Evaluating the historiography of the Great Depression: explanation or single‐theory driven?

Abstract

Following James Rule and Colin Clark, a single‐theory‐driven approach to scientific inquiry which focuses on testing particular theories can be distinguished from an explanation‐driven approach which is open to all observations and whose results do not cease to have value with the passing of a particular theory. Several ‘decision points’ in the historiography of the Great Depression are examined and it is shown that the decisions made at each point reflect a single‐theory‐driven orientation. It is argued that the single‐theory‐driven approach likely characterizes other fields of Economics, and also that this paper's method of applying a ‘test’ to various decision points in the evolution of economic thought is of widespread utility. The likely cost of a single‐theory‐driven bias in terms of our collective understanding is discussed.

Keywords:

• Great Depression
• historiography
• methodology
• macroeconomics
• structuralism
• technology

ACKNOWLEDGEMENTS

The author thanks Ross Emmett, David Laidler, Thomas Mayer, and Roger Sandilands for advice and encouragement. Kevin Hoover made interesting remarks as a discussant at the History of Economics Society, Vancouver, July 2000. I have most of all appreciated the sage counsel of Roger Backhouse, and the referees he consulted; the paper is much better because of their efforts.

Notes

This paper will provide evidence that, at least in its approach to understanding the Depression, the core favoured by economists has included ‘use only macroeconomic theory’, ‘use only mathematical modeling and/or econometrics’, and perhaps ‘defend the market system’.

Fox (1997) and Hausman (1992) are among many who worry that data limitations in Economics are such that economists may never reject a theory because anomalies can too easily be attributed to the data rather than the theory. This must enhance the importance of engaging alternative theories and evidence.

The advantages of utilizing multiple theories are outlined with respect to the scholarly enterprise as a whole in Szostak (2004), and with respect to economic history in particular in Szostak (forthcoming).

The final component of Tobin's definition has been weakened of late by the search for ‘micro‐foundations’ of macroeconomic behaviour (inspired by a laudable concern that macroeconomics violates elements of the microeconomic ‘hard core’; Hausman 1994). The attempt to wed microeconomic and macroeconomic theory can still be distinguished (at least so far) from openness to novel theories. It is an open question as to whether an exploration of sectoral interactions will lead to a greater general appreciation of structural approaches; the analysis in this paper suggests pessimism.

For most of this century, technology was viewed as exogenous by economic theory, and largely ignored – with some exceptions such as Allyn Young (Coombs, Saviotti and Walsh 1987). In recent decades there has been a flurry of activity concerned with rendering technology endogenous. I would argue that the profession needs to recognize both that technology is important and that the rate of innovation is influenced by a range of non‐economic phenomena. Modern Real Business Cycle theories are macroeconomic theories, despite claiming to embrace technological innovation: they use measures of aggregate productivity as a proxy for ‘technological innovation’, and thus add one more aggregate variable to macroeconomic analysis. They have in any case been applied almost exclusively to postwar data. As noted in Szostak (1999), the assumption of intertemporal substitution does not fit interwar unemployment experience.

Temin (1976) does refer to ‘structural deflation,’ an idea he credits to Kindleberger that deflation has differential effects by sector. Temin (1976) does briefly mention Schumpeter – the most famous economist to push a technological explanation in the 1930s – but only with respect to his discussion of cycles of varying length (and one comment on expectations). Friedman and Schwartz (1963) also briefly mention Schumpeter, but only his views on the Stock Market and Federal Reserve. Haberler (1941) does mention in passing Schumpeter's suggestion that irregularities in innovation might cause investment bunching; Haberler (1958) refers to Schumpeter only with respect to price rigidity.

Hansen welcomed the policy implications of Keynesianism; he had, like many other structuralists, advocated fiscal stimulus long before. Postwar policy makers were more receptive to Keynesian than structural arguments, and Keynesian policies appeared immensely successful. Structural analyses might have been useful in the 1970s; Roosevelt had claimed that he wished to employ workers on public works while structural problems in the private sector were worked out.

‘Since recent [c. 1969] postwar prosperity has been founded upon Keynesian policy, the limitations of Keynes' understanding of the thirties has hardly seemed fascinating to economists’ (Stoneman 1979: 3). One possible explanation here is that McCarthyism frightened economists from any argument that might suggest the market system was imperfect. Applying Keynesian theory to post‐war cycles was dangerous enough: Hansen and other Keynesians were ‘purged’ by the Eisenhower administration. To have explicitly blamed the market for the Depression would invite unwanted attention (see Leeson 1997).

