did not prevent British banks from lending. Rather, the fall in real interestrates as expectationsof deflationwere eliminatedstimulatedthe demandfor bank credit. Both cases thus suggestthat the failureof the demandfor bank loans, not any liquiditytrap-relatedobstacle to their supply, was the problem in the Depression.27 Similarargumentscan be madefor othercountries;for example,Della Paolera and Taylor(1998)show that abandonmentof the gold standardled to a sharp changein expectationsin Argentina,again consistentwith the effectivenessof monetary-cum-exchange-rate policy. All these cases suggestthat this problem could be solved by a devaluationthat convincinglytransformedexpectations of future prices and monetarypolicies and gave the authoritiesthe leeway neededto validatethose expectations. Once devaluationtransformedexpectationsand freedup monetarypolicy, prices began to rise, or at least they stopped falling. Eichengreenand Sachs (1985) identifieda numberof channelsthrough which the effects were felt. Raisingimportand exportpricesstimulatednet exports.Increasedprofitability and sales, futureas well as current,raisedthe value of productivecapacity relative to its replacementcost, encouraginginvestment- which, as noted, tended to lead the recovery in countries pursuing this policy. Stabilizing outputpricesmade productionmore profitableand stimulatedlabor demand. Bernankeand Carey(1996)focus on this aspectof the response.They observe, on the assumptionthat differencesin economic performanceas of the mid1930sweredue to differencesin aggregatedemandflowingfromgold standard policiesand that producerswerecontinuouslyon theirlabourdemandcurves, that the cross-sectionrelationshipbetweenrealwagesand productionidentifies a componentof the aggregatesupply relation. If the main thing happening in the 1930swas shocks to demandcoming from domesticand international monetarypolicies, in other words, then the resultingshifts in the aggregate demandscheduleenable us to trace out the aggregatesupplycurve.That real wagesfluctuatedcountercyclically(in contrastto the periodsbeforeand since, when their cyclical propertiesare more ambiguous)is consistent with the premise,common to these studies, that the Great Depressionwas mainly a demand-inducedphenomenon, with aggregate demand shocks driving the economyup and down along its aggregatesupplycurve. These conclusions have not gone unchallenged.Figure 3, adapted from Cole, Ohanian,and Leung (2002), shows that the cross-sectionrelationship betweenthe changein real wages and the changein output over the 1929-33 period is relativelydiffuse. Superimposedon the scatterplot is their labour 27 Of course,factorsotherthan the liquiditytrap(adversebalance-sheeteffectsdue to deflation, the destructionof informationdue to bankfailures,the eliminationof thicknessexternalities) could havedisruptedthe supplyof financial-intermediation services,as emphasizedby Bernanke(1983)and furtheranalysedbelow. Understandingthe Great Depression 21 o 0.2-0- - o_0 o0O - tzerland.... 0.0-. ..........Sw ... i - ++ ..... ........... + 0.0 0.6.... 0.2 . 0-0.2. {). .....................+ LRWCH 3 The real wage-output relationshi, FIGURE -J o 0.6 ... ..... ........ Netherlands 0.4 France i ,= + ............... ? o0. 2 US: -0.6 -0.2 0.0 0.2 0.4 0.6 Proportional ChangeinRealWages,1929-33 LRWCH 1929-33 FIGURE3 Therealwage-output relationship, demand schedule, derived on the assumption of a Cobb-Douglas production function with a labour share of two-thirds. (The schedule goes through the origin and has an elasticity of negative one-half, such that labour's share is constant.) In some countries, notably the United States, the fall in output is larger than the rise in real wages would lead one to predict. In others, notably France, Holland, and Switzerland, it is smaller. Indeed, assuming, as we have done so far, that the Great Depression resulted from demand-side factors alone, it is hard to imagine any plausible production function that would be consistent with the fact that output fell so much more in some countries than in others experiencing roughly comparable increases in real wages. The implication would seem to be that the observed real-wage/employment relationship was simultaneously perturbed by supply-side disturbances. This interpretation is attractive if it is possible to identify plausible supply shocks in the countries concerned. As noted, four countries in Cole, Ohanian, and Leung's 17-country sample lie especially far from the predicted real-wage/ employment relationship. The United States lies far below, while France, the Netherlands and