(The Causes of the 1929-33 Great Collapse: A Marxian Interpretation, by James Devine)

[introduction] [section I] [section II] [section III] [section IV ]

[end of this section] [appendices] [references] [short version]

* Earlier versions of this paper were presented at the Delta Marsh Seminar of the Economics Department of the University of Manitoba on October 25, 1991, at the departmental seminar at Loyola Marymount University on February 3, 1992, and at the "New Urban and Regional Hierarchy" conference at the University of California, Los Angeles, on April 25, 1992. Thanks to Clarence Barber, Robert Chernomas, Ardeshir Sepehri, and other participants in those seminars for their extremely useful comments. Thanks also to Paul Burkett, Gérard Duménil, Mark Glick, Sanford Jacoby, Michael Lebowitz, Dominique Lévy, Tom Michl, Michael Perelman, Paul Zarembka, and the anonymous editors of Research in Political Economy who commented on earlier drafts. All sins, either of commission or omission, venial or mortal, are mine alone.

(1) This work goes further than my earlier [1983, 1988] papers on the subject, because it deals with recent literature, presents a more complete perspective, especially concerning international political economy, and introduces new data. Compared to the 1983 article, there is a greater emphasis on uneven development among sectors on both the national and international levels.

(2) Capitalism is implicitly assumed to be stable. Typical is Temin [1976] who starts with "precipitating factors" and moves to analyze their effects without discussing the nature of the system being hit by those shocks. In his later work [1989, 1993], he stresses conjunctural institutions as causing the Collapse.

(3) This summarizes some of the conclusions of Devine [1992]. The most important institutions also can produce a surplus, have relatively unchanging laws of motion, and are independent of other structures. Which structures are most important also depends on what question one is asking (one's standard of importance).

(4) One could look for deeper causes than capitalism, for example, class society in general, bound up with the institution of the state per se, or the conditions of scarcity dominating human history so far. However, these seem too general to be relevant to the specifics of the 1930s. The demand-driven business cycle and depressions (as opposed to hard times due to bad harvests) arose with capitalism (and not with class society or scarcity) and seem specific to that mode of production.

Similarly, though patriarchy and racial domination are key (and durable) institutions in our social formation, it is hard to see how they contributed to the origins of the Collapse, except perhaps by dividing and weakening labor and thus encouraging stagnant consumption.

(5) For more on the Crash, see sections III.E.

(6) The rest of this paper does not stress this element, because excessive leveraging was restricted to only a few industries, as indicated by the corporate debt figures reported by Goldsmith [1955].

(7) The following survey is not meant to describe Marx's theory of crisis or Marxian crisis theory in general, but my version of Marxian crisis theory. On both of the former, see Clarke [1993], whose excellent critical survey helped my presentation immeasurably.

(8) It should be stressed that even if crises are inevitable, revolution is not: unlike capitalism, the development of working-class and other opposition forces does not behave in a law-like or inevitable manner.

(9) See Crotty [1993] for a recent extension of Marxian investment theory, which is consistent with the discussion here.

(10) Moseley [1991] argues for the use of the price-based profit rate. On the other hand, he sees the conventional profit rate as empirically relevant, but as determined by the Marxian rate. This paper skips Moseley's step of first analyzing the Marxian rate and then examining the effects of its fluctuations on the conventional rate. The existence of "indirectly productive" labor (unproductive labor that promotes the productivity of productive labor) makes the distinction between productive and unproductive labor -- and thus the two profit rates -- fuzzy [cf. Laibman, 1992, ch. 4]. Also, the role of unproductive expenditure in lowering conventional rates seems relatively unimportant when those rates are rising, as in the 1920s.

(11) A falling conventional profit rate tightens constraints on funds available for accumulation, decreases cash-flow compared to previously-fixed interest-payment obligations, decreases the incentive to invest in fixed capital (as opposed to, say, real estate speculation), encourages pessimistic expectations, and (when relative to other countries) spurs movement of capital out of that country.

(12) This presentation originates from Weisskopf [1978].

(13) Devine [1990] presents a preliminary examination of the nature of the connections between value and price categories on the macro-societal level.

(14) Devine [1983] rejects one Marxian underconsumption theory specific to the twentieth century, that of Baran and Sweezy [1966].

(15) This is an extension of Marx's discussion in the Grundrisse cited above and volume III of Capital [1894: ch. 15]. It is not a theory of over-accumulation per se.

