real business cycle models of the great depressionperso.uclouvain.be/luca.pensieroso/tesi...

Universite catholique de LouvainDepartement des Sciences Economiques

These presentee en vue de lobtention du grade deDocteur en Sciences Economiques

Real Business Cycle Models of the GreatDepression

Luca Pensieroso

Juin 2007

Composition du jury:

Michel De Vroey (Universite catholique de Louvain) (directeur)David de la Croix (Universite cahtolique de Louvain)Franck Portier (Universite des Sciences Sociales Toulouse 1)Henri Sneessens (Universite catholique de Louvain)Raf Wouters (Banque Nationale de Belgique)

i

Ai miei genitori.

Acknowledgements

Eventually, five years after I first arrived in Louvain-la-Neuve, I see theend of my graduate studies. With them, a period of my life comes to theend, while a new one, whose story is still to be written, is already on thehorizon.

Before drawing the final line, it is time to thank all the people that madethis achievement possible.

My first thought goes to my supervisor, Michel De Vroey. He believedin me since I was a Master student. Tough but fair, he taught me to ques-tion my certainties, often to discover that they were more based on ideo-logical a-priori than on purely rational basis. He has been the best super-visor I could ever wish.

I would like to thank the other members of the Jury, David de la Croix,Franck Portier, Henri Sneessens and Raf Wouters for their interesting re-marks on this thesis. Their suggestions resulted in appreciable improve-ments of this work.

The Department of Economics of the Catholic University of Louvain isan ideal working place, where the rigour of scientific research meets withthe pleasure of an informal and lively ambience. I would like to thank allthe professors, the staff and the students for their precious daily coopera-tion.

A special thank goes to Raouf Boucekkine and David de la Croix. Theywere ready to fund a young Italian economist with a heterodox back-ground, showing a laudable open-mindedness, and trusting his capabil-ities as researcher. I hope to have been able to pay them back with thevaluable research they expected from me.

Part of this work was carried out at the GREMAQ - University of Toulouse

iv

1. I would like to thank Franck Portier for inviting me, and Aude Schloesingfor the excellent organization of my visiting. Matteo Bassi, Sara Biancini,Wolfgang Heisen and Dorte Klaus, among others, made my staying inToulouse enjoyable.

The story of a PhD thesis is also a story of days (and nights) spent onapparently unsurmountable problems. If I have not got depressed, it alsobecause I could count on the support of many colleagues, or friends, as Ishould say. My memory goes to the days of the Master, at the Saturday-night dinners organized by Augusta Badriotti; to the countless (and end-less) discussions of economic theory, methodology and policy with GabrieleCardullo, Mario Denni, Giulio Nicoletti and Alessandro Sommacal; to theunforgettable olehh! said by Cecilia Vergari; to the nights spent onGame Theory with Alessandro, secretely waiting for the phone calls ofour respective girlfriends to give us a pause; to the 19-hours non-stop im-mersion in Business Cycle Theory with Mario before the Macro II exam; tothe Mundialito football-tournament, where the Italian Master studentsshowed their taste for epics with their Belli Masteronzi team, lead bytheir captain Oscar Amerighi . . . Other faces, other places, other namescross my mind, all equally important, just too many to mention themall: Anna Batyra, Mauro Bambi, Carmen Camacho, Gianfranco Casanova,Gul Ertan, Filomena Garcia, Antonio Minniti, Giordano Mion, Andrea Sil-vestrini, Marie Vander Donckt, Benteng Zou. . . Thank to all of you, guys,from the deepest of my heart.

When you leave your country and go abroad for several years, youwould like the hypothesis of ceteris paribus to hold at home. This is notalways the case. But it was so for my best friends ever: David, Francescaand Antonello. Grazie, ragazzi.

Dulcis in fundo, the final lines are for my family. To my sister Barbaragoes a big thank for her joie de vivre, for all those chats that had the in-credible property of making me feel better, no matter how hard were thetimes. My fiancee Alessandra has been the love, joy and support of mylife for all these long years. To her goes my love and my deepest gratitude.Finally, I have always thought that I own to my parents much of what I amtoday. This work is dedicated to them, to the memory of my father, and tothe love and care of my mother.

Preface

When I started this project, my orientation towards this new trend in theliterature was admittedly rather critical. I had the feeling that RBC authorshad gone too far. One thing is to apply dynamic optimization and theequilibrium method to the mild post-war business cycles. A completelydifferent thing ought to be the analysis of the Great Depression, an eventcharacterized by pervasive political and financial turmoils, with evidentdisruptions in the exchange system. Isnt that to commit a hybris sin, Iasked myself?

If this was my mood at the start of this work, I have been graduallyrealizing that things were more complex than that. Modelization had itsown advantages, often allowing to assess the quantitative contribution ofexistent competing explanations. The crude and aseptic data analysis pro-vided for by RBC authors shed new lights on the event, reminding us thateven statements that are apparently obvious need to be proved. Indeed,in the process of proving them, it often happens to realize that those state-ment were not so obvious, after all. In one word, the application of theRBC methodology to the analysis of the Great Depression put a renewedaccent on the discipline-aspect of the theory, which consists in forcing theeconomist to take into account the general equilibrium effects of any sup-posed explanation of the Great Depression.

Moreover, the fact of building models geared towards quantitative as-sessments makes results open to direct contest, which is not immediatelythe case for narrative historical accounts.

On the other hand, I keep on thinking that too narrow a focus on themodeling aspect is often detrimental to the general understanding of thephenomenon. Models are built on exclusions, as economists focus their

vi

analyses on those aspects deemed to be crucial. Other aspects, which maybe important as well, are often overlooked, not least for the sake of analyt-ical tractability. Here is where history matters. Having the broad picturein mind may indeed help to be cautious in deriving implications from eco-nomic models.

To conclude, using a metaphor from Alan Blinder,1 I still think that theexplanation of complex economic facts like the Great Depression of the1930s is as much an art as it is a science. What I have learned in thoseyears spent working on the Great Depression is that while practising thedark art, science is rather useful.

1See Blinder (1997), p. 17

Contents

1 Introduction 1

2 A Critical Survey 52.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52.2 Assumptions and methodology . . . . . . . . . . . . . . . . . 7

2.2.1 Assumptions . . . . . . . . . . . . . . . . . . . . . . . 72.2.2 Dating the Depression . . . . . . . . . . . . . . . . . . 82.2.3 Methodology . . . . . . . . . . . . . . . . . . . . . . . 92.2.4 The normality view: history and economics . . . . 102.2.5 The national dimension of the phenomenon . . . . . 11

2.3 The US Great Depression . . . . . . . . . . . . . . . . . . . . . 122.3.1 Cole, Ohanian and Prescott on the Great Depression 142.3.2 Other RBC models of the Great Depression . . . . . . 232.3.3 A sunspots neoclassical interpretation of the US Great

Depression . . . . . . . . . . . . . . . . . . . . . . . . . 302.4 Great Depressions worldwide . . . . . . . . . . . . . . . . . . 32

2.4.1 The critique of the consensus view . . . . . . . . . 322.4.2 Case studies . . . . . . . . . . . . . . . . . . . . . . . . 34

2.5 A critical perspective . . . . . . . . . . . . . . . . . . . . . . . 402.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

3 The Abandonment of the Abstentionist Viewpoint 493.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493.2 Lucas on the Great Depression . . . . . . . . . . . . . . . . . 50

3.2.1 The Lucas and Rapping (1969) paper . . . . . . . . . . 503.2.2 Reess criticism . . . . . . . . . . . . . . . . . . . . . . 51

viii CONTENTS

3.2.3 Lucass reaction . . . . . . . . . . . . . . . . . . . . . . 523.2.4 Lucass subsequent standpoint . . . . . . . . . . . . . 53

3.3 Prescott on the Great Depression . . . . . . . . . . . . . . . . 553.3.1 Prescotts early view . . . . . . . . . . . . . . . . . . . 553.3.2 Prescotts present standpoint . . . . . . . . . . . . . . 56

3.4 Explaining Prescotts change of mind . . . . . . . . . . . . . . 593.5 Cole and Ohanian on the US Great Depression . . . . . . . . 603.6 An assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

3.6.1 The originality of Cole and Ohanians analysis . . . . 653.6.2 One or several great depressions? . . . . . . . . . . . 683.6.3 Contrasting the approaches of equilibrium macroe-

conomics and economic history to the Great Depres-sion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

3.7 Lucas after Cole and Ohanian . . . . . . . . . . . . . . . . . . 743.8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

4 The Belgian Great Depression 774.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 774.2 A look at the data . . . . . . . . . . . . . . . . . . . . . . . . . 794.3 History of interwar Belgium . . . . . . . . . . . . . . . . . . . 85

4.3.1 1920-1926 . . . . . . . . . . . . . . . . . . . . . . . . . 854.3.2 1926-1930 . . . . . . . . . . . . . . . . . . . . . . . . . 874.3.3 1930-1935 . . . . . . . . . . . . . . . . . . . . . . . . . 884.3.4 1935-1939 . . . . . . . . . . . . . . . . . . . . . . . . . 89