Switzerland lie far above. For the United States, the obvious factor disrupting aggregate supply was the financial crisis.28This emphasis on 28 Conceivably supplemented by the tendency for New Deal policies to restrict competition and push up production costs, although the comparison of 1933 with 1929 is a bit early for the latter set of effects plausibly to have made themselves felt. 22 B. Eichengreen the collapse of financialintermediationas a supply shifteris consistentwith Bernanke'semphasison the non-monetaryeffectsof the financialcrisisand on the impactof increasesin the cost of creditintermediation.Earlyanalysestreat bankfailuresmainlyas a determinantof aggregatedemand,operatingthrough the monetarychannel.The presentperspective,in contrast,suggeststhat the bankingcrisis had an impact on both blades of the aggregate-supply/aggregate-demandscissors. As noted, France,the Netherlands,and Switzerlanddid better,in termsof the change in output, than the rise in their labour costs would lead one to predict.As membersof the gold bloc, they wereon the receivingend of capital inflows once the gold-standardparitiesof other countriesbegan to crumble. This renderedtheir financialsystemsmore liquid and creditconditionsmore accommodatingthan would have been the case otherwise.29This, then, could have amountedto a positivesupplyshock - a mirrorimage of the U.S. case. Thatsaid,noneof thesecountriesentirelyescapedfinancialsectorproblems.In France,a majordepositbank,theBanqueNationalede Credit,collapsedin 1932.30 In Switzerland, the BanquePopulairehad to be restructured in November1933. Thatmost of theseeventsoccurredrelativelylate in the 1929-33periodsuggests that the longerthesecountriesstayedon the gold standard,the morefragiletheir financialsystemsand economiesbecame.An implicationis thatthesethreemembersof the goldblocwouldnot haveperformedso unusuallywellin a longer-term comparison,whilethe UnitedStates,now off the gold standardandableto put its banking-sector problemsbehindit, would not have performedso exceptionally This poorly. predictionis borneout for the UnitedStates,as shownin figure4, whichis the equivalentof figure3, exceptthat 1929is comparedwith 1935rather than1933.U.S. outputis now 30%belowpredictedlevels,ratherthanfully45%,a considerable injust two years.Bordo,Erceg,and Evans(1997)note improvement that standardmodelsof monetary-induced inflationand deflation,togetherwith stickiness due to contracts,cannotexplainwhythe UnitedStates wage overlapping recoveredas quicklyas it didafter1933,giventhattherewasno significantdecline in realwages.Thespeedof recoverybecomeseasierto understandonceone admits a role for the restorationof financialstabilityand reintermediation due to the loosercreditconditionsbroughtaboutby capitalinflows. France,the Netherlands,and Switzerland,on the other hand, remainpositive outliersby about the same extent in 1935 as in 1933. But by 1935 these countrieswere no longer on the receivingend of gold and capitalinflows, as worries mounted about the sustainabilityof their gold standardpegs. This suggests that there must have been additional factors behind the fact that output in the gold bloc countriesdid not collapse more dramatically.Cole, 29 Cooperand Ejarque(1995)providea modelin whichcapitalflows like thesecan affectthe but, as theynote, their supplyside,owingto thicknessexternalitiesin financialintermediation, modelcan reproduceonly some of the macroeconomicfluctuationsof the period. 30 Therewerealso threebankfailuresand runson provincialbanksin 1930,althoughBouvier (1984)questionswhethertheseeventsconstituteda bankingcrisis Understandingthe Great Depression 23 0 .2 . 0.2 Q) A ........... . ^..... .....................--.......| - 0.0-- . + \ ..... ..... >ic \ .r?~~~ SwitzeFland - i + + ~Netherlands ..................... +..... -0.2- + + ...... \ 0 U a. -0.4 -0.2 + 0 France + \ 0.2 0.4 0.6 ProportionalChange in Real Wages, 1929-35 LRWCH35 FIGURE 4 The real wage-output relationship, 1929-35 Ohanian, and Leung encourage one to search for additional positive supply shocks to these countries. Finding them is not easy, especially in France. In early 1935 the newly formed Flandin Government, inspired by Roosevelt's New Deal, encouraged the adoption of legislation to cartelize French industry. The subsequent year was one of growing political tension, culminating in the electoral victory of a leftist government, the adoption of statutes mandating a reduction in weekly hours, and sharp increases in unit labour costs. It is hard to believe that there was a sharply positive aggregate supply shock lurking behind these adverse