(16) This might be seen an extension of Engels' (and Marx's early) theory summarized by Clarke [1993: chs. 2,3]. In this theory, capitalist competition drives accumulation and production ahead of consumption, whose growth is determined by quite different laws of motion. As developed in section I.B, the growth of workers' consumption depends on labor-power market conditions.

(17) This is the presentation in Marx [1867].

(18) Again, the emphasis on fixed rather than constant capital arises from the emphasis on imbalances that block accumulation.

(19) Gordon, Weisskopf, and Bowles [1983] see this as the distinction between a "reproductive" and a "nonreproductive" cycle. In the former, the crisis helps reproduce long-cycle prosperity.

(20) See Aglietta [1979], Lipietz [1987: ch. 2], and Piore and Sabel [1984: ch. 4]. For the similar "Social Structure of Accumulation" (SSA) school, see Gordon, Edwards, and Reich [1982] and Bowles, Gordon, and Weisskopf [1983]. For an effort to synthesize these views, see Marglin and Schor, eds. [1991].

(21) In turn, Regulationists argue that the Depression helped spur the rise of "monopoly regulation" that allowed more stable growth after the second World War. Together, intensive accumulation and monopoly regulation form "Fordism," which is seen as persisting from World War II until the 1970s.

(22) Lipietz [1987: 39-40] introduces the importance of a world-hegemonic power to Regulation theory. See also Marglin [1991].

(23) Included as part of the increasingly overt socialization of production is the concentration and centralization of capital. This tendency also involves the greater role of governmental, quasi-governmental, and other non-market institutions (formal and informal) in organizing and regulating production and circulation.

(24) Part of the theoretical debate between the partisans of the three crisis theories arises because of Marx's deliberate decision in Capital not to deal with the determination of the value or supply of labor-power (the "laws" of motion of working-class struggle) in any detail [cf. Lebowitz, 1992]. Instead of developing a full-scale political economy of the working class in order to explain labor abundance and scarcity, however, this paper takes labor-power supply conditions as historically given. For one effort at such explanation, see Lembke [1991-92].

(25) The first hegemonic war (the Thirty Years' War), established the Netherlands as hegemon, while the second, the wars of 1793-1815, put the U.K. in that role.  

(26) This Monopoly Capitalism differed drastically from the stage of the same name described by Baran and Sweezy [1966], in which competition (even among nations) was severely limited.

(27) The increasing size and integration of world markets, along with the development of new industries, are counter-tendencies to the centralization of capital. So monopolization of markets is hardly permanent or universal. But during the period before 1929, the centralization tendency dominated. This was partly due to the absence of a hegemon playing the role of breaking down trade barriers, i.e., the absence of a G.A.T.T.

(28) Here and below, the term "elite" is used to refer to policy-makers and power-holders in the government. These are often representatives of fractions of the capitalist class; though enjoying a certain relative autonomy, their policies are always limited and shaped by class pressures from outside the state.

(29) Kindleberger's [1986: 291] explanation of Britain's commitment to free trade by "cultural lag and the . . . tradition of Adam Smith" is insufficient, given the U.K.'s denial of free entry of many colonial goods to Britain, as with Indian textiles, and its suspension of free trade during World War I.

(30) However, when the European power lost, as in Italy's Ethiopian debacle (1896) or the Russo-Japanese war (1905), it stimulated domestic revolt.

(31) Military build-up also helped moderate industrial over-production, but it is hard to imagine that this was a conscious military Keynesian policy.

(32) For one standard summary and analysis of that treaty, see Smith [1965: ch. 9]. See Williams [1959] for a leftist interpretation.

(33) This is similar in certain ways to recent events in the former Soviet Union: the split has conflicted with the strong economic interdependence of the old system, further undermining the ability to produce.

(34) Moreover, Japan was treated as a second-class world citizen (as later, when the League of Nations rejected Japan's proposed "Declaration on Racial Equality"), encouraging enmity in the Pacific.  

(35) The Collapse encouraged the colonization of the periphery to restart in Manchuria in 1931 (for Japan) and Ethiopia in 1935 (for Italy).

(36) After the resources and land have been taken from the indigenous populations, and much of the latter incorporated in forced-labor systems, returns are lower and the maintenance of social order increasingly costly.

(37) According to these authors, the trends that were disrupted by the War were the pattern of world trade and the stable gold standard.

(38) On the myth of U.S. isolationism, see Williams [1959: ch. 4]. For example, the U.S. continued its regular intervention south of the border (as in Nicaragua).