Colin Clark (1940 and elsewhere) could be considered another exception, for he attempted to investigate international regularities in areas such as income distribution and productivity; it was seen above that he fought a lonely fight for a ‘scientific’ approach. Textbooks also spoke of structural theory somewhat longer than did leading theorists. Still, by the early 1970s (and likely earlier), a random sample of textbooks finds no mention of structural explanations: Harvey and Johnson (1971) discuss technology once in passing before their discussion of growth in the second last chapter, Hanson (1972) – the sixth edition – ignores technology but devotes three pages very late in the book to underconsumption, Hosek (1975) makes no mention of either technology or the Depression, and Moore (1973) ignores technology in the macroeconomic section.

Lawrence Summers has indeed criticized large‐scale exercises in macro‐econometrics for failing to inform theory (Backhouse 1994: 145).

It is not entirely clear what Friedman and Schwartz countenance when they speak of an underlying business cycle on which Fed miscues were superimposed. Perhaps they would have been open to arguments in favour of technological shocks, for example. They did speak of harvest cycles.

Weinstein (1980) suggested that the National Industrial Recovery Act was responsible for much of the unemployment after 1933. This argument, like Darby's, could potentially bridge much of the gap between theory and reality. Margo (1993) argues that Weinstein greatly exaggerated the effect of the NIRA. To accept Darby's argument, and to some extent Weinstein's, requires ignoring all sorts of evidence that the Depression was a truly horrible event. Were all those men riding the rails, all those soup kitchens, and the decline in marriage and birth rates and life expectancy just a figment of Hollywood imagination?

If I fall and break my leg, I can attribute the break to the fall without knowing the precise causal mechanism. If, however, I awake one morning and find that I cannot move my leg, I should be careful of deducing the cause of the malady without understanding the mechanism. When there are multiple possible causes, we cannot arrive at what without also identifying how.

Earlier in the book, Temin (1989: 7) discusses Keynes' emphasis on the role of investment in the Depression. While Keynes proffered many explanations for declining investment, Temin argues that the most important of these were clearly changes in monetary policy. He dismisses Keynes' arguments about the declining return to investment as ‘elusive.’ Since the rate of return on investment depends on such structural variables as technological innovation, population growth, and income distribution, discussion of why it might have fallen ceases to be elusive if economists widen their gaze beyond macroeconomics.

Reliance on expectations also solves a problem in international comparisons: the poor correlation between the rate of recovery and size of stimulus given. Since almost all countries were on the Gold Standard and practiced contractionary policies early in the Depression, and thereafter moved off the Gold Standard and pursued expansionary policy, a theory that can translate miniscule stimulus into major recovery can seemingly fit all ‘data.’ But so could many theories if they were freed from the necessity of explaining the size as well as direction of effect.

Sandilands argues that later monetarists ignored Currie's use of the quantity theory because they, unlike he, viewed monetary policy as a substitute for fiscal policy. Currie not only celebrated the New Deal but blamed the 1937 recession on misguided fiscal policy (Sandilands 1990: 38, 89). Currie's monetary approach also emphasized the structure of the banking system, the effects of differential reserve requirements for rural banks, and the economic effects of unit banking.

‘The Friedman‐Schwartz interpretation of the Depression was crucial, moreover, to the revival of confidence in market‐based economies. The Great Depression, and the way it was interpreted by Keynesian economists, convinced a generation of American intellectuals that only socialism (or near‐socialism) could save the economy from periodic economic meltdowns’ (Rockoff 2000).

Some modern macroeconomists, notably Olivier Blanchard, David Lilien, and Mark Bills, have begun to develop mathematical models in which sectoral interactions play a role in generating macroeconomic fluctuations. If this line of research continues, the methodological motive for ignoring the sectoral history of the Depression may be alleviated (though not all models lend themselves to empirical application). If so, macroeconomists will face another choice: whether or not to explore the structural explanations of that sectoral history.

Szostak (2004) details the strengths and weaknesses of each of the dozen methods used by scholars, and urges all scholarly communities to embrace multiple methods. In particular, it is noted how particular methods are biased in favour of particular types of theory. A classification of the biases and errors that can afflict all scientists is developed as well: scientists can combat each of these but only if they first recognize their existence.

Margo (1993) notes that macroeconomic theories shed little light on the sectoral experience of unemployment, and urges that these theories be supplemented by microeconomic (that is, sector‐specific) analyses. Rosenbloom and Sundstrom (1999) survey regional experience, arguing that this can be understood in terms of trend growth rates and the differential distribution of industry by region, but not the regional distribution of bank failures; they note that their conclusions are consistent with some structural approaches.

Prognostications in the popular press generally hinge on the sorts of variables that macroeconomics excludes: Are we on the verge of a microelectronic revolution?; (When) will the biotechnology revolution have a significant economic impact? Who can doubt that the answers to such questions will have a significant impact on growth and employment? So why not look beyond macroeconomics in understanding the past? Phelps (1993) argues that non‐macroeconomic arguments are needed to comprehend periods of prosperity or decline longer than the standard business cycle. In particular, ‘…the experience of the Depression needs to be restudied with alternative interpretations, structuralist theory being a strong candidate’.