4.4 The Data through the lens of the theory . . . . . . . . . . . . 894.4.1 Detrending . . . . . . . . . . . . . . . . . . . . . . . . 894.4.2 Total factor productivity . . . . . . . . . . . . . . . . . 91

4.5 A baseline RBC model of the Belgian Great Depression . . . 924.5.1 The model . . . . . . . . . . . . . . . . . . . . . . . . . 924.5.2 Calibration . . . . . . . . . . . . . . . . . . . . . . . . . 944.5.3 Simulation . . . . . . . . . . . . . . . . . . . . . . . . . 944.5.4 Comments on the results . . . . . . . . . . . . . . . . 97

4.6 War expectations and investments . . . . . . . . . . . . . . . 984.7 Money and sticky wages . . . . . . . . . . . . . . . . . . . . . 102

CONTENTS ix

4.7.1 The model . . . . . . . . . . . . . . . . . . . . . . . . . 1054.7.2 Calibration . . . . . . . . . . . . . . . . . . . . . . . . . 1084.7.3 Simulation . . . . . . . . . . . . . . . . . . . . . . . . . 1114.7.4 Comments on the results . . . . . . . . . . . . . . . . 1124.7.5 Asymmetric wage rigidity . . . . . . . . . . . . . . . . 1164.7.6 TFP shocks and sticky wages . . . . . . . . . . . . . . 126

4.8 A small-open-economy RBC model . . . . . . . . . . . . . . . 1264.8.1 The model . . . . . . . . . . . . . . . . . . . . . . . . . 1274.8.2 Calibration and simulation . . . . . . . . . . . . . . . 1324.8.3 Extension: shocks on r . . . . . . . . . . . . . . . . . 136

4.9 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1374.10 Appendix: Data . . . . . . . . . . . . . . . . . . . . . . . . . . 139

5 Prospects for Future Research 1455.1 The Belgian case: an open-economy perspective . . . . . . . 1455.2 Involuntary unemployment and the Great Depression . . . . 1465.3 Expectations and the recovery from the US Great Depression 148

Bibliography 153

List of Figures 165

List of Tables 169

x CONTENTS

Chapter 1

Introduction

To understand the Great Depression is the Holy Grail of macroe-conomics (Bernanke (1995), p. 1).

So far, for all the bravery of the knights, the quest has proved to bechallenging, and the task is not yet fully attained. In the span of almosteighty years, several theories claimed to have clinched the matter. Inex-orably, each time that claim proved to be wrong, or partial at least. Eachtime new theoretical developments, or new data successfully challengedthe dominant story. What is more, each time the change in the perspectiveunder which the economists looked at the Depression signaled a shift inthe dominant paradigm in macroeconomics.

This thesis deals with a new stage of the quest. Few years after Eichen-green (1992)s synthesis between the Keynesian and monetarist explana-tions was believed to have provided the final word on the Depression his-tory, Harold Cole and Lee Ohanian brought it all up for discussion. Indaring to face the Great Depression by means of a full-fledged neoclas-sical model in the real business cycle (RBC) tradition, they broke whatwas perceived by many authors as a taboo (Prescott (2002)): for the firsttime in the history of macroeconomics, an analytical neoclassical modelentered the battlefield of the Great Depression, once considered a themedefying any equilibrium explanation. What is new about the RBC expla-nation of the Great Depression? Are we witnessing the setting of a newdominant paradigm in macroeconomics? How did the paradigm need to

2 Introduction

be emended in the endeavour to tackle the Depression issue?This thesis elucidates the fundamentals of RBC modeling on the Great

Depression; investigates its origin back in the history of the new classicalrevolution; provides some critical thought on its whereabouts; presentsthe analysis of the Belgian case through the lens of RBC theory.

As it is natural for the questions at stake, this work is on the border linebetween history of theory and economics.

The first two chapters are more on the history side. Chapter 2 criti-cally reviews the RBC interpretation of the Great Depression, clarifyingits theoretical and methodological foundations, and paving the way foran assessment of its contribution. The focus of the chapter is on the out-standing methodological innovation brought about by the attempt to castthe Great Depression within an equilibrium framework. While suchan attempt deserves to be hailed, many problematic issues are put to thefore. What does total factor productivity (TFP) means in historic accounts?How reliable are linearized models to grasp output decreases as importantas those observed during the 1930s? Could economic theory do without acausative perspective, and be content with the ability to mimic a given setof data, in tackling historical issues like the Great Depression? These andother questions are singled out in the chapter.1

Chapter 3, which is coauthored with Michel De Vroey, documents andassesses the reversal of opinion by new classical macroeconomists aboutthe ability of their methodology to explain the Great Depression. We showthat in the 1970s and 1980s Lucas and Prescott took an abstentionist stance.They maintained that, because of its exceptional character, an explanationof the Great Depression was beyond the grasp of the equilibrium approachto the business cycle. However, while Lucas stuck to this view, Prescottchanged his mind at the end of the 1990s, breaking his earlier self-imposedrestraint. We argue that such a reversal of opinion has been made possi-ble by the progressive evolution of RBC theory in the last decades, from astrictly Walrasian framework to a general method of analysis that encom-passes also non-Walrasian features, like imperfect competition and price

1A slightly modified version of this chapter was published in the Journal of EconomicSurveys, 21, 110-142.

3

stickiness.In this chapter we also produce a first assessment of RBC models of

the Great Depression. We claim that the fact of having constructed anequilibrium model of the Great Depression constitutes a methodologicalbreakthrough. However, as far as substance is concerned, we argue thatthe contribution of real business cycle literature on the Great Depressionis slim, and does not gain the upper hand over the work of economic his-torians.2

If the first two chapters are more on the history side, chapter 4 is defi-nitely on the economics side. This chapter casts the Belgian Great Depres-sion of the 1930s within a dynamic stochastic general equilibrium (DSGE)framework. After a brief detour in the data and history of the interwarBelgium, I show that a TFP shock within a standard RBC model givesunsatisfactory results. Given the enormous drop in investments, and theinternational political environment of the time, I investigate whether fearsof the outbreak of a war could play a role in accounting for the Depres-sion, particularly for what concerns its protracted character. The exerciseshows that this is not likely the case, as introducing war expectations inthe baseline model produces little improvement.

Simple growth accounting suggests that the onset of the Great Depres-sion is almost entirely accounted for by a big drop in hours worked. Thisobservation, together with the data on nominal and real wages and theevidence on monetary turbulence put forward by historians, lead me toexplore the performance of a DSGE model with sticky wages a la Taylorand monetary shocks. Results confirms that high real wages due to a sud-den big deflation might have been the transmission mechanism explainingthe onset of the Great Depression in Belgium. However, the model fails toaccount for the absence of recovery after the 1935 devaluation, a failure itshares with all the other model considered in this chapter.

In view of the peculiar small-open-economy characteristic of Belgium,the chapter concludes with a first extension to the open-economy dimen-sion, though with no appreciable gain in terms of explanatory power.

2A slightly modified version of this chapter was published in Contributions to Macroe-conomics, 6, issue 1, article 13.

4 Introduction

Chapter 2

A Critical Survey

2.1 Introduction

The Great Depression of the 1930s was undoubtedly the most importanteconomic crisis ever witnessed in the twentieth century. Its extension andduration convinced several contemporary observers that it might well sig-nal the approaching collapse of the capitalist production system.

The Great Depression plays an outstanding role in the history of ideas.Keyness General Theory, in effect, dates to 1936, and the Great Depressionunquestionably paved the way for Keyness work.

The Keynesian approach to economic theory concentrates on the con-cept of market failure, as opposed to the classical laissez faire theory. Conse-quently, in the eyes of most contemporary observers, the experience of theGreat Depression seemed to confirm the correctness of Keyness intuition,that, in the short run at least, a capitalist economy does not gravitate to-wards full employment. The Keynesian approach to economics remainedthe mainstream theory until the end of the 1960s, when it was first chal-lenged by Friedman and the monetarists, and subsequently replaced bynew classical macroeconomics. The new theory is an equilibrium busi-ness cycle theory, meaning that the analysis is cast in a Walrasian frame-work, and grounded on individual rational choices. The main message ofthis new trend in economic literature, with respect to the history of eco-nomic thought, is that there is no need to resort to any market failure idea

6 A Critical Survey

in order to provide a thorough explanation of the business cycle. A prop-erly defined neoclassical model can provide a plausible explanation of thephenomenon.