(39) Kindleberger distinguishes between "hegemony" and "leadership," noting the latter's positive connotations and thus embracing it [1986: 289n]. But hegemony and leadership are typically complementary, with the former being more crucial. It is a country's economic and military hegemony that allows it to have world influence and to be able to easily afford the costs of leadership. It is true that any hegemon that does not act in some positive ways that legitimate its domination -- i.e., acts as a "leader" -- does not last for long: the Roman Empire's Pax helped it to survive. Still, a country can be hegemon without being a leader, while it is the former which allows the latter.

(40) The Bretton Woods plan was in fact based on an analysis of the world economy strongly akin to Kindleberger's.

(41) For a long time after World War II, its power made the U.S. able to simply print dollars and have them accepted as being "as good as gold." This seignorage power made a more successful Restoration easier.

(42) On corporate liberalism, see Weinstein [1968]; on War-time planning, see Scheiber et al. [1976].

(43) See Hirschman [1968: 208, 222]. For one formal model of the process of the fight over the nominal product leading to inflation, see Rowthorn [1977].

(44) A similar formulation (for more moderate inflation) can be found in Bowles et al. [1983: 116f].

(45) From the point of view of nations and classes dominated by the Global Cop, the story is much less rosy. The benefits of this stability have been concentrated in the advanced capitalist nations.

(46) Marx himself applied this insight to the domestic arena, seeing world money as being inherently metallic.

(47) The idea of a freely-floating exchange rate system does not seem to have been seriously considered in the 1920s. Policy-makers lacked the optimism of Milton Friedman, the major advocate of totally-free exchange rates. The stresses and inflation of the floating exchange-rate system that prevailed (by default) during World War I may have undermined such optimism. Further, being deflationists, they valued the international discipline the gold standard imposed on the domestic economy.

(48) Winch [1970: 75] argues that "When the Cunliffe Committee recommended that the overriding aim of British policy in the post-war period should be to return to the gold standard at the pre-war parity, they commanded the overwhelming support of informed opinion and sentiment of the day. The issue was settled as soon as it was raised; there was no mention of any alternative." Another government committee saw this step as being "a decisive step towards the reconstruction of the international economy. It would restore London to the centre of the international financial network, and, it was hoped, create conditions favourable to the revival of multilateral trade" [Winch, 1970: 84]. Winch also points to fear of inflation as central to the decision.

(49) This dependency arises because both the demand for and the supply of primary products is typically price-inelastic.

(50) This differs from the 1990s, in which the Bush and Clinton administrations seemed willing to sacrifice the U.S. farming sector's subsidies in the G.A.T.T. discussions. Not only is the farm sector now relatively weaker in terms of political clout, but perceptions now are that U.S. agriculture (or at least the larger and more influential farmers) would win with freer trade. 

(51) Recent authors who do not fall for this fallacy include Dowd [1974: 91-6], Scheiber et al. [1976], and DuBoff [1989: 86-9].  

(52) For more information on all diagrams, see the Appendix A.

(53) This refers to the mining, electric light, power and gas, construction, transport, communication, trade, finance, services, and miscellaneous sectors.  

(54) See Scheiber et al. [1976: 348-50], Devine [1983], Jacoby [1985: ch. 6], and Lazonick [1990: ch. 8] for more details and references on this era. 

(55) This rate is measured by union membership (either series D-940 or D-941 from USDC [1975: 177]) divided by the labor force (series D-4, p. 126). The downward slide far exceeded the degree of the falls after 1904 and 1913. Work stoppages [USDC, 1975: 179, series D-977] fell drastically after their peak in 1917 and their minor peak in 1919-20.

(56) The ratio of immigrants aged 14-44 to the labor force fell from 3.2% in 1907 to about 0.5% in 1929 (calculated from USDC, 1975: series C141, D1 (pages 112, 126), assuming that the ratio of the 14-44 age group to the 16-44 age group is constant). However, among the advanced industrial countries, population growth in the U.S. during the 1920s was second only to Australia's [Maddison, 1982: 184-5].

(57) In the 1920s, the increase in the labor-force participation rate of women canceled out much of the effect of the men's declining labor-force participation rate. Between January 1920 and April 1930, women's LFPR rose from 22.7% to 23.6%, especially among those older than 20 years of age. Men's LFPR fell, from 84.6% to 82.1%. Over-all, the LFPR stayed relatively constant, at 54.3% and 53.2% [USDC, 1975, volume 1: 132, series D-29, 30, 36].

