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

[introduction] [section I] [previous section] [end of this section]

[section IV] [appendices] [references] [notes] [short version] [a new analysis of the future]

[[p. 141]]

Global Keynesians such as Kindleberger and Temin minimize the role of U.S. internal events in causing the Collapse. But even though this country was a major bright spot in the world economy, all was not well. To some extent, the hidden rot reflected the world situation while the U.S. represented a microcosm of the world pre-Depression: the U.S. participated in the Restoration, the deflationary policies, laissez-faire, and the gold standard. And to some extent, the U.S. economy's problems were unique and internal.

These statements conflict with the common image of the 1920s as a "New Era," a golden age perhaps analogous to that of the 1950s and 1960s. But this common vision is based on the fallacy of treating the U.S. as a homogeneous bulk. But uneven development affected the U.S. as much as it did the world economy. Much of the anecdotal evidence about "roaring 20s" prosperity comes from the sectors on the high end of this uneven development, so it is easy to forget the sectors and classes left behind.51

Diagram 1 shows the shares of different sectors in total income.52 [note on diag. 1] While the relative claims on national income of the agricultural and non-farm entrepreneurial (i.e., proprietorship, small business) sectors were shrinking, those of non-entrepreneurial (corporate) manufacturing and "other"53 were growing steadily. The usual stories of 1920s prosperity either ignore the former two sectors and focus on the latter two areas (the leading sectors) or deal only with aggregate numbers.

[[p. 143]] These stories also ignore the employers' offensive against labor. The success of this assault was partly due to the workers' weakness, which in turn reinforced that weakness, as did the technological unemployment encouraged by rising labor productivity. This allowed rising profit rates in the leading sectors and an unstable boom.

As mentioned in section I.B, this paper develops and presents evidence for a historically-specific version of a theory of underconsumption resulting from a falling profit rate, but with a long delay. The lag is crucial to the story, since the accumulated fixed capital, luxury goods, and consumer debt that allowed this long delay (roughly a decade) also meant the deepening of imbalances that helped make the U.S. Collapse so Great [Devine, 1983]. This suggests that if the Collapse had been delayed even further, it would have been even worse.

The 1920s "employers' offensive" against labor occurred partly in reaction to high War-time wages and the post-War strike wave. This offensive was part and parcel of laissez-faire in practice, and continued the War-time repression (and was in some ways preminiscent of the period of the late 1970s until the present). It was given extra fervor by low profits: as Duménil et al. [1987] and Duménil and Lévy [1993] point out, the U.S. economy as a whole suffered from declining profit rates during the first two decades of this century (1900-1920). [They call this growth à la Marx, because it fits his story of falling profits due to rising K/Y.] Most data also show a steep profit-rate fall immediately after the War. This offensive -- part of the more general Restoration -- was seen most dramatically in the Wilson-Palmer raids after the War, but also in the "American Plan" of company unions and open shops plus the repeated anti-union injunctions by the courts.54 The post-War recession and deflation (starting in January 1920) was part of this offensive, an effort to cut wages in order to raise profit rates in the longer term: in the dominant economic-policy view of the time, such recessions were seen as a "natural" way to deal with the economic imbalances including high wage demands and inflation. Given this attitude, the power elite needs not have a conscious desire for an anti-labor offensive for it to have such effects. However, the main macro-policy-making institution of the time (the Federal Reserve) was clearly anti-labor, in that it was controlled by bankers.

As a result of the offensive and the conflicts within the labor movement, workers' organizations lost power. The unionization rate peaked in 1920 and fell steeply after that, reversing all gains after 1916.55 The A.F. of L. shrank to the most elite of craft sectors, losing its political influence with the end of corporate liberalism and war-time planning. The I.W.W. and the Socialist Party (which had scared business before and during the War) faded from the scene. Workers' ability to capture productivity gains or to resist wage cuts shrank. The organizational frailty of unions and workers' political parties resulting from the [[p. 144]] employers' offensive helped create a situation of "labor abundance" for the leading sectors.