Nevertheless, even after the Keynesian model had lost its predomi-nance and been replaced by new classical macroeconomics, the Great De-pression still appeared to be an example of market failure, whose causeswere mainly attributed to the complex social and institutional situation af-ter World War I (Kindleberger (1973); Eichengreen (1992)), and whose endcould be ascribed to the intervention of public authorities (Romer (1992);Vernon (1994). New classical macroeconomists themselves considered theGreat Depression a phenomenon somehow beyond the reach of equilib-rium theory. In particular, Lucas, whose distinctive contribution to eco-nomic theory consists of having stated that all cycles were alike and couldbe studied as equilibrium phenomena (Lucas (1977)), wrote:

The Great Depression. . . remains a formidable barrier to a com-pletely unbending application of the view that business cyclesare all alike. (Lucas, 1980, p. 273)

If the Depression continues, in some respects, to defy explana-tion by existing economic analysis (as I believe it does), perhapsit is gradually succumbing under the Law of Large Numbers.(Lucas (1980) p. 284)

However, at the end of the 1990s, attempts to overcome this limitationsaw the light of day: a new interpretation of the Great Depression, whichtried to explain it within a real business cycle (RBC)1 framework, began to

1Throughout this article, I will use the epithets RBC and New Classical as synony-mous. The focus here is on the methodological aspect of tackling the Great Depressionwithin a neoclassical equilibrium framework. In view of this methodological focus, thedistinction between real or monetary shocks is less important than the general methodof analysis. This is also the reason why I do not make any a priori distinction betweenRBC and Dynamic Stochastic General Equilibrium (DSGE) models (DSGE being the mostrecent label adopted in the literature to denote such models). I will, however, refer to asimilar distinction when discussing a specific methodological aspect, namely whether themodels considered are or are not involved in causal perspectives. Preference is given tothe RBC nomenclature.

2.2 Assumptions and methodology 7

gain ground. Instead of viewing the Great Depression as a phenomenonlying beyond the grasp of the equilibrium discipline, authors working inthis direction believe that the new classical methodology and theory mightbe able to tackle it.

The aim of this paper is to present a critical review of this RBC in-terpretation of the Great Depression, by singling out its theoretical andmethodological foundations. The paper will be organized as follows. InSection 2.2, I will explain some methodological premises for the applica-tion of RBC theory to the Great Depression. In Sections 2.3 and 2.4, areview of existing RBC papers about the US and international Great De-pressions will be presented. Section 2.5 provides some critical remarks,and indicates guidelines for future research, while Section 2.6 summarizesthe argument.

2.2 Assumptions and methodology

2.2.1 Assumptions

The distinctive feature of RBC theory is its attempt to explain cyclical fluc-tuations of income and employment by two fundamental hypotheses: theequilibrium hypothesis and the exogenous shock hypothesis.

The equilibrium hypothesis is the postulate that an economic cyclecan be studied as an equilibrium phenomenon, or, in other words, that itcan be studied in a framework postulating market clearing and agents op-timizing behaviour (Lucas (1977)). Under this assumption, business cyclesare the aggregate result of the optimum response of individuals to changesin the economic environment (Hartley, Hoover, and Salyer (1997)).

I will label as exogenous shock hypothesis the assumption that thesource of any economic cycle is exogenous to the growth process. In anRBC perspective, the economic cycle is conceived as a stochastic oscilla-tion around a trend. Such a trend is determined by savings, demographyand technology, as in Solow (1956).2 This hypothesis characterizes the con-

2For surveys of the RBC literature see, inter alia, Hartley, Hoover, and Salyer (1997),Mankiw (1989), Plosser (1989), Ryan (2002) and Stadler (1994).

8 A Critical Survey

ception of economic cycles within the RBC framework as being due to ex-ogenous shocks to the fundamentals of an economic system, as opposed totheories in which fluctuations are endogenous or to animal-spirit driventheories, in which fluctuations result from the indeterminacy of the long-run growth path.

This conception of economic cycles has important implications for thedefinition of depressions. Researchers in the RBC tradition define a de-pression as a period in which the rate of growth of the economy is sud-denly and significantly below that which it would have been if the ex-ogenous random shock that hit the economy had never occurred. As tothe notion of a Great Depression, Kehoe and Prescott (2002) consider, as aworking definition, that a recession is a Great Depression if output fallscumulatively by more than 20% with respect to its trend level, droppingby more than 15% in the first decade of the depression. These numbersserve to give a quantitative definition of the borderline between a busi-ness cycle, and a business cycle that has become a Great Depression. Ofcourse they contain a good dose of arbitrariness, and although they maybe reasonable, no theoretical meaning should be attributed to them.3

2.2.2 Dating the Depression

It might be thought that the dating of the Depression would be an issueon which consensus existed, but this is not the case. As a matter of factRBC theorists have changed the general way of thinking on this issue.Traditionally, economists tended to consider the Great Depression as start-ing with the stock market crash of 1929, and ending with the election ofRoosevelt in 1933 (Robbins (1934); Friedman and Schwartz (1963); Temin

3Moreover, these definitions produce some odd results. Kehoe and Prescott (2002)argue that Switzerland has been experiencing a Great Depression since 1973, on thegrounds that detrended output per person of working age fell by more than 30% be-tween 1973 and 2000, with a decline of more than 18% between 1973 and 1983. Anyonecan witness, however, that life in Switzerland in the last 30 years has had very little incommon with life in the USA during the 1930s!


(1989); Eichengreen (1992)).4 However in an RBC interpretation, the GreatDepression is defined as covering the entire decade of the 1930s. This re-sults from the definition of a Great Depression given above: US de-trendedoutput dropped more than 35% in 4 years, while in 1939 it was nearly 27%below its 1929 de-trended level (Cole and Ohanian (1999)). As Prescott(1999) points out, this change in the timing of the event shifts the natureof the central question to be addressed from Why was there such a bigdecline in output and employment between 1929 and 1933? to Why didthe economy remain so depressed for the entire decade? In other words,according to RBC theoreticians, a new issue should be added to the tra-ditional question of what caused the Great Depression, namely, Whatexplains the slowness of the recovery phase?

2.2.3 Methodology

As to methodology, RBC theorists tread in Lucass footsteps by arguingthat the central purpose of a theory of the economic cycle is to make the ar-tificial, modelled economy reproduce the actual behaviour of a real-worldeconomy (Lucas (1980)). Elucidating the origin of a particular cycle is per-ceived as secondary. The logic of this methodological premise must betraced back to the fundamental hypotheses we have singled out. If anyeconomic cycle starts with an exogenous shock, studying the specific char-acteristic of this shock serves little purpose for the task of elaborating ageneral theory of the business cycle. It is much more important to under-stand the regularities that will ensue after the shock occurs.

RBC theoreticians build models in the SolowRamsey tradition, mod-ified to allow for stochastic shocks that hit the economy at random. Any

4Here I refer to the dating of the event called the Great Depression, not to the dat-ing of its alleged causes. In effect, many of the authors quoted in the text consider thecauses of the Great Depression to be rooted in events that occurred well before 1929.Eichengreen (1992) and Friedman and Schwartz (1963) are examples. An exception tothis general tendency to date the Great Depression between 1929 and 1933 is Galbraith(1995), who criticises this idea from a post-Keynesian point of view, asserting that theGreat Depression never ended, but was swept away by the outbreak of the Second WorldWar.

10 A Critical Survey

stochastic shock of this nature is called an impulse mechanism of thebusiness cycle. The typical impulse mechanism considered in standardRBC models is a technological shock, represented as an autoregressivestochastic shock on the total factor productivity (TFP). TFP is a parameterof the production function, which embodies a broad concept of efficiencyin combining inputs to obtain output.5

Having defined the impulse mechanism of the business cycle, RBC the-oreticians compute the equilibrium reaction to the impulse mechanism.That is, they study the qualitative and quantitative response of the modeleconomy to the random shock, on the basis of the set of relationships pos-tulated by the model that allows them to identify a propagation mecha-nism for the shock. Such a propagation mechanism is standard in all themodels reviewed here, and is typically based on both the intertemporalsubstitution in leisure and consumption, and the intratemporal substitu-tion between labour and leisure.

This simulation technique requires the model to be calibrated, that is, anumerical value must be assigned to each parameter on the basis of econo-metric estimates, or, if reliable econometric data are absent, on the basis ofeconomic plausibility. If the perturbed model economy reproduces ag-gregate fluctuations reasonably well, it can be considered as a plausibletheory of the cycle. That is, the ability of an artificial model to reproducea set of stylized facts after being hit by an exogenous random shock isthe methodological litmus test by which the robustness of the theory isjudged.

2.2.4 The normality view: history and economics

RBCs methodological premises have important implications for the anal-ysis of historical events such as the Great Depression. In the RBC ap-proach, economic theory and economic history are perceived as pertainingto different, though possibly complementary, realms. Economic theory, asstated above, is concerned with regular co-movements in the behaviourof economic variables. Economic history, on the contrary, is a different

5Solow (1957). See Hulten (2000) for a review of TFP.


branch of social science, naturally inclined towards considering singulari-ties. In this way, a specific event is amenable to economic theory if, duringthat event, economic variables co-moved as predicted by the theory. If,however, the event is peculiar, it should simply be left to historians, andconsidered beyond the grasp of economic theory.

Once this consequence of their methodological assumptions is speltout, the breaking of the earlier limit to equilibrium theory assumes newconnotations. To all intents and purposes, it amounts to considering theGreat Depression as a business cycle, possibly of greater than usual mag-nitude; the alternative view is that the Great Depression was a singularity.That is, in assuming that the Great Depression is amenable to RBC the-ory and method, these authors implicitly assume that during the GreatDepression economic aggregates behaved as in any other business cycle,although with greater variance in their oscillation. In this paper, I will referto this implicit assumption as the normality view.