(58) Missed is the extent to which small business was "semi-proletarianized," i.e., the extent to which they had to work part-time for wages. Also not measured is the degree to which labor-abundance and unemployment may have encouraged people to seek or preserve self-employment, so that the slowdown in proletarianization was a symptom of labor abundance.

(59) USDC [1975], series D-831 (vol. 1, p. 172) started rising. Series D-803 (vol. 1, p. 170), on the other hand, leveled out. Because they are collected from employers rather than from workers, these numbers do not capture the effects of moonlighting.

(60) The effective labor supply is non-agricultural proletarian labor-hours times output per hour, while its growth rate is Harrod's "natural" rate of growth (representing the dynamic labor supply constraint on the "warranted" growth of output).

(61) Rural-urban migration contributed 13%, with proletarianization trivial (1%) and hours per worker actually hurting (-7%). These do not add to 100% due to rounding.  

(62) Given data problems, it was necessary to include the self-employed (entrepreneurs) as part of the effective supply. But as noted, this was trivial. 

(63) In all cases, ratios of the unemployed to the employed were converted to ratios of the unemployed to the labor force (employed plus unemployed).

(64) This ratio shows a steady and relatively smooth decline over the period. Replacing the actual ratio with a exponential or linear trend in the calculation of the "estimated Romer" series does not change the results substantially. 

(65) We cannot reject the hypothesis that the Douglas numbers (either non-extrapolated or using the two different extrapolations) have no time trend (t-stats = 1.29, 1.32, 1.32). (Against the log of time, these t-stats are 1.16, 1.27, and 1.25. His two other series [1930: 458, 460], which add construction and mining (but do not include 1927), are highly correlated with this series and have similar correlations with time.) On the other hand, the t-statistic on the time-coefficients for the Lebergott and estimated Romer series are significant (t-stat = -2.00, -2.48), indicating a downward trend in unemployment. Against the log of time, these t-stats are -1.55 and -2.24.

(66) One can only speculate about the extent to which unemployment was structural as opposed to being part of the (floating) reserve army of the unemployed. Gordon's [1987: 582] numbers for structural-frictional (mis-named "natural") unemployment are entirely conjectural and only for the economy as a whole.

(67) This effect is similar to the way in which high unemployment rates in the early 1980s had effects that persisted into the late 1980s and 1990s.

(68) Following the conflict theory of inflation presented in section II.D., the downward drift of prices during the 1920s is also a symptom of the absence of worker power. The wholesale price index for nonfarm, nonfood commodities went from 161.3 in 1920 to 100.6 in 1923 to 100.0 in 1926 to 91.6 in 1929 [USDC, 1975: 200, series D41].

(69) See Kendrick [1961: 65-71]. A similar (and more meaningful) break can be seen in output per labor-hour in manufacturing: using Kendrick's [1961: 465-6], this productivity measure rose 1.6% per year for 1899-1919 (t-stat = 14.37) and 2.5% for 1919-1957 (t-stat = 25.48). Regressing the log of productivity against a time variable for the whole period and also one for the period after 1919 gave a t-stat of 5.74 for the coefficient of the second time trend, suggesting that the kink was significant.

(70) The differences seem to arise from the latter series' emphasis on long-term trends (so that it does not fall in the post-War recession, unlike all other series) and more importantly from different data used in the calculation.

A simple regression of the Duménil/Lévy profit rate against time for the period 1899 to 1929 indicates a positive but insignificant trend. The profit rate fell from the beginning of their series (1869) until 1929 and generally rose from 1890 until 1929. Unfortunately, the quality of the data falls as we go further back.

(71) Unfortunately, data limitations mean that a variety of different methods of calculation are used. When different series lead to the same result, however, this strengthens the certainty of results.

(72) Discussion of the third determinant of r in equation (1) is delayed until section D.  

(73) The regression coefficient against time using the D/L series was insignificant, that of the merged manufacturing series was significant.  

(74) This concept (a hypothetical rate measured at full capacity utilization) is analogous to the "high employment government budget." See Appendix B, equation B-3.

(75) This equation is derived in part 1 of appendix B. Marx's "reproduction schemes" [1967b] for expanded reproduction is an obvious precursor of the Harrod-Domar model and the present formula. The present model is far simpler than Marx's because it is not split between departments I and II, and leaves out the role of money.

(76) See part 2 of appendix B. 

(77) The ratios were derived from USDC [1975: 231-2], with I from series F112, H from F108+F109, and C from F100. These calculations suffer from the obvious problems arising from the inclusion of Collapse years in the end-point data. 