This abundance occurred despite the racist and highly restrictive 1921 and 1924 immigration laws which slowed the growth of the supply of labor.56 First, as Jacoby [1985: 170] notes, migration from Mexico helped prevent labor-power shortages, even in northern industry. Second, as part of the general phenomenon of U.S. domestic uneven development, labor migrated from the countryside to the city. The stagnation of the farm sector created an ample flow of labor to the city, continuing the trend of the pre-War period [cf. Jacoby, 1985: 171; Lazonick, 1990: 254-5]. As shown in table I, though the civilian labor force was only 68.3% larger in 1929 than in 1900, the non-agricultural civilian labor force rose by 114.8% between those years. Table II, column (6), shows that this migration slowed during the 1920s compared to the pre-War period. But it still had a significant impact on the urban labor-power supply.57 [notes on tables I and II]

Other potential sources of labor-power to the leading sectors are examined in tables I and II. First, it might be hypothesized that the secular shrinkage of the [[p. 145]] petty bourgeoisie predicted by Marx (reflected in the decline of non-farm entrepreneurs in diagram 1) provided a labor supply. In the table I, column (3) shows the growth of the estimated proletarian work-force (excluding the self-employed). Though this adjustment had an effect, it was very minor in the 1920s (see column 8 of table II), in that the nonfarm "proletarianization rate" was constant between 1920 and 1930 [M. Reich, 1986: 125].58 Next, changes in hours worked can raise the labor-power supply. Column (4) of table I and column (9) of table II show that changes in work-hours actually reduced labor-supply growth during the 1920s. Not shown, however, is a significant change, i.e., the cessation (or even reversal) after 1925 of the long-term trend toward shorter hours as shown in the series on weekly hours of work in manufacturing.59 The fact that the long-term trend halted or was even reversed is in itself evidence for labor abundance in the late 1920s, since it shows labor's low or falling ability to resist the extension of the working day. Finally, and most importantly for the 1920s, the accelerated growth of productivity during this period helped preserve labor abundance, by reproducing a high level of technological unemployment over time [cf. Jacoby, 1985: 168-70].

The effect of productivity growth on raising the "effective labor supply" is shown in column (5) of table I and column (12) of table II.60 Productivity growth contributed to 51% of the increase in the effective labor supply in the 1920s; the next most important was the growth of the civilian labor force (41%).61

Table II compares the growth of the effective labor supply with that of labor demand (non-farm Gross Domestic Product growth).62 In all three decades, demand (column 13) grew more slowly than supply (column 12). The "undershoot," i.e., the amount that labor demand growth grew less than supply, fell from decade to decade, so that during the 1920s, demand grew at only slightly more slowly than supply. Errors in measurement could mean that, in reality demand grew more than supply in the 1920s. Does that mean that a labor shortage was about to develop? Possibly. In fact, the closing gap between demand growth and supply growth might be interpreted as a harbinger of things to come in the 1950s and 1960s, when for a time, profits were squeezed by high wages [cf. Devine, 1987]. After all, according to Lebergott [1964: 512], in 1926 the nonfarm unemployment fell to less than 3%.

But a more complete examination of unemployment in the 1920s leads to the rejection of this preliminary conclusion. First, examine part A of table III. [notes on table III.] First, as shown by column (14), the average of Lebergott's nonfarm unemployment rates was high during the period 1919-29, though lower than in previous decades.63 Romer [1986] recalculated his national unemployment rates to take into account pro-cyclical changes in the labor force (to make the data consistent with current estimation practice). This makes unemployment rates less volatile, meaning that her measured unemployment is lower than Lebergott's in 1921 and higher in 1926 (i.e., 8.7% and 4.0%). She does not calculate urban unemployment rates, but column (15) shows estimates of what her numbers would look like, assuming that the ratio of urban to total unemployment rates is as with Lebergott's data.64 In [[p. 146]] general, this "estimated Romer" urban unemployment rate is higher and shows less of a downward trend than does the Lebergott series. In 1926, this number is 6.4%, significantly higher (and closer to the decade average) than Lebergott's 2.8% of the urban labor force. [[p. 147]]

Next consider an even smaller level of aggregation, which shows the most important part of the urban leading sector. Column (16) of table III shows unemployment rates in manufacturing and transportation, as estimated by Douglas [1930]. As seen in simple regressions against time, unlike the other two series, the Douglas numbers show a slight upward trend. Despite the statistical insignificance of the coefficient, even the absence of a trend is evidence for labor abundance in the 1920s: there was a continuation of the labor-power market conditions of the pre-War period. As the numbers in parentheses show, this is so even when Douglas' numbers are extrapolated to 1929, using the Lebergott and estimated Romer numbers (and a time trend) in simple regressions.65