2.2.5 The national dimension of the phenomenon

The RBC interpretation of the Great Depression differs from previous in-terpretations in the role assigned to the international political and eco-nomic environment during the 1930s. While earlier leading authors (Kindle-berger (1973); Eichengreen (1992); Bernanke (1995)) stressed the interna-tional dimension of the Great Depression, and went so far as to say that afull understanding of that phenomenon could not be reached without con-sidering the international dimension, RBC researchers reversed this posi-tion by concentrating their analysis on isolated country studies. Severalreasons for this change of perspective may be given:

The first work on the Great Depression from an RBC perspective isthe paper by Cole and Ohanian (1999), which is strictly concernedwith the Great Depression in the USA. Data prove that the GreatDepression hit harder in the USA than in other industrialized coun-tries; output fell relatively more, and the state of depression of theeconomy lasted longer than in any other country. This evidence per-suaded the authors to assume that the shock that affected the US


economy must have been far bigger than the shocks that affectedother economies and, in addition, that the slowness of the US recov-ery was probably due to some idiosyncratic shock, since other coun-tries recovered earlier. Moreover, the USA is notoriously an almostclosed economy as far as international trade is concerned. Conse-quently, a national dimension appeared to them sufficient to analysethe US Great Depression.6

From a methodological point of view, the mathematical formaliza-tion that is typical of RBC research forces the economist to leave outmany aspects of reality in order to concentrate on the aspects thatare considered essential. Given that RBC models explain recessionsby means of a shift in the labour-demand schedule (Mankiw (1989)),exogenous shocks to TFP (i.e. exogenous variations in the Solowresidual) are an easy way to reproduce such a shift, while keepingthe model sufficiently compact. This implies that the internationaldimension need not be the main focus of the analysis.

2.3 The US Great Depression

RBC models of the US Great Depression can be split into two classes. Thefirst includes Cole and Ohanian (1999), Cole and Ohanian (2000), Cole andOhanian (2004) and Prescott (1999). In this interpretation, the explanationof the plunge of the early thirties (that is, the historical identification ofthe shock that caused the Great Depression) is considered methodologi-cally less interesting than the explanation of its long duration (that is, whythe Great Depression did not behave in the same way as business cyclesobserved in the post-war period). The causes of the productive collapseof the USA economy in the 1930s are mostly traced back to some exoge-nous supply shock, embodied for simplicity in a parameter of the pro-

6This is also the position held by Romer (1993). Although working from a differentbasis, she argues that the Great Depression in the US was due to a mixture of bad mone-tary policy and aggregate demand shocks, both with idiosyncratic characteristics specificto the American case.

2.3 The US Great Depression 13

duction function (TFP). As for the protracted character of the depression,these models charge New Deal policies with having been responsible forit. These policies produced substantial distortions in the economy, thusimpeding the otherwise inevitable recovery.

The second class of models includes all the other RBC papers on theGreat Depression. These models are more interested in the traditionalquestion of what caused the Great Depression. To this end, they resort toa variety of variables (monetary shocks and sticky wages in Bordo, Erceg,and Evans (2000); preference for liquidity in Christiano, Motto, and Ros-tagno (2004); exogenous demand shocks in Weder (2006)). As to the longduration of the Great Depression, all these authors accept the implicationsof the normality view: either the Great Depression would have been anormal business cycle of greater magnitude had distorting State interven-tions been absent (Bordo, Erceg, and Evans (2000); Christiano, Motto, andRostagno (2004)); or the Great Depression was a normal business cycle ofgreat magnitude that lasted a long time because the shock producing thecycle was extremely long lasting (Weder (2006)).

Other papers exist, the classification of which under the RBC label ismore uncertain. Sunspots models are a case at hand.7 In sunspots mod-els, there is multiplicity of equilibria. Business cycles are defined as theeconomys swinging from one equilibrium to the next. Such swinging

7In the taxonomy adopted here, it is doubtful whether this last class of models shouldbe included in the RBC category. First, expectation shocks are neither exogenous norendogenous, in the sense that their nature is ambiguous. On the one hand, the veryconcept of animal spirits suggests that variations in expectations should be consideredas an exogenous shock. On the other hand, it defies credulity to assume that peopleform expectations without looking at reality, or framing it in some causal perspective,i.e. in a model. In this sense, the shock must be at least partially endogenous. Secondly,be that as it may, by definition expectation shocks are not shocks to the fundamentalsof the economy, which we have assumed to be a distinctive feature of RBC theory. Thecounterargument that leads me to include sunspots models in my taxonomy of RBC isthat they are equilibrium models a la Lucas, they use the Slutzky intuition of businesscycles as summation of random shocks, and they distinguish between the impulse andtransmission mechanisms of the business cycle. Being unable to choose whether or notto include sunspots models in the RBC category, I have decided to include at least one ofthem for the sake of completeness.


movements are determined by self-fulfilling (rational) expectations. TheGreat Depression appears as a demand-driven phenomenon arising frompeoples unexplained, though plausible, pessimistic behaviour. I will re-view one paper in this tradition, that by Harrison and Weder (2002), in aseparate subsection.

2.3.1 Cole, Ohanian and Prescott on the Great Depression

The RBC interpretation of the US Great Depression stems from the work oftwo leading authors, Harold Cole and Lee Ohanian. Initially, they focusedon the standard issue of explaining the origins of the Great Depression.This inquiry led to a rather frustrating result. Neither the standard RBCstory of technological shocks, nor other standard real and monetary fac-tors, could properly account for both the observed magnitude and the longduration of the Great Depression. Cole and Ohanian soon turned their at-tention to the protracted character of the Great Depression, a theme thateventually proved more congenial to RBC methodology and theory. Thedistorting elements of some New Deal policies helped to explain why theeconomy remained depressed for so long. This position has been author-itatively espoused by Prescott, who, in a short comment article in 1999,gave a clear picture of the basic elements of the RBC interpretation of theGreat Depression. As it is representative of the whole of this stream of lit-erature, I will start by discussing Prescotts paper, and turn my attentionto Cole and Ohanians article in a second step.

Prescotts assessment

According to Prescott (1999), the RBC interpretation of the US Great De-pression consists of two building blocks. First, some of the exogenous fac-tors usually described in terms of shocks to TFP caused a strong recessionat the end of 1929. Second, misconceived economic policies, attempting toimprove the disastrous economic performance of that time, impeded thenormal adjustment of market forces. These policies introduced strong dis-torting elements into the US economy: by increasing de jure the real wage


rate, they lowered the normal employment level and the growth path. InPrescotts words:

In the Great Depression, employment was not low because in-vestment was low. Employment and investment were low be-cause labour market institutions and industrial policies changedin a way that lowered normal employment. (Prescott (1999), p.27)

The interest of Prescotts comment is that it highlights the basic ele-ments of the RBC methodology, which I have spelt out in Section 2.2.A dividing line is drawn between the realm of history, which includesthe historical identification of shocks, and the realm of economics, whichstudies the propagation mechanism of the business cycle. In Prescottsmethodological approach, the origin of a shock (i.e. the concrete historicaldetermination of the impulse mechanism of the business cycle) is outsidethe scope of economics. What is more puzzling for an economist is the ex-planation of the slowness of the recovery. By sticking to this view, Prescottreduces the explanation of the Great Depression to the explanation of the19341939 episode. The following quotation illustrates this point.

The fundamental difference between the Great Depression andbusiness cycles is that market hours did not return to normalduring the Great Depression. Rather, market hours fell andstayed low. In the 1930s, labor market institutions and indus-trial policy actions changed normal market hours. I think theseinstitutions and actions are what caused the Great Depression.(Prescott (1999), p. 27)

The point is that this method of analysis might make sense, from atheoretical point of view, when the development of a general theory ofthe business cycle is considered. In that case, the theory can conceivablybe more concerned with the regularities of the business cycle (that is, inhow a business cycle arises from an exogenous shock) than in studyingthe peculiarities of each particular shock. However, things should be dif-ferent when a specific event, such as the Great Depression, is analysed.


In that case, explaining the Great Depression must be tantamount to ex-plaining both its onset and its long duration. By sticking to the normalityview, and trying to cast the Great Depression within the RBC framework,Prescott is led instead to overlook a priori any explanation of the plungeof the early 1930s, a standpoint which is not acceptable from an historicpoint of view. All the more so in that, as shown by Ohanian (2002), inthe specific case of the Great Depression, the exogenous shock requiredto reproduce the data is abnormally large (the next Section illustrates thispoint). It is important to make the point that at least this abnormal dimen-sion deserves more detailed historical analysis.

Cole and Ohanian on the onset of the Great Depression

Cole and Ohanians early work was mainly negative, consisting of show-ing that, when closely scrutinized, earlier explanations of the Great De-pression are unsatisfactory. In their 1999 paper, they started by describingthe behaviour of the main detrended macroeconomic aggregates duringthe decade 19291939; subsequently, they tried to identify, from among themany different explanations in the literature purporting to explain busi-ness cycles, the models that best fit these data. Cole and Ohanian (1999)found that stochastic shocks to the growth rate of the TFP could explainroughly 40% of the 19291932 drop in output. They obtained this result bytaking a suitable specification of the model, and feeding in the observedlevel of TFP as a measure of technological shock.