developments. The other way of interpreting figures 3 and 4 is that firms were thrown off the notional demand curves for labour to which Cole, Ohanian, and Leung assume that they clung faithfully and continuously throughout the period. Whereas the New Classical analysis of these authors assumes that a reduction in real wages would have led firms to hire additional workers, since the reduction in costs would have made it profitable to produce and sell more output, disequilibrium analysis in the tradition of Barro and Grossman (1976) suggests that even if firms had produced more, they might have found themselves unable to sell it. In this view, countries lay to the left or the right of the neoclassical labour demand curve plotted in figure 3 not because of adverse and favourable aggregate supply shocks, but because producers were pushed off their notional labour demand curves by different degrees of effective demand failure in different countries. In the United States, where the policy failure and therefore the demand failure were greatest, the lingering impact on employment was still evident in 1935. In contrast, in France, where the Flandin 24 B. Eichengreen governmentpursuedexpansionaryfiscal initiativesto head off politicalopposition on the left and a high level of reservesenabledthe Bank of Franceto maintaininterestrate at artificiallylow levels, the consequenceswere least. Ultimately,then, evidencelike that in figures 3 and 4 does not permitus to determineonce and for all whetherthe GreatDepressionis entirelya story of demandfailure, or whetherthere also were importantcompoundingsupplyside factors associatedwith the collapse of the financialsystem. More likely than not, the answeris not either/or,but both. 4. Conclusion One way of characterizingrecent researchon the Great Depression is as 'normalscience.'There is now a steady streamof publicationsadding incrementallyto existingknowledge.The onset of the Depressionremainsthe stage about whichthereis probablyleast agreement,perhapsunavoidablygiven the limited ability of macroeconomiststo explain turning points. The debate continuesto revolvearoundthe relativeimportanceof inappropriatenational policiesversusan unstableinternationalsystem.If thereis anythingresembling a consensus,it is a syntheticview, which admits a role both for monetary policy blundersin the United States, Germany,and France but also for the unstable internationalmonetary and financial system in amplifying these negativeimpulsesand transmittingthem to the rest of the world. It explains the speedand extentof the subsequentoutputdeclinein termsof both banking crisesand the collapseof the gold exchangestandard.It explainsthe eventual recoveryin termsof the abandonmentof the gold standard,which facilitated the pursuitof stabilizingmonetarypolicies,but also in termsof the restoration of stabilityto bankingand financialsystems. The challengefor proponentsof this consensusview is that differentelements dominatedin the United States and other countries.In the case of the United States,thereis no denyingthe role of policymistakesin the onset of the Depression,whereasfor other countriesinternationaltransmissionvia capital and gold flows playedthe moreimportantpart.Whereasthe bankingcrisiswas the main motor for the downwardspiralin the United States,in other countries the disintegrationof the gold-exchangestandardhad more profound effects. For most countries,the movement of wages and prices can largely explain the course of slump and recovery,but industrialproductionin the United States fell more rapidlythan the behaviourof real wages would lead one to predict,and it remaineddepressedrelativeto the internationalnormfor severalyearsfollowingdevaluationof the dollar. One explanationfor the disputesthat characterizethe literatureis thus that Americanistscontinue to export the conclusions of researchon the United Statesto the restof the world,whileothercountryspecialistscontinueto do the opposite. The challengefor those seekingconsensusis thereforeto recognize Understandingthe Great Depression 25 that national experiences differed while at the same time not abandoning the effort to place individual country experiences in a broader framework. In this paper I have sought to demonstrate that this effort to reconcile these heretofore competing strands of thought may not be in vain. References Balderston,Theo (1993) The Originsand Course of the GermanEconomicCrisis, 1923-1932.Berlin:Haude& Spener Barro, Robert, and Herschel Grossman (1976) Money, Employmentand Inflation. Cambridge:CambridgeUniversityPress Bernanke,Ben (1983)'Nonmonetaryeffectsof