One might quarrel with the use of a decade averages and long-term trends. One alternative way of looking at matters appears in part B of table III, which shows average unemployment rates by business cycles. All three series show upward trends from 1919-20 to 1927-29. On the other hand, if we start the comparison with 1921-23, which includes the severe post-War recession, matters appear differently: unemployment is relatively low in the late 1920s. But when comparing 1927-29 to 1924-26, both the Lebergott numbers and the extrapolated Douglas numbers moved upward.66

At this point, we must remember that the effect of the reserve army of the unemployed is not simply a matter of the current year's unemployment. Record unemployment rates such as those seen in 1921 (see part C of the table) have a sustained impact on the morale, institutions, and relative bargaining power of workers, even as the unemployment rate fell temporarily) in the mid-to-late 1920s.67 The early 1920s recession represents a "sneak preview" of the depression and an indication that U.S. urban workers shared in the world pre-depression: crucially, the 1921 rate for manufacturing workers is close to the 1933 overall rate (25%). This point brings us full circle, to the beginning of this subsection: wages and the relative scarcity of labor-power do not simply depend on "market forces"; they also depend on working-class institutions and consciousness, which are part of the mechanism by which supply and demand affect wages. These institutions were in full retreat during the 1920s. The unemployment rates of the late 1920s may have been lower than during the early parts of that decade. But the institutional weakness of the working class also meant that unemployment was less necessary to protect and boost profits. That is, capitalists the fragility of the unions and other working-class organizations meant that capitalists could afford to eschew some of disciplining power of the reserve army of the unemployed and allow relatively high employment. [This is similar to the experience of the U.S. in the 1990s, where until 1998, low unemployment did not imply much of a rise of real wages. Unemployment seems to have become more powerful at cowing labor and preventing inflation.]

In theory, the U.S. economy might have been moving toward a regime of labor scarcity during the middle 1920s. But even in the low-unemployment year of 1926, the numbers cited below show no profit squeeze.68 Further, this move toward scarcity was not persistent enough to reverse the general urban labor abundance. With each of the three measures in table III, unemployment rose between 1926 and 1929. In each case, this rate fell from 1928 to 1929, but not [[p. 149]] enough to negate the upward trend. Key to the problem is that the government had not yet embraced military Keynesianism, which might have counteracted the almost automatic slowdown in capitalist accumulation that results when unemployment falls toward more rational levels.

Before turning to the effects of labor abundance on profits, we need an explanation of the "kink" in long-term labor-productivity growth that unites the period after 1919 with the 1950s and even the 1960s rather than with the pre-War era.69 Most of the literature trying to explain this change stresses generalities about the institutionalization of science and improvements in education [cf. Rosenberg, 1972: ch. 5; Abramowitz, 1993]; others emphasize the importance of the introduction of electric and internal combustion engines [cf. Oshima, 1984]. But without denying these points, one can point to some specific facets of this change related to the present argument.

The depressed profit rate may have encouraged capitalists to accelerate the introduction of new technologies, e.g., "pent up" technology denied to the civilian sector during the War and the steep post-War recession, which had the desired effect of speeding up labor productivity growth. Further, technical spin-offs from the War, many of which were developed partly in response to labor scarcity and thus were labor-saving, allowed this to occur.

Technology was simple compared to today's and the wasteful military-industrial complex was very undeveloped at the time, so unlike in recent years, these spin-offs were relatively inexpensive. For example, the War-time effort to build a fleet using mass production and prefabricated parts [Scheiber et al.: 322] probably accelerated the spread of such techniques in civilian production.

Third, during the 1920s, business followed two tracks to promote productivity using management techniques. On the one hand, Lazonick [1990] argues that for some important large firms, "welfare capitalist" efforts to limit employee turnover and to gain greater control over the work-place more than paid for their costs with productivity increases. On the other hand, for the vast majority of firms, such a "carrot" approach was unneeded or untried: they increasingly returned to the "stick"-oriented management systems of the pre-War era, involving speed-up and the like [Jacoby, 1985: 174-205; Gitelman, 1992]. This return to "stick" techniques seems a symptom of labor abundance and low worker bargaining power.

Either way, greater capitalist power sapped any resistance to the introduction of new technologies into existing factories. The success of the employers' initiatives in general was partly due to the workers' weakness, and in turn reinforced that weakness, as increased productivity boosted effective labor supply.