An interesting point, highlighted by Ohanian (2002), is that the dropin measured TFP during the Great Depression, although not sufficient toreproduce in the model the magnitude of the decline in output, was stillrelatively high when compared with the drops in measured TFP that havenormally accompanied recessions in the post-World War II period. Thisfeature means that the behaviour of the TFP during the 1930s was pecu-liar, for reasons still to be elucidated (see Ohanian (2002) for further dis-cussion).

Alternative real explanations, such as shocks to international trade,public expenditure and distorting taxes, are presumed to have had a lesser


impact, if any, on the crisis. For international trade, Cole and Ohanian(1999) note that the United States was at that time a relatively closed econ-omy. Moreover, the presence of tariffs suggests that US imports mighthave had a high elasticity of substitution with domestic intermediate goods.Consequently, international trade disruptions had no appreciable or en-during negative effects on the US Great Depression. As to public expendi-ture, Cole and Ohanian (1999) report data showing that detrended publicexpenditure in the USA remained above the trend level during almost theentire decade. So a negative crowding-out effect of public expenditure hasto be dismissed. As far as taxes are concerned, Cole and Ohanian (1999)ran two further simulations using data on the average marginal tax rateson factors income: the first with the 1929 average tax level, and the secondwith the 1939 average tax level. In the second simulation the steady-statelevel of labour input was 4% lower than in the first. The authors there-fore concluded that negative fiscal policy shocks did not have appreciableeffects on the 19291933 crisis, but that they can explain some 20% of theweak 19341939 recovery.

Monetary shocks, financial disruptions and nominal rigidities arealso considered to have had little impact on the Great Depression. Coleand Ohanian (2000) reviewed the main mechanisms identified by economiststo explain possible real effects of monetary policy during the 1930s, namelythe Lucas and Rapping (1969) unexpected deflation model, the debt defla-tion model of Fisher (1933), the sticky-wage hypothesis and theories cen-tred on the role of banking disruptions induced by deflation (Bernanke(1983)). By comparing deflation in 19291933 to that in 19201921, the au-thors first excluded Lucas and Rappings and then Fishers hypotheses.8

8Their objection to the first theory is that deflation was more widely anticipated in the1930s than in the 1920s because the nominal interest rate was lower during the 1930s.This weakens Lucas and Rapping (1969)s propagation mechanism, which is based onunexpected deflation. As to Fisher (1933)s debt deflation model, they note that, althoughthe level of private debt as a proportion of output was higher in 1929 than in 1920, outputdropped more sharply during the 1930s than during the 1920s, despite deflation beingless severe. Prices went down by 19.4% in 19201922 and by 11.5% in 19291931, whereasdetrended real income dropped by 3.8% in the 19201922 and by 22.4% in the 19291931period. See Cole and Ohanian (2000), p. 6, Table 3.


To test the sticky-wage hypothesis, Cole and Ohanian (2000) built a twomacrosector general equilibrium model, in which a final good is producedby means of two different types of intermediate goods. Each intermediategood is produced by means of capital and labour. There are two sectorsproducing intermediate goods: one, n, is a competitive sector, with wagesset at the market-clearing level; the other, m, is a noncompetitive sector,where wages are fixed above the market-clearing level. Both sectors usethe same constant returns to scale CobbDouglas technology. The finalgoods sector uses a constant elasticity of substitution (CES) technology.Both capital and labour are assumed to be immobile. The preferences ofthe representative household are specified through a logarithmic utilityfunction. The household can allocate its working time between the twosectors, and it is assumed that wage fixity in the noncompetitive sectoris perceived as a nonrecurring phenomenon (i.e. the model assumes thateach wage shock occurring in any of the Depression years is completelyunexpected).9 The model was calibrated using, as far as possible, standardvalues from the RBC literature for the parameters. A calibration for themodel-specific parameters is also provided. The values of these parame-ters were chosen by considering the manufacturing sector as the empiricalcounterpart of the noncompetitive sector in the model. Running two sim-ulations, one with a benchmark model without nominal wage rigidities,and the other with the model as described above, and comparing theirresults with the data, Cole and Ohanian conclude:

These results suggest that the high wage was not the primarycause of the Great Depression . . . This wage accounts for abouta 3 per cent decline in output at the trough of the Great Depres-sion, compared to an actual 38 per cent decline. Increasing thesize of the distorted sector to 50 per cent, or reducing the sub-stitution elasticity to 0,1 did not significantly change the result.(Cole and Ohanian (2000), p. 20)10

9This is a technical assumption needed in order to be able to compute the equilibriumin the simulation recursively.

10The economic rationale for this result is as follows. In this two-sector model, wage


Cole and Ohanian (2000) also exclude the possibility that wages mightbe significantly underestimated, and argue that in fact the contrary is likelyto be true. Referring to Margo (1993), they assert that wages were proba-bly also well below the trend line in the manufacturing sector, because ofthe compositional bias in favour of high-skilled workers that affected theUS economy in the 1930s.11

As to the analysis of banking shocks, Cole and Ohanian (2000) first de-fined banking shocks as bank closures affecting the information capital.Then they built a model in which information capital was used by banksas input, together with deposits, to obtain a banking output. This bank-ing output appears, in the end, as an input to the production of the finalgood. Both these productive processes are assumed to be constant returnsto scale. This model is built so that, in each sector, the ratio of inputsto outputs is equal for all inputs. Consequently, the loss of informationcapital relative to output due to bank closures is equal to the fraction ofdeposits on output loss due to bank closures. As the US data reported byCole and Ohanian (1999) show this to have been pretty low during the

rigidity has both a direct and an indirect effect on employment. In the distorted sector, m,firms employ labour up to the point where the marginal product of labour equates to thereal wage. Because, by definition, the real wage in this sector is above the market-clearinglevel, production in the distorted sector will be below its potential level. It follows thatpart of the labour force potentially employable in the distorted sector will remain unem-ployed. Such a direct effect is clearly negative. To understand the indirect effect, it isworth considering that output in the distorted sector is an input in the production of thefinal good. Cole and Ohanian (2000) assume that technology is such that Ym and Yn areimperfect complements in the production of the final good, rather than substitutes. Thismeans that, as Ym diminishes, its relative scarcity will increase, and so will its relativedemand. Firms cannot substitute Yn for Ym beyond a certain level. Thus pmpn , the relativeprice of the distorted sector, must increase. According to the authors, this means that,given a monetary wage wm, the real wage wmpm will decrease. In other words, the realwage will decrease in spite of the nominal rigidity, thus producing an upward shift in thevalue of the marginal product of labour (i.e. the marginal product of labour multipliedby the price of output schedule). Thus the indirect effect would tend to counteract thedirect one.

11This point is actually controversial. For instance, Bordo, Erceg, and Evans (2000) ar-gue that data at the industrial level suggest that there was no significant skill compositionbias.


Great Depression, the authors conclude that, because the loss of informa-tion capital was also low during the Great Depression, it only affected theeconomy slightly.

Cole and Ohanian on the long duration of the Great Depression

According to standard RBC theory, the Great Depression should have endedmuch earlier than it actually did. Once the effects of the TFP negativeshock were exhausted, the economy should have returned to its steady-state growth path. In Cole and Ohanian (1999)s simulations, output wouldhave recovered to its trend level by 1936, if the measured shocks to TFP inthe 1930s had been the sole impulse mechanism for the economic cycle.The TFP returned to its trend level that year. However, detrended datashow that in 1939 output was still a good 25% below its trend level. Thisobservation led Cole and Ohanian (1999) to argue that the Great Depres-sion was not only the result of a temporary shock that caused a fluctuationaround the trend-growth path, but was also rather the outcome of a mix-ture of a temporary shock and some permanent shocks that caused thegrowth path itself to shift downwards. At the end of their paper, Cole andOhanian (1999) suggest that a likely culprit could be the New Deal policiesintroduced after 1933.

While this line of research is only alluded to in Cole and Ohanian(1999)s paper, the link between New Deal policies and the Great Depres-sion is the central object of their subsequent research (see Cole and Oha-nian (2004) and an earlier and more detailed working paper version, Coleand Ohanian (2001)). Their basic claim is that New Deal competition andlabour market policies are to blame for the duration of the Great Depres-sion. In particular, they consider two important reforms: the NationalIndustrial Recovery Act (NIRA),12 and the National Labour Relations Act

12The NIRA was enacted in 1933 and declared unconstitutional by the Supreme Courtin 1935. The act aimed to ensure that all sectors were covered by codes of fair com-petition, which would put an end to substantial price deflation and increase workersincome, so promoting greater consumption expenditure. The NIRA also suspended anti-trust laws, and encouraged cooperation between firms, and collusion in price setting; itheavily discouraged price competition, requiring administrative approval for price cuts.


(NLRA).13 These measures had a relatively high coverage in the economy:about 52% of total employment was in sectors covered by the NIRA, whilethis figure reached 77% in the private nonfarm sector (Cole and Ohanian(2001), p. 67, Table 2). Cole and Ohanian (2004) present a model to showthat the rise in prices and wages actually curbed the recovery in produc-tion, rather than boosting it (as Roosevelts economic advisers thoughtit would).14 The model is explicitly oversimplified insofar as it assumesNIRA and NLRA to be the same thing, and does not consider the effectsof other New Deal policies. This is done in order to predict output for thewhole 19341939 period more easily.