the financialcrisisin the propagationof the GreatDepression,'AmericanEconomicReview73, 257-76 -(1995) 'Themacroeconomicsof the GreatDepression,'Journalof Money,Creditand Banking27, 1-28 - (2000)Essayson the GreatDepression.Princeton:PrincetonUniversityPress Bernanke,Ben,and KevinCarey(1996)'Nominalwagestickinessand aggregatesupply in the GreatDepression,'QuarterlyJournalof Economics111, 853-83 Bernanke,Ben, and HaroldJames(1991) 'The gold standard,deflationand financial crisesin the Great Depression:an internationalcomparison,'in FinancialMarkets andFinancialCrises.ed. R. Glenn Hubbard.Chicago:Universityof ChicagoPress Bernanke,Ben, and Ilian Mihov (2000) 'Deflation and monetarycontractionin the Great Depression:an analysis by simple ratios,' in Ben Bernanke,Essays on the GreatDepression.Princeton:PrincetonUniversityPress, 108-160 Bernstein,Michael (1987) The Great Depression:Delayed Recoveryand Economic Changein America,1929-1939.Princeton:PrincetonUniversityPress Bordo, Michael, Ehsan Choudhri, and Anna Schwartz(1999) 'Was expansionary monetarypolicy feasible during the Great Contraction?An examinationof the gold standardconstraint,'NBER WorkingPaperNo. 1725(May) Bordo,Michael,BarryEichengreen,DanielaKlingebiel,and SoledadMariaMartinezPeria(2001)'Is the crisisproblemgrowingmore severe?'EconomicPolicy 32, 51-82 Bordo, Michael,ChristopherErceg,and CharlesEvans (1997) 'Money, sticky wages, and the GreatDepression,'InternationalFinanceDiscussionPaperNo. 591, Board of Governorsof the FederalReserveSystem(September),Washington,D.C. Borio, Claudio,and PhilipLowe (2002)'Assetprices,financialand monetarystability: exploringthe nexus,'unpublishedpaper,Bankfor InternationalSettlements(March) Borio, Claudio,Craig Furfine,and Philip Lowe (2001) 'Procyclicalityof the financial systemand financialstability:issuesand policyoptions,'BISPapersNo. 1, BIS,Basel Bouvier,Jean(1984)'TheFrenchbanks,inflation,and the economiccrisis,1919-1939,' Journalof EuropeanEconomicHistory13, 29-80 Calomiris,Charles,and DavidWheelock(1998)'Wasthe GreatDepressiona watershed for Americanmonetarypolicy?'in TheDefiningMoment:TheGreatDepressionand theAmericanEconomyin the TwentiethCentury,ed. MichaelBordo,ClaudiaGoldin, and EugeneWhite.Chicago:Universityof ChicagoPress Cassel,Gustav(1936) TheDownfallof the GoldStandard.Oxford:ClarendonPress Cecchetti,StephenG., and StefanKrause(2001) 'Financialstructure,macroeconomic stabilityand monetarypolicy,'NBER WorkingPaperNo. 8354 (July) Clarke, S.V.O. (1967) Central Bank Cooperation,1924-1931. New York: Federal ReserveBank of New York 26 B. Eichengreen Cole, Harold L., Lee E. Ohanian, and Ron Leung (2002) 'Deflation, real wages, and the international Great Depression: a productivity puzzle,' unpublished manuscript, Federal Reserve Bank of Minneapolis Cooper, Russell, and Joao Ejarque(1995) 'Financial intermediationand the Great Depression: a multiple equilibrium interpretation,' NBER Working Paper No. 5130 (May) Della Paolera, Gerardo, and Alan Taylor (1998) 'Economic recovery from the Argentine great Depression: Institutions, Expectations and the Change of Macroeconomic Regime,' NBER Working Paper No. 6767 (October) DeLong, J. Bradford (1990) 'Liquidation cycles: old fashioned real business cycle theory and the Great Depression,' NBER Working Paper No. 3546 (December) Eichengreen, Barry (1992) Golden Fetters: The Gold Standard and the Great Depression 1919-1939. New York: Oxford University Press Eichengreen, Barry, and Kris Mitchener (2003) 'The Great Depression as a credit boom gone wrong,' unpublished manuscript, University of California, Berkeley, and Santa Clara University Eichengreen, Barry, and Jeffrey Sachs (1985) 'Exchange rates and economic recovery in the 1930s,' Journal of Economic History 45, 925-46 Einzig, Paul (1937) The Theory of Forward Exchange. London: Macmillan Ferguson, Thomas, and Peter Temin (2001) 'Made in Germany: the German currency crisis of 1931,' unpublished manuscript, MIT Field, Alexander (1992) 'Uncontrolled Land Development and the Duration of the Great Depression in the United States,' Journal of Economic History 52, 785-805 Flandreau, Marc (1997) 'Central bank cooperation in historical perspective: a skeptical view,' Economic History Review 50, 735-63 Flandreau, Marc, Jacques Le Cacheaux, and Frederic Zumer (1998) 'Stability without a pact?Lessons from the Europeangold standard, 1880-1914,' EconomicPolicy 26, 115-62 Friedman, Milton, and Anna J. Schwartz (1963) A Monetary History of the United States, 1860-1963. Princeton, NJ: Princeton University Press Galbraith, John Kenneth (1972) The Great Crash. Boston: Houghton-Mifflin Grossman, Richard (1994) 'The shoe that didn't drop: explaining banking stability during the Great Depression,' Journal of Economic History 54, 654-82 Hayek, Freidrich A. von (1925) 'Die Wahrungspolitik der vereingten Staaten seit der Uberwindung der Krise um 1920,' Zeitschrift fur Volkswirtschaftund Sozialpolitik N.S. 5, 25-65, 254-317 Hsieh, Chang-Tai, and Christina Romer (2001) 'Was the Federal Reserve fettered? Devaluation expectations in the 1932 monetary expansion,' NBER Working Paper No.8113 (February) James, Harold (2001) The End of Globalization: Lessons from the Great Depression, Cambridge: Harvard University Press Johnson, H. Clark (1998) Gold, France and the Great Depression, 1919-1932. New Haven, CT: Yale University Press Jonung, Lars (1981) 'The Depression in Sweden and the United States: a comparison of causes and policies,' in The Great Depression Revisited. ed. Karl Brunner Boston: Martinus Nijhoff Kaminsky, Graciela, and Carmen Reinhart (1998) 'The Twin Crises: The Causes of Banking and Balance of Payments Problems,' American Economic Review 8, 473-501 Kindleberger, Charles P. (1973) The World in Depression 1929-1939. Berkeley: University of California Press Krugman, Paul (1979) 'A model of balance of payments crises,' Journal of Money, Credit and Banking 11, 311-25 McCallum, Bennett T. (2001) 'Inflation targeting and the liquidity trap,' NBER Working Paper No. 8225 (April) Understanding the Great Depression 27 Meltzer, Allan H. (2002) 'A History of the Federal Reserve, Volume 1, 1913-1951', unpublished manuscript, Carnegie Mellon University Mises, Ludwig von (1924) Theorie des Geldes und der Umlaufsmittel. 2nd ed. Munich. Trans. and reprinted as The Theory of Money and Credit. New Haven, CT: Yale University Press (1953) Moure, Kenneth (2002) The Gold Standard Illusion: France, the Bank of France, and the International Gold Standard, 1914-1939. New York: Oxford University Press Mundell, Robert A. (2000) 'A reconsideration of the twentieth century,' American Economic Review 90, 327-40 Obstfeld, Maurice, and Alan Taylor (2002) 'Globalization and capital markets,' NBER Working Paper No. 8846 (March) Ritschl, Albrecht (1997) Deutschlands Krise und Konjunktur, 1924-1934. Habilitationsschrift, Economics Department, University of Munich - (1999) 'Peter Temin and the onset of the Great Depression in Germany: a reappraisal,' unpublished manuscript, Pompeu Fabra University Ritschl, Albrecht, and Ulrich Woitek (2000) 'Did monetary forces cause the Great Depression? A Bayesian VAR analysis for the U.S. economy,' Working Paper No. 50, Institute for Empirical Research in Economics, University of Zurich (July) Robbins, Lionel (1935) The Great Depression (New York: Macmillan) Romer, Christina (1992) 'What ended the Great Depression?' Journal of Economic History 52, 757-84 Rothbard, Murray N. (1975) America's Greatest Depression 3rd ed. Kansas City: Sheed and Ward Sprague, Otto M.W. (1910) History of Crises Under the National Banking System. Washington, D.C.: Government Printing Office Sumner, Scott (1997) 'News, financial markets, and the collapse of the gold standard: 1931-1932,' Research in Economic History 17, 39-84 Svennson, Lars E.O. (2000) 'The zero bound in an open economy: a foolproof way of escaping from a liquidity trap,' NBER Working Paper No. 7957 (October) Szostak, Rick (1995) Technological Innovation and the Great Depression. Boulder, CO: Westview Press Temin, Peter (1971) 'The beginning of the Great Depression in Germany,' Economic History Review 24, 240-8 -(1976) Did Monetary Forces Cause the Great Depression?New York: Norton Temin, Peter, and Barrie Wigmore (1984) 'The end of one big deflation,' Explorations in Economic History 27, 483-502 Thomas, Mark (1983) 'Rearmament and economic recovery in the late 1930s,' Economic History Review 36, 552-79 Tobin, James (1965) 'The monetary interpretation of history,' American Economic Review 55, 464-85 United States Congress, House Committee on the Currency (1935) Hearings on H.R.5357, Banking Act of 1935. 74th Congress, 1st Session. Washington: GPO Vila, Anne (2001) 'Asset price crises and banking crises: some empirical evidence,' www.bis.org/pubs/confer08.pdf Voth, Hans-Joachim (2002) 'With a bang, not a whimper: pricking Germany's "stock market bubble" in 1927 and the slide into depression,' unpublished manuscript, Pompeu Fabra University Wilcox, James (1984) 'Excess reserves in the Great Depression,' NBER Working Paper No. 1374 (June) Wigmore, Barrie (1984) 'Was the bank holiday of 1933 caused by a run on the dollar?' Journal of Economic History 47, 738-55