Who benefited from the abundance of labor? Not surprisingly, it was primarily the owners of companies in the leading sectors of the "roaring 20s," i.e., the industrial and corporate sectors. The flow of relatively cheap labor encouraged [[p. 156]] high and rising profit shares and rates in these sectors. Of course, this encouraged the leading sectors to continue to lead: since "nothing succeeds like success" (a slogan that captures the 1920s spirit), relative profit-rate rises attract accumulation. Moreover, the dominant political power of the most laissez-faire-oriented sections of the capitalist class meant that the government policies reinforced the trend toward inequality. The epitome of this process was the role of millionaire Andrew Mellon as Secretary of the Treasury, who pushed through tax cuts for the rich (including himself).

So even though the profit share (R/Y) fell for the economy as a whole between 1909 (or 1916 or 1925) and 1929 [Duménil et al., 1987], the share of profits rose for the leading sectors [Devine, 1988]. Whereas before the War, there is no large difference between the behavior of profit shares when one breaks down the data into sub-sectors, in the 1920s uneven development is manifest, favoring these leading sectors. Figure 2 shows this process: the income share of property (non-wage) income in the economy as a whole (the "unadj. share") and in the non-agricultural sector ("non-agr. share") trended downward during the whole period.[note on diag. 2] But these two sectors show substantial (but not complete) recovery during the 1920s, once self-employed or "entrepreneurial" income is deducted ("non-ent." and "non-agr., non-ent."). The behavior of unadjusted property income in the manufacturing sector was similar to these last two series, but once entrepreneurial income is subtracted (the solid triangles), the manufacturing sector had a profit boom that exceeded that of the War. The non-entrepreneurial manufacturing sector roughly tracks the behavior of the corporate sector as a whole. In general, the behavior of the different indices is similar before the War and then exhibits clear uneven development. Over the entire period, non-entrepreneurial manufacturing trends upward, while all the rest trend downward.

The increasing inequality between sectors according to profit share was reflected in estimates of the profit rate, as shown by Diagrams 3-A through 3-D. Diagram 3-A shows two estimates of profit rates for the national economy by Gérard Duménil and co-authors.[note on diag. 3-A] Between 1909 and 1916, the two series show similar trends, a decline before the War, followed by a profit-surge. After 1916, the two diverge. Duménil et al. [1987] estimates show a continuation of the downward trend after the War, while Duménil and Lévy's [1993] series shows a recovery, so that the 1900-29 trend was generally flat.70 Because of the current paper's emphasis on the leading sectors, the difference between these two total-economy series is unimportant here. The point is that for the economy as a whole profit rates fell up until the War and, including the impact of the post-War recession, until 1920 or so.

Diagram 3-B shows uneven development of profit rates between sectors, which turns out to be similar to the uneven development of profit shares in diagram 2. [note on diag. 3-B.] Most series show a flat trend or a very slight downward trend during the period.71 The exception is the corporate rate, which shows large profit [[p. 157]] booms during the War and the 1920s and a general upward trend. The manufacturing number, which includes entrepreneurial income as part of the numerator, did more moderately, rising in the 1920s but not recovering fully. Diagram 3-C shows that if the manufacturing profit rate is adjusted to exclude entrepreneurial (proprietor's) income, that rate showed a greater boom in the 1920s (and a upward trend over the entire period) than did the unadjusted rate.[note on diag. 3-C] Diagram 3-D shows three estimates of the corporate profit rate, including the adjusted manufacturing profit rate from 3-C. [note on diag. 3-D] The "merged series" represents several series based on limited samples of companies. Despite the clear limitations of such numbers, it turns out this profit rate follows much the same pattern as Taitel's series, representing the corporate sector as a whole, which starts only in 1909. If we can assume that the Taitel series would have followed the merged series path, we can conclude (in tandem with diagram 3-B) that corporate profit rates boomed in the 1920s, relative to the rest of the economy, and showed a long-term upward trend. But only the 1920s profit boom is central to this paper.