The codes, though different for each sector, had to be negotiated under the guidance ofthe National Recovery Administration, and required the approval of the President. Coleand Ohanian (2001) stress that Roosevelts political inclinations, as well as the deep con-viction of his advisers that an increase in prices and nominal wages would be the bestway to counteract the Depression, led him to guarantee approval to those codes that in-cluded collective bargaining over wages, and minimum wages for low-skilled workers.

13The NLRA was enacted in 1935, and its constitutionality was upheld by the SupremeCourt in 1937. It gave workers the right to organize themselves into trade unions inde-pendent of their employers; it prohibited discrimination based on union affiliation, aswell as coercive enrolment in companies unions. The Act also established a NationalLabour Relations Board (NLRB), which had the authority to guarantee the legal enforce-ment of wage agreements.

14It is very interesting to note that the view that the NIRA policy probably had a neg-ative impact is not the prerogative of RBC theory. J.M. Keynes, in an open letter to Roo-sevelt published in The New York Times in 1933, expressed his disagreement with thispolicy as a means of producing a recovery. He argued that the fact that an increase inprices and monetary wages generally characterizes the recovery periods does not meanthat it causes the recovery to happen. So, in Keyness view, the US administration hadconfused causes with effects. In Keyness opinion the NIRA was probably an obstacleto recovery, because it increased the costs of production, whereas the appropriate mea-sure for ending the recession was a policy of large government expenditure, financed bylong-term public debt, together with a monetary policy that fixed low nominal interestrates. Keyness diagnosis was that people were not spending money, and that this wascausing the cumulative deflation that resulted in depression. To restart a virtuous circleof development, people had to be induced to spend. If this were not possible, a goodsurrogate for the missing private expenditure would be government expenditure. In theend, the increase in the aggregate demand would generate an increase in the general levelof prices.


The benchmark specification of the model is a multisector version ofa standard real business cycle model, in which a final good in period tis produced using a variety of intermediate goods. These intermediategoods are produced by different industries, each belonging to a sector. Allthe production technologies exhibit constant returns to scale. Labour is as-sumed to be perfectly mobile across industries and sectors, whereas capi-tal is considered sector-specific. Households are supposed to maximize alogarithmic utility function in which labour is assumed to be indivisible.

To model New Deal policies in this setup, Cole and Ohanian (2004)modified the model in three ways. First, they assumed that, in the econ-omy, a fraction of the sectors producing intermediate goods forms acartel. In these sectors there is, therefore, a rent to be shared betweenworkers and firms arising from the monopolistic extra profits. Second,they assumed that, as a consequence, wages in these cartelized sectors arethe result of bargaining between workers and firms; the relative bargain-ing power of the two parties is embodied in a parameter that gives theprobability of a firm gaining monopolistic extra profits without acceptingworkers wage demands. The cartelized sector behaves in the same way inan insideroutsider model, where all insider workers are paid the samewage. Third, Cole and Ohanian assumed that there are frictions in thelabour market, which allow for flows of workers between the competitiveand the cartelized sectors. Considering that jobs in the cartelized sectorsare better paid, workers prefer to move to these sectors rather than to simi-lar jobs in the competitive sectors. A search process for these jobs thereforeensues.

These three modifications were intended to emphasize the characteris-tic of the New Deal policies that Cole and Ohanian consider essential: aconnection between collective bargaining (allowing de facto for the greaterbargaining power of unions and workers) and price control by cartelizedfirms. They also reproduce the equal pay for equal work principle, acornerstone of union policy in the 1930s. Calibrating and simulating theirmodel, Cole and Ohanian fed in the sequence of observed TFPs as mea-sures of technological shocks, and compared the results of the cartel mod-ification with the competitive benchmark, both relatively and in terms of


reproducing the actual data.Their main result was that cartelization caused a greater drop in output

the greater the bargaining power of workers, i.e. the lower the calibratedvalue for the parameter , and, ceteris paribus, the higher (the share ofthe economy involved in such a policy). However, the effects of varying were much larger than those induced by variation in ; as Cole andOhanian observe:

The key depressing element of the policy is not monopoly perse, but rather the link between wage bargaining and monopoly.(Cole and Ohanian, 2004, p. 805)

As far as a comparison with the actual data is concerned, while thecompetitive model failed to reproduce the observed trend of economic ag-gregates during the recovery, the cartel model made predictions that wereconsiderably closer to the facts. On the basis of the figures obtained, Coleand Ohanian (2004) argue that the cartel model is able to explain a good60% of the slow recovery. The rationale for this result is that the nega-tive effects of higher wages and lower production propagated from thecartelized sectors to the competitive sectors, insofar as the reduced outputin the cartelized sectors tended to lower wages and employment in thecompetitive sectors where, moreover, people were looking for better paidjobs in the cartelized sectors. So, they conclude,

. . . New Deal labor and industrial policies did not lift the econ-omy out of the Great Depression. . . Instead, the joint policiesof increasing labors bargaining power, and linking collusionwith paying high wages, prevented a normal recovery by creat-ing rents and an inefficient insideroutsider friction that raisedwages significantly and restricted employment. (Cole and Oha-nian (2004), p. 813)

2.3.2 Other RBC models of the Great Depression

The debate about sticky wages

Cole and Ohanian (2000)s conclusion that sticky wages were irrelevant


in accounting for the onset of the US Great Depression is far from uncon-troversial. Empirical evidence on cross-sectional international data (pre-sented by Eichengreen and Sachs (1985)) suggests that currency-devaluatingcountries experienced relatively lower real wages and higher industrialproduction, a finding consistent with the sticky-wage hypothesis.15 Onthe other hand, Christiano, Motto, and Rostagno (2004) point out that

. . . There just does not seem to be a tight negative relationshipbetween the real wage on the one hand, and output and em-ployment on the other. (Christiano, Motto, and Rostagno (2004),p. 11)

This point is also debated theoretically among RBC authors. Bordo,Erceg, and Evans (2001) and Gertler (2001) argue that Cole and Ohanian(2000)s result follows from the unjustified assumption of perfect wageflexibility in the non-manufacturing sector. As Gertler (2001) points out,this model excludes nominal wage rigidity by definition, and thus ex-cludes the decrease in the aggregate demand for labour that is necessaryif the sticky-wage hypothesis is to produce real effects. Moreover, Bordo,Erceg, and Evans (2001) emphasize that there is no justification for thischoice, either theoretically or empirically, because it is based on a question-able extension of the wage flexibility observed in the farming sector to thewhole non-manufacturing sector. According to Bordo, Erceg, and Evans(2001), imposing noncompetitive wages in the non-manufacturing sector even lower, perhaps, than the manufacturing sectors wages completelyreverses Cole and Ohanian (2000) results.

In an earlier article, Bordo, Erceg, and Evans (2000) showed that thesticky-wage hypothesis could provide an explanation of the onset of theGreat Depression within an RBC framework. They built a simple one-sector real business cycle model with monetary shocks and fixed wagesa la Taylor (1980). Running a simulation on this model, they found thatit could explain approximately 70% of the 19291932 drop in output, a

15Expansionary monetary policy generates price inflation; provided that nominalwages are rigid, real wages will go down. This will produce an increase in labour de-mand and hence in output.


result in sharp contrast to Cole and Ohanians result. However, Bordo,Erceg, and Evans (2000) admitted that their results clearly show that onits own the sticky-wage hypothesis can account neither for the recoveryphase of the US Great Depression (characterized by a strong monetary ex-pansion (Romer (1992)), nor for the final year of the recession, 19321933.According to them, some financial disruption of the kind envisaged byBernanke (1983) might have been responsible for the crisis in the final year.They suggest a more detailed explanation for the recovery phase, built onCole and Ohanian (1999)s early suggestion about the possible distortingrole of New Deal policies. In particular, they focused (as Cole and Oha-nian did a year later) on the NIRA. Bordo, Erceg, and Evans (2000) thenmodified the process of wage formation in their model by splitting it intotwo processes: a Taylor setting, for the period 1929:31933:2;16 and a levelof wages fixed to their 1933:2 level later on. This modified model showsthat

As long as real wages were legislatively mandated at levelswell above the marginal product of labour that would prevailat full employment, monetary expansion alone could not leadto recovery. (Bordo, Erceg, and Evans (2000), p. 1461)

Christiano, Motto, and Rostagno (2004)

A further development in the application of RBC methodology to the GreatDepression is the recent work by Christiano, Motto, and Rostagno (2004).This paper attempts to build a realistic dynamic stochastic general equi-librium model able to tackle contemporary policy questions. The authorsconsider the US Great Depression as the toughest possible test for such amodel. Christiano, Motto, and Rostagno (2004)s main conclusion is thatwhile the Great Depression was certainly the result of many joint shocks,it is mainly attributable to two factors: a preference for liquidity shock(which induced a shift away from demand deposit towards money, thusin large part causing the onset of the depression); and the increased mar-ket power of workers during the New Deal (which explains why, during

16Quarterly data are used here.


the recovery phase, employment was still so low, thereby shedding somelight on why the recovery phase itself was so slow).