(78) Corporate debt may also be a barrier to recovery, but this was not a major problem (until debt-deflation set in) because the existence of abundant profits for internal financing and the emphasis on equity-financing in the late 1920s. Goldsmith (1955a: 210) presents ratios of debt to total assets for different sectors over the longer haul: for nonfinancial corporations, there is a downward trend after 1912, equaling 41.9% in 1900, 50.0% in 1912, 40.8% in 1922 and 40.5% in 1929. This ratio rose for farm and non-agricultural households, banks, and miscellaneous financial institutions over this period. It fell for unincorporated businesses and nonfinancial nonprofit institutions. For a deeper analysis, see Isenberg [1987]. 

(79) After the collapse of the leading sectors, both the exports and the imports of the U.S. fell, as world trade imploded.  

(80) The reasons for the fall are beyond the scope of this paper, while the exact timing of the peak is not crucial here.

(81) This fits with Keynes' view in the Treatise [quoted in Temin, 1976: 31]. 

(82) Strangely, the latent instability of the U.S. economy seems implicit in Friedman and Schwartz's view that the stock market was more robust than the economy: in the late 1920s, Federal Reserve policy was "too easy to break the speculative boom, yet too tight to promote healthy economic growth" [quoted in Temin, 1976: 23].

(83) Unlike Romer, Galbraith [1961: 182-91] looks for and finds factors that made the U.S. economy "fundamentally unsound" in 1929, so that the Crash, a relatively small trigger compared to the size of the U.S. economy, could trigger the Collapse. These were the highly unequal distribution of income, the highly leveraged corporate structure, the inability of an extremely decentralized banking system to transfer funds to units suffering from illiquidity, the difficulty foreign debtors were having making payments, and the poor state of economic policy knowledge. These are unobjectionable and hardly contradictory to this paper's view.

 (84) Here and elsewhere, the "accelerator" simply refers to the negative effects of decreasing aggregate demand on investment, rather than to some more specific theory.

 (85) While the debt-deflation process eventually clears the system of many imbalances (debt overhang, inefficient firms, etc.), the process is extremely painful and can cause social disorder or worse. It also does not solve the problems of unused capacity and pessimistic expectations automatically, but can make them worse for a long period.

 (86) In this view, the dividing line between the powerful hegemonic power willing to make short-term sacrifices because of the perceived benefits of world economic stability and the current tendency to subordinate world stability to short-term domestic concerns seems to be the "Nixon shocks" of the early 1970s (in which Nixon unilaterally jettisoned the Bretton Woods system).  

(87) As with the U.K. during the 1930s, however, a serious depression could cause a shift in political power and encourage the end of free trade, which in turn would make the depression worse.

(88) On the other hand, there is no reason to expect trade liberalization to progress quickly on all fronts, since the farm sectors of many nations oppose it. Part of the fuzzy thinking in the area arises from the "either liberalization or trade war" rhetoric used by some of its partisans.

(89) Though the former Yugoslavia and Soviet Union are major sites of armed conflict, it seems unlikely that conflicts there would spark war between, say, France and Germany or Japan and the U.S. It is also hard to see them as ever having been in the core of the world system.

(90) "Autocentric" (articulated) growth is based on domestic consumer markets, while dependent (disarticulated) growth is based on exports [de Janvry and Garramon, 1974].

(91) The emphasis is much more on automatic stabilization as opposed to discretionary policy.

(92) Full discussion of the downfall of this system is beyond the scope of this paper; only its outlines are sketched.

(93) It should be mentioned, however, that in many cases, the "trickle down" did not work. For example, many periphery nations and the core's minority groups were left out of the prosperity. As usual with "golden ages," the prosperity looks better in retrospect than it did at the time.

(94) This trend even included Mitterand's Socialist Party government in France. 

(95) This is likely to be too optimistic in that such spending would have to paid for out of wages -- in order to avoid alienating transnational capital.

 (96) See also Bluestone and Harrison [1988], Winnick [1989], and Nasar [1992a,b] for the U.S.

 (97) An important caveat must be stated here: it is possible that a disaggregated examination of profit rates would come up with a different story, as did the study in section III. This is a subject for further research.

 (98) See Armstrong et al. [1991: 248-54] for similar profit rate behavior for the core capitalist countries as a whole, and the impact of this on accumulation.

[introduction] [section I] [section II] [section III] [section IV ]

[top of this section] [appendices] [references] [short version]