Turn next to the issue of capacity utilization (Y/Z), a possible determinant of profit-rate fluctuations (see equation 1) and an important datum for estimating the full-capacity profit rate. Existing statistics for Y/Z are not very good, but we can reach some tentative conclusions.72

Diagram 4-A shows the trend between 1900 and 1929 for two series that use the ratio between actual and trend production as a proxy for more direct measures of Y/Z. [note on diag. 4-A] Duménil and Lévy's numbers represent Y/Z for the economy as a whole, showing a similar but smoother trend than the merged manufacturing series. The trends for both series are downward, with manufacturing falling more steeply and more significantly.73

The long-term downward trend in Y/Z is not severe, while it is quite possible that an upturn in the 1920s might have negated this trend. For that decade, as shown in diagram 4-B, all available series except that of Bassie and Foster show an upward trend in Y/Z. [note on diag. 4-B.] The cyclical behavior of all series except Bassie-Foster also show remarkable similarities. This suggests first that the S-C-H and Hickman-Coen series, which cover only part of the period, would have shown behavior similar to the Duménil/Lévy and merged manufacturing series if they could be extended backwards to 1900. Second, we might conclude that the Bassie-Foster series (which has been used to indicate a long-term stagnation hypothesis) is somehow mismeasured.

However, rather than throw out such data, it is plausible to conclude that, in the long-term trend, either the Y/Z ratio either stayed roughly constant or fell. This, along with the profit-rate data, suggests that the "potential" or full-capacity profit rate (r*) of the leading sectors also rose.74 If we assume that the long-term trend of Y/Z was downward, potential profit rates would have been even higher than in charts 3-A through 3-D. If the trend is flat, then the trend of potential profit rates would have matched that of actual profit rates. [[p.158]]

Because of the potential profit rate's rise, conditions necessary to sustained stable growth in the leading sectors (and the unequally-distributed "prosperity" of the economy as a whole) became increasingly unlikely to be met. It meant that behind the rising profit rates of the late 1920s lurked a severe but latent instability. This instability can be seen as "over-investment relative to demand," in which investment grows too far relative to other sources of demand.

To understand the conditions necessary for a leading sector's stable growth, examine a simple equation in the tradition of the Harrod-Domar knife-edge.75 The leading sector is seen as part of a world economic system that is developing unevenly: while that sector has rising r* (as argued in the last sub-section) and growth rates, the lagging sectors are stagnant. The leading sectors can pull up the lagging, but any prosperity of the latter is totally dependent on that of the former. It is presumed that the structural problems of the world were so bad (as described in section II) that a world-wide process of self-sustained growth was impossible during this period. That is, it is assumed that the leading sector does not pull the lagging sectors beyond the notional threshold that allows a general sharing of prosperity on relatively equal terms among the core capitalist countries (perhaps as in the 1950s and 1960s), so that each country's prosperity can reinforce that of the others.

The equation is stated in terms of necessary conditions for the maintenance of steady growth at a "normal" rate of capacity utilization (v*, near full capacity), because persistent actual utilization (v) below normal discourages investment. This assumes that utilization rates affect investment directly (as with the stock-adjustment model of investment), in addition to their indirect effect of determining expected profit rates via actual profit rates (r).76 To keep capacity utilization constant at the normal rate, that is, to keep v equal to a constant v*,
 

where

As is standard with knife-edge models, condition (2) is met only by a happy accident in which the value of the five parameters just happen to be "right." But this simple model need not be seen as an absolute knife-edge in which either v = v* or the economy spins into stagnation (or hyperinflation). Instead, given the nature of investment's response to capacity use, this condition should be interpreted [[p.160]] probabilistically: the constant v* should be seen as allowing the greatest probability of stable growth, while the greater the deviation of v from v*, the greater the likelihood of recession. The greater and more persistent the deviation, that is, the more likely it is that the gap will affect expectations and investment decisions.

Assume that we have shown convincingly that for the leading sectors, r* rose during the late 1920s. This means that to maintain v = v*, either (1) the growth rate of the stock of means of production (g) must rise; (2) capitalist (luxury) consumption must rise, i.e., a must fall; or (3) external demand must rise relative to the stock means of production (e must rise). The "right" combination of these trends options would also allow v = v* to be maintained.

The first option is possible, given the rising full-capacity profit rate -- if capacity utilization indeed stays stable or rises. For this to be true, condition (2) must be met, on average. If so, the rise in r* boosts the actual profit rate (r), and -- through the indirect route of affecting expectations of profitability -- increases the growth of the leading sector's stock of means of production. This is "bootstrap" [or profit-led] growth, because it has elements of a self-fulfilling prophecy (or the "Tugan-Baranowsky Path" [Devine, 1983], a process of investing to serve investment demand). This seems to be part of the late-1920s growth process.