These results are obtained by means of a very complex RBC model. Itsbasic structure is as follows. It is assumed that a final good Yt is producedby a perfectly competitive representative firm, using a number of inter-mediate goods Yj,t. These intermediate goods are produced by monop-olists who set their prices Pj,t subject to Calvo (1983)-style friction. Theintermediate-good firms need labour lj,t and capital Kj,t for their produc-tive activity. They buy working hours from households, paying a wagerate Wt. They rent capital from entrepreneurs, paying a rental price ofcapital Prkt for capital services. Moreover, each intermediate-good firmmust finance in advance fractions k and l of capital and labour services,respectively. They do this by asking for loans from banks, and paying anet interest rate of Rt. Entrepreneurs buy capital x from capital producers,paying for it at the price Qk,t. In order to pay these amounts they use theirnet worth Nt and they borrow Bt = Qk,t Nt from banks, paying a grossinterest rate Zt. At the end of the period, they sell the undepreciated capi-tal back to capital producers, at the same price Qk,t. Entrepreneurs can bebankrupted during each period with a probability 1 t, which also rep-resents the fraction of the new entrepreneurs entering the market duringeach period.

Capital producers produce units of new capital good x by means ofpreviously installed capital x and investment goods It. They buy invest-ment goods from the final-good sector, paying them Pt. Banks use capi-tal and labour to produce their services and hoard reserves. They buyworking time lbt from households, and rent capital Kbt from entrepreneurs,paying Wt and Prkt , respectively. They hold demand deposits D

ft and

Dht from firms and households, respectively, paying them an interest rateof Rat . They also hold time deposits, Tt, from households, which pay anon-state-contingent expected rate of returnRet+1. Finally, households con-sume an amount Ct of the final good, paying Pt per unit; they hold high-powered money M b; they pay lump-sum transfers to entrepreneurs, inorder to guarantee free entry to entrepreneurship and they receive lump-sum transfers corresponding to the net worth of entrepreneurs leaving the


economy.Households are modeled as maximizing a complex utility function en-

compassing, inter alia, habit persistence, shocks to the preference for leisureand shocks to liquidity preference. Households are assumed to be able toexert some monopoly power over labour, so that they set wages within aCalvo contract setting. There is also a non-modeled Government, whichbuys Gt unit of the final good, at the price of Pt per unit.

Next, Christiano, Motto, and Rostagno (2004) introduce eight exoge-nous shocks and study their joint and individual impact on the model,comparing their outcomes with data for the US Great Depression. Theseshocks affect the monopoly power of intermediate-good firms, the monopolypower of wage earners, households preference for currency versus de-mand for deposits, the preference for liquidity, productivity shocks forintermediate goods, the survival probability of the entrepreneur, the rela-tive value of excess reserves in the banking sector and the willingness ofentrepreneurs to take risks. These shocks are drawn from stochastic pro-cesses, and estimated with a maximum likelihood procedure. Christiano,Motto, and Rostagno (2004) assume that the shocks influence the rate ofgrowth of money, because of the monetary authority reaction function.

After having estimated all the parameters and calibrated the model,Christiano, Motto, and Rostagno (2004) ran a simulation, including esti-mated values for the shocks. They found that their model reproduced keyfeatures of the data properly. As anticipated at the beginning of this sec-tion, they also found that two shocks are crucial in explaining the GreatDepression in the United States: preference for liquidity and workersmarket power. While workers market power resembles the traditionalhigh-wages story, which we have discussed above, the preference for liq-uidity deserves some further explanation. An exogenous shock to thepreference for liquidity leads to a decrease in the ratio between demanddeposits and money demand, D

ht

Mt, in consumption and in time deposits.

The aggregate M1 falls, causing the interest rates to increase. The higherinterest rates cause an increase in the debt burden and a decrease in therental price of capital,17 leading to a higher probability of bankruptcy for

17Because consumption demand decreases.


entrepreneurs. As a consequence, entrepreneurs drop their demand forcapital goods, and so capital-goods producers lower their level of pro-duction. Their prices, therefore, go down. The fall in the price of capi-tal worsens the drop in investments, because it causes the net worth ofentrepreneurs to diminish.

At the end of the paper, Christiano, Motto, and Rostagno (2004) mod-eled a counterfactual example in which the monetary authority activelyreacts against the shocks, allowing the growth rate in the monetary baseto overcompensate for the reduction due to the eight shocks. This led themto argue that, had an appreciably more expansive monetary policy been inplace in the 1930s, the size and duration of the Great Depression wouldhave been much less.

Weder

The last papers to be considered in this section are both by Weder, whohas produced two accounts of the same idea (Weder (2001); Weder (2006)).The two papers share the same model, but differ in the narrative part,which is much more developed in the earlier one. I will mostly refer to themore recent (2006) paper, while occasionally referring to the earlier (2001)paper. In these papers, a dynamic stochastic general equilibrium modelof the RBC type is modified to allow exogenous shocks to the aggregatedemand for consumption to be the only impulse mechanism of the busi-ness cycle. The aim is to evaluate the impact of the consumption shockon the Great Depression quantitatively, by simulating the model. As forthe methodological concern, the model is in the RBC tradition. Neverthe-less, it has a clearly Keynesian flavour, all the more so in that Weder (2001)defines his model as an RBC formalization of Temin (1976)s view of theGreat Depression as a phenomenon mainly caused by a contraction of theautonomous components of aggregate demand for consumption.

In Weders model, households are thought of as maximizing a logarith-mic utility function with a random variable affecting the subsistence levelof consumption. The model also includes variable capital utilization, or-ganizational synergies and increasing returns to scale in the production


function. Weder (2006) identifies the preference shifter econometrically.18

He calibrates his model, largely on the basis of Cole and Ohanian (1999),and runs a simulation. It turns out that the model with increasing returnsmatches the trend in US output, explaining around 59% of the collapse,and almost all of the slow recovery and the 19371938 recession.

An interesting point, which is developed in Weders (2001) paper butabandoned in the drier (2006) one, is the explicit comparison betweenWeders model and the competitive and cartel models discussed by Coleand Ohanian (2001). Weder (2001) points out that his model can mimic theonset of the Great Depression as well as the slowness of the recovery (re-producing about 80% of the variance in the data correctly), whereas Coleand Ohanian (2001)s competitive model can explain only about 40% of theonset of the depression, and very little of the recovery phase. Moreover,Cole and Ohanian (2001)s cartel model can only explain 5060% of the re-covery phase. In addition, Weder argues that his model can reproduce the1937 recession, which other models cannot.

To investigate further which model explains the data in a statisticallymore appropriate way, Weder (2001) runs a regression of actual US outputon the predicted output of three models (his own, Cole and Ohanian(2001)s competitive model and Cole and Ohanian (2001)s cartel model).He finds that the predictions of his model are statistically more significantthan those of the other two. When output from his model is added to theregression, the other two lose any explanatory power, meaning that thenull hypothesis (that they do not explain US output at all) cannot be re-jected. When only the recovery period is considered, the explanatorypowers of his model and Cole and Ohanians cartel model are equally sta-tistically significant. For both models the null hypothesis is rejected at 1%significance level, with regression coefficients of 0.57 and 0.41 for the cartel

18Weder first derives a Euler equation from the first-order conditions for the house-holds utility maximization problem. He then linearizes the Euler equation, taking aTaylor approximation of it. Finally he uses ordinary least squares to regress the formulahe obtained on the data, and takes the residual from the regression as the preferenceshifter. The dynamic process of this preference shifter is then found econometrically tobe second-order autoregressive, of the kind t = 1t1 + 2t2 + dtc . Weder (2006)uses this AR(2) to compute a shock series {dtc }

19391930 from the data.


and the demand-driven model, respectively. Weder (2001) concludes that:

Judging the overall performance, the demand-driven modelfares at least as good [sic] as its considered contenders. (Weder(2001), p. 18)

2.3.3 A sunspots neoclassical interpretation of the US GreatDepression

All the models reviewed above fit the definition of RBC models developedin Section 2.2, in that they respect the equilibrium discipline, and businesscycles are assumed to arise from exogenous shocks to the fundamentals.A slightly different approach to the Great Depression was proposed byHarrison and Weder (2002). In this paper the authors stuck to the equilib-rium hypothesis, but business cycles were assumed to be driven by animalspirits (or sunspots), i.e. self-fulfilling expectations not related to the fun-damentals of the economy.

This model is a variation on Weder (2001)s theme. Instead of assum-ing an exogenous shock to preferences, and setting parameters to solve thedynamics for a saddle path, Harrison and Weder (2002) set the parametersso as to allow for bubbles. The possibility of animal-spirit-driven businesscycles arose in their model because they assumed sufficiently increasingreturns to scale to ensure the existence of multiple equilibria. A high de-gree of increasing returns to scale actually ensures that optimistic or pes-simistic expectations will be self-fulfilling. Consumers will move savingsaccordingly, labour supply will shift and capital utilization will vary. Vari-ations in capital utilization will mean variations in labour demand up tothe new equilibrium, at which point expectations will actually have beenfulfilled.

Harrison and Weder (2002) identified non-fundamental shocks to thedegree of confidence by means of a vector auto-regression (VAR) model.They assume that the interest rate spread between high-risk and low-riskbonds is a reasonable proxy for the degree of confidence. Running twoalternative versions of the VAR, plus a Granger causality test, they claim


that residuals from the VAR specification, in which the spread of interestrates is the most exogenous shock to the system, do Granger-cause out-put.19 Subsequently, they used the sunspot shock series generated by thisprocedure to compute the output, consumption, investment and produc-tivity series implied by the model.

The findings fit the data well. Provided the increasing returns to scaleare large enough, the model reproduces stylized facts better than Coleand Ohanians competing models. Both the sharpness of the downturnand the slowness of the recovery are accounted for by sunspots shocks.Moreover, Harrison and Weders model reproduces the 19371938 reces-sion, which all the other models fail to do.

These results led them to give the following account of the Great De-pression:

The 1929 stock market crash was followed for about a year bywhat appeared to be the start of a normal recession. Only later,during the summer of 1930, did confidence began to deterio-rate dramatically. Hence the recession was transformed into adepression. In 1932, faith in the economy hit bottom; and thecontinuing sequence of pessimistic animal spirits are a primecandidate in the quest to explain the subsequent stagnationthat only ended with the onset of World War II. (Harrison andWeder, 2002, p. 26)

Extending the analysis, Harrison and Weder tested their model overa longer period starting from the end of Great Depression and ending in2000. Results in this case were poor. They concluded that

demand shocks [that is sunspots] were either less important orsmaller in the postwar period or were partially neutralized byactive fiscal and monetary policies. (Harrison and Weder, 2002,p. 25)

19In the sense that the null hypothesis that the residuals do not Granger-cause outputis rejected at a confidence level between 5% and 2%, depending on whether 4 or 8 lagsare used (Harrison and Weder (2002), Table 3, p. 17)


2.4 Great Depressions worldwide

The RBC interpretation of the Great Depression outside the USA is madeup of two elements. The first is a critique of what RBC theorists callthe consensus view,20 stressing the role of deflation and nominal wagestickiness in the diffusion of the depression from the USA to the rest ofthe world. The basic idea of the consensus view is that adherence tothe gold exchange system induced restrictive monetary and fiscal poli-cies in the presence of serious deficits in the balance of trade, or in or-der to avoid them. These policies are normally deflationary, and deflationcauses unemployment, unless nominal wages decrease. The second ele-ment is a case-study analysis of a number of countries, applying an iden-tical methodology and theoretical setup to each country. These studies,it is claimed, demonstrate that idiosyncratic shocks to TFPs and country-specific economic policies provide a fairly good explanation for the GreatDepression in each country, without any reference to an international di-mension.

2.4.1 The critique of the consensus view

The arguments presented by Cole, Ohanian, and Leung (2005) for rejectingthe consensus view are empirical and mainly based on the signs of thecorrelations between log deviations from the trend-lines of real wages andoutput, and prices and output. According to these authors, if the con-sensus view were right, there should be a positive correlation betweenthe rates of growth of prices and real output, and a negative correlationbetween the rates of growth of real wages and real output. In other words,pinning things down to a traditional labour supply and demand graph, weshould observe an upward and leftward movement along the labour de-mand schedule, with increasing real wages and decreasing employment.

Studying cross-sectional data on 17 OECD countries, Cole, Ohanian,and Leung (2005) noted that when regressions were performed on the

20The term consensus view is used by Cole, Ohanian, and Leung (2005) to refer topapers by Bernanke (1995), Bernanke and Carey (1996) and Eichengreen and Sachs (1985).

2.4 Great Depressions worldwide 33

cross-sectional averages for 19291932, the correlation between the log de-viations of prices and real output turned out to be slightly negative, whilethe correlation between the log deviations of real wages and real outputwas moderately positive. This observation led them to conclude that thecause of the international Great Depression could not be sought in a move-ment along the labour demand curve, but rather should be found in amovement of the labour demand curve. To model this hypothesis, theyconsidered an RBC model with money a la Lucas. In this model, the econ-omy can be hit by two shocks: a monetary shock, causing a movementalong the labour demand curve, and a productivity (TFP) shock whichshifts the labour demand curve. Cole et al. calibrated the parameters ofthe model so that the two shocks taken together reproduced the data setas exactly as possible. They then tried to work out, for different orthog-onalizations21 of the two shocks, how much of the movement of the totalquantities during the Great Depression could be explained by each fac-tor. They found that a country-specific TFP shock orthogonal to deflationcould explain two-thirds of output variation in each country, while mone-tary shocks explained the remaining third. Moreover, their artificial seriesof TFP shocks matched the small amount of data available for economy-wide productivity during the 1930s.22 On the other hand, the same simula-tion carried out with only the monetary shock (that is without TFP shocks)produced a strong negative correlation between real wages and real out-put (in log deviation terms), which is at odds with the cross-sectional evi-dence.

On the basis of this analysis, Cole, Ohanian, and Leung (2005) con-cluded that an RBC account of the international Great Depression shouldbe based on a shock that works like a productivity shock, that is orthogo-nal to deflation, and that looks like a productivity shock in the data. They

21Two random variables x and y are said to be orthogonal if their cross moment E(xy)is zero. In the present case, the favoured procedure is one that orthogonalizes the TFPshock on deflation. This means that the authors regress TFP on deflation, and then sub-tract the value of TFP obtained by the regression from the actual TFP value. In this waythe residual TFP is not correlated with deflation, as the effect of deflation on TFP hasalready been taken into account by the regression.

22These data refer to the USA, Canada, the UK and Australia.


suggest that natural candidates for such a shock are the financial disrup-tions stressed by Bernanke (1983), the decrease in information capitalhypothesized by Ohanian (2002) and policy interventions that obstruct thenormal working of the market forces, as in Cole and Ohanian (2004).

The analysis presented by Cole, Ohanian, and Leung (2005) deservessome critical discussion. The 19291932 data show that a positive log de-viation from the trend of real wages was accompanied by a negative logdeviation from the trend of output in 13 of the 17 countries considered.This means that the relationship between real wages and output was neg-ative in the vast majority of countries. True, the interpolation of the plotteddata gave an upward-sloping line. But the observations in the plot werehighly dispersed, so that the R2 was very low. Moreover, considering thatthe countries under consideration differed substantially in this period, thefact that an international increase in the rate of growth of wages was ac-companied by a diminishing rate of decrease in output does not necessar-ily mean that as long as real wages increase in each country, we shouldexpect a parallel increase in real output. Many other factors that have notbeen taken into account here could influence the results for example, in-ternal political factors (such as the role of unions and of socialist parties),international political factors (such as war reparations and war debts) andexchange problems in connection with problems in the balance of trade.

2.4.2 Case studies

The case study analyses are all contained in a special issue of the Review ofEconomic Dynamics. Four of them, concerning, respectively, Canada, Ger-many, France and the United Kingdom in the 1930s, will be consideredhere.23

23For the sake of completeness, the other papers in the issue concern Italys mild de-pression of the 1930s, Japans crisis in the 1990s and analyses of South-American coun-tries depressions in recent years. I omitted Italy, because the Italian depression wassmaller than the others and a bit peculiar (Perri and Quadrini (2002)). Instead I havefocused on Canada, to compare it with the United States.

2.4 Great Depressions worldwide 35

Canada

Amaral and MacGee (2002) carried out a comparative analysis of the GreatDepression in Canada and the USA, using an RBC model that is formallyequivalent to that used by Cole and Ohanian (1999). Their principal re-sult is that an exogenous shock to TFP could reproduce about 50% of theCanadian depression. This shock also performed well in accounting forthe slow recovery. Moreover, building on arguments by Cole and Oha-nian (2000), they excluded the possibility that monetary factors could haveplayed a major role in causing the Canadian Great Depression. Finally,they tested the importance of terms-of-trade shocks in explaining the de-pression. During the 1930s, Canadas economy was small, and trade con-stituted a high proportion of GDP; trade shocks were certainly appreciableat that time. The test was done by running a simulation on a two-countryRBC setup, under the limiting assumption that inputs are non-tradablegoods. The results show that terms-of-trade shocks are unable to accountfor the Great Depression in Canada.

The comparison between Canada and the USA is interesting, althoughpuzzling. It shows that, in spite of some similarities in the general eco-nomic trend between the two countries, the USA experienced a recoverystarting in 1933, while Canada did not. The US recovery was character-ized by a strong TFP recovery. TFP, in effect, came back to its trend levelby 1937 in the USA, while it remained below the trend level throughoutthe 1930s in Canada. Interestingly enough, the time of recovery coincidedwith the implementation of New Deal policies in the US, while Canadahad no such policy. On the other hand it is surprising to note that, whileall the aggregate variables suggest that from 1933 onwards the USA wason the path to recovery (unlike Canada), the total hours worked increasedmore rapidly in Canada than in the USA during this time.

Amaral and MacGee (2002) tried to solve these problems by using Coleand Ohanian (2001)s ideas; they argue that New Deal policies in the USAaffected labour employment negatively, and therefore measured TFP (whichis a residual) tended to be, ceteris paribus, higher in the USA.

In my opinion this explanation conceals some logical pitfalls. Amaral


and MacGee (2002) argue that the economy in the USA recovered earlierthan in C

real business cycle models of the great depressionperso.uclouvain.be/luca.pensieroso/tesi...

Documents