Over the long haul, the behavior of investment relative to consumption is ambiguous. Between the turn of the century (1897-1901) and the late 1920s (1927-31), the ratios of net and gross producer durable investment (I) to consumption (C) rose. But on the other hand, if housing investment (H) is treated as a form of consumption, the trends are generally downward.77 (Residential construction is interpreted as a form of consumer durable purchase.) But our concern here is not with the long period, but with the growth process of the 1920s. Diagram 5 shows the trends during the 1920s. [note on diag. 5] As indicated by the total-economy data, the investment boom occurred despite the relative decline in personal consumption, either including or excluding housing [cf. Devine, 1983]. These numbers (which do not capture the level of aggregation we are seeking) show a slowdown in investment relative to consumption in the late 1920s, and in 1929 itself, but the ratios are far from stagnant relative to the early 1920s. Now move to a more appropriate level of aggregation: in the last years of the decade, the ratio of manufacturing investment in plant and equipment rose relative to manufacturing production. Interestingly, the ratio of manufacturing investment to industrial production surged in the last quarter of 1929. Investment was over-shooting (as production fell) and would soon be cut back drastically.

This over-investment process is akin to a "bubble" in financial markets: the economy is growing because of faith that the economy will continue to grow, while profitability is maintained because high profit rates encourage investment. Similarly, the "bubble" of high profits and high growth rates is latently unstable and can easily be popped. Investment is more volatile than is consumption, so the average degree of instability of the sector's growth process increases with g. This [[p. 162]] means that the probability of an economic crisis rises. Further, this type of growth creates imbalances that can barriers to recovery: the production capacity created so abundantly during the boom is a factor discouraging the rebound of investment once the crisis occurs.78

The second option, rising consumption rates (falling a), is possible and are part of the story of the 1920s. This may be due to rising capitalist consumption, fitting with the model of Appendix B.2. Further, in a more complex story, we could drop the assumption that workers consume all of their income: condition (2) could be met despite rising r* if workers consumed a larger percentage of their income, by cutting down on saving or by accumulating debt [or by running down accumulated savings]. In fact, luxury and debt-financed consumer spending partially substituted for investment in promoting the economy's growth during the 1920s, as seen in Devine [1983] and DuBoff [1989: 88-9]. But that does not change the result. Because it is easier to delay, capitalist consumption of luxury goods is more unstable than is workers' consumption. For the second case, the economy would also become more unstable over time as interest payments weigh increasingly on workers' stagnant incomes.

For the third case to occur, the growth rate of leading-sector "net exports" must exceed the growth rate of its means of production (g). If this happens, it means that the sector becomes increasingly dependent on the health of the rest of the world. This was especially problematic during the late 1920s, given the stagnation of the world outside the U.S. leading sectors. It is possible that there were other leading sectors in the world besides the U.S. corporate sector, but these sectors suffered from the same problem as that sector. In a generally stagnant world economy, bootstrap growth becomes increasingly unstable.

In fact, for the U.S. economy as a whole, the growth of real exports had fallen or leveled off (depending on the deflator used) between 1927-28 and 1928-29, while the growth rate of real imports became positive in 1928-29. Thus the export/import ratio fell between 1928 and 1929 [Temin, 1976: 145]. Given the leading sectors' growing role in the economy as a whole (shown in diagram 1), this also made their growth more fragile.79

As a final point on the pre-Collapse trends in the U.S. economy, notice that this paper changes the classical Marxist attitude toward the "composition of capital" and K/Z. A rising composition of capital is usually seen as depressing the profit rate, ceteris paribus, discouraging accumulation. Not noticed, however, is that a falling capital-output ratio can also have a negative effect on accumulation. Though this fall promotes the production of profit, it can hurt the realization of profit, the ability to sell what can be produced: rising "capital productivity" means that the new factories and machinery created by investment have a larger output at full capacity utilization. This process (as one factor boosting actual and potential profit rates) intensifies the likelihood of excess capacity arising. [[p. 162]]

Diagram 6 shows the behavior of four capital-output ratios; because of the long time periods covered, these estimates of K/Y approximate K/Z. [note on diag. 6] Until about the end of the War, these ratios fit Marx's prediction, as does the series (not shown) presented by Duménil et al. [1987: 30, 37-38] which peaks in 1918-1921. A minor exception is the Duménil/Lévy series which peaks in 1911.80 One might think that an investment boom such as that in the 1920s would lead to a rising K/Z ratio, but all of these ratios reverse their movement. This may be due to the falling age, and thus rising quality, of the capital stock in the late 1920s [USDC, 1975: 261, series F-516, 519], the technological boom discussed above, and/or the abundance of labor, which encourages the use of labor-intensive technology. Further, the shift of aggregate demand toward investment could have also encouraged this aggregate fall, if the sectors producing means of production have lower capital-output ratios.

No matter the explanation of the reversal of the trend in K/Z, growth with a sharply rising full-capacity profit rate is difficult to sustain, especially when the rest of the world is stagnant, the stage was set for the Great Collapse. Once the conditions necessary to maintain v = v* are no longer met, the actual profit rate falls below the full-capacity profit rate. As a result, the expected profit rate and investment fall, hurting profits further. So the economy goes down a steep slope, which is discussed in greater detail in section III.F.  

The collapse of the U.S. economy might have been delayed if certain familiar shocks had not occurred. However, since the nature of capitalism as a system makes "shocks" very likely if not normal, it is hard to imagine a shaky growth process of the sort described in the sub-section C as persisting. As noted in section I, the Marxian vision does not deny as much as play down the role of depressionary triggers: the stock market Crash and contractionary fiscal, monetary, or tariff policy were not structural causes but mere sparks causing a greater conflagration. These types of events need not cause a Collapse except when the economic structure is initially unstable. Further, these events are to a large extent explained by the theory rather than being diabli ex machinae.

After a long period of downplaying its influence [cf. Temin, 1976: 69-83], some mainstream economists have recently given the Crash more attention as a cause of the Collapse:

 "[B]oth the initial recession in the United States in the summer of 1929 and the acceleration of the decline in late 1929 and 1930 are ultimately attributable to the stock market boom and bust of the late 1920s. The stock market boom is the prime explanation for why the Federal Reserve was pursuing tight monetary policy starting in 1928. The stock market crash is the prime source for the collapse of durable goods starting in November 1929" [Romer, 1993: 31].81

As argued above, tight monetary policies in 1928 should be seen as part of the policy bias of the time. On the crash itself, Romer sees the 1929 Crash as simply [[p.164]] an exogenous "bubble that burst," claiming to follow White [1990]. But White attributes the Crash to the adjustment of speculator expectations to news of the recession that had started earlier [1990: 78-81]. That is, the Crash was partly explained by the beginnings of the Collapse and cannot be seen as purely exogenous. This can be extended to the degree that uncertainty about the late 1920s growth process -- its latent instability -- spread to the stock market.

Further, as argued elsewhere [Devine, 1983], the Big Bull market itself could be explained partly by the uneven development of the 1920s: increasing income inequality spurred the Big Bull, by giving income to those most able and apt to speculate; at the same time, rising corporate profit rates inspired the price of equities to rise. Speculation (normal to capitalism) and a laissez-faire system of finance (more specific to this era) also encouraged this boom. The Bull was able to continue partly because it could attract funds from places as far away as Europe and Japan. [This flight from the financial instability outside the U.S. is nowadays termed the "safe haven" effect.]

Because of the latent instability of profit rates discussed above, and the way in which equity prices roughly reflect the profit rate, the Bull was teetering on the edge despite all appearances.82 This is an addition to the common story of a speculative bubble: when the real economy itself resembles a bubble, the question of whether or not price-equity ratios were justified by "fundamentals" is not very important to the tale. However, as with any stock-market process, the actual timing of the Crash cannot be explained by the fundamentals in the economy.

The changes described in sections III.B through D made spending more responsive to stock-market events. Romer [1990] concludes that stock-market uncertainty had an immediate negative effect on consumer durables purchases. But there are missing links in this story [cf. Devine, 1983]. First, the large share of consumer durables that were purchased by the speculating, i.e., richer, classes (the less than 2% of the U.S. population that owned significant amounts of stock) made private spending more prone to being affected by stock-market events. Second, the growing importance of durables purchases in aggregate demand -- partly or wholly reflecting the wealthy's rising share of income -- made such effects more important.83

This links up with the discussion of section III.D: growth that is promoted by greater luxury demand or working-class debt accumulation is also more fragile. Given the more general fragility of the growth process, a financial shock like the Crash could have major effects.

As discussed above, the government and Fed policies often seen as causes of the Collapse were not mere "policy mistakes" (as in Romer [1993: 34]). Rather, they were products of an era of Restoration of order and of pro-business politics of the most narrow-minded sort. That is, these policies -- such as the tax hikes (e.g., in June 1932) aimed at balancing the federal budget during the Collapse -- were poor only when seen through the omniscient lens of hindsight, and were hardly "mistakes." Avoidance of Keynesian expansionary fiscal policies was a product of conservative rule (reinforced by ignorance) and a relatively non-militaristic international environment. If a war occurred or was being planned, there is little doubt [[p. 165]] that balanced-budget strictures would have been forgotten, as with previous and later wars (or as in Germany in the 1930s). The same can be said if the working class and other dominated groups had been pushing for social programs despite increases in deficits (as in the middle-to-late 1930s or in Sweden in this era).

What about monetary policy, as emphasized in the long quote above and by Friedman and Schwartz [1965], who see a "Great Contraction" of the money supply resulting from the Fed's passivity in the face of waves of bank failures? Post-Keynesians and Marxians have argued that such policy can be rendered impotent (and thus unable to prevent or reverse a Collapse) because of the uncontrollability of the supply of or demand for credit and also the unresponsiveness of fixed investment to interest rates. Whether we accept these argument or not, it is hard to see how the Great Contraction could not have made matters worse. But these authors presume that the Fed actually cared about the health of the domestic economy. Epstein and Ferguson [1984] present an excellent empirical analysis of the Fed's policies during this period, which indicated that the Fed was concerned with external balance rather than domestic prosperity and was unwilling to go against the interest of banks (except the small and less influential banks catering to farmers).

First, the Fed waited a long time before any serious efforts to reflate because

Conventional doctrine among businessmen, bankers, and economists in the period held that occasional depressions (or deflations) were vital to the long-run health of a capitalist economy. Accordingly, the task of central banking was to stand back and allow nature's therapy to take its course. As one well-known voting member of the Fed's Board of Governors, Treasury Secretary Andrew Mellon, expressed it, the way out of a depression consisted of a sustained effort to 'liquidate labor, liquidate stocks, liquidate the farmers, liquidate real estate' [1984: 963].

As with the international deflationary consensus mentioned above, the inflationary experience of the U.S. during and immediately after the War, though incomparable to that in Germany, encouraged deflation. The post-War inflation was scarier than that during the War, as U.S. bankers saw inflation as normal during war-time. On the other hand, the success of the post-War recession in taming said inflation (effectively stabilizing the price level until 1929) encouraged the use of deflation in later years. The Restoration had been successful so far, so why change orientation?

Even though the deepening recession spurred calls for reflation, public works spending, the veterans' bonus, and so forth, concern with the gold standard restrained any Fed effort to reflate. This was especially true after the run on the dollar that followed Britain's leaving the gold standard. After the 1932 Glass-Steagall Act, on the other hand, the gold-standard constraint was loosened and the Fed moved to use open-market operations to stimulate the economy.

But this program was abandoned in the face of three effects [Epstein and Ferguson, 1984: 968-83]. The Collapse pushed banks to liquidate their loans and invest in short-term securities (especially T-bills), so that falling short rates directly and significantly hurt their earnings. The bankers then successfully pushed the Fed [[p. 166]] to abandon expansionary policies.

Second, a growing number of Federal Reserve branches resisted expansion since their gold reserves were threatened. Third, foreign depositors -- led by France -- started pulling their balances from U.S. banks, especially those in New York. In the end, the Fed was pushed to deflate.

 The process of learning the wrong lesson from past success was also applied to tariffs. In the past, domestic stimulus had arisen from tariffs: the most recent case was the tariff hike of 1921, which had been followed by the "roaring 20s" (which benefited the politically-relevant populations). The effect of the tariff making the economy less flexible (emphasized by Gordon and Wilcox [1981]) was balanced by the tariff's promotion of domestic demand, which lowers the need for flexibility, at least in the short run. Further, if the tariff had not limited price flexibility, the debt-deflation (see below) might have been more devastating. President Hoover and Congress, therefore, instituted the Smoot-Hawley tariff of 1930, which helped to convert creeping protectionism into a trade war. The destabilizing impact of protectionism was international.

To summarize, U.S. prosperity was fragile even before late 1929, due to the process of over-investment relative to demand and the international environment. Then the Crash, restrictive fiscal and monetary policy, and protectionism interacted to break the unstable prosperity and to accelerate the downward movement. This movement involved the famous "multiplier/accelerator" interaction, reinforced by wage-cut induced underconsumption, debt deflation, and international interactions.

The multiplier-accelerator interaction is familiar and a version of this process is implicit in the conditions for stability introduced in section III.D.84 On the second point, there exists some preliminary evidence for an underconsumption trap during the first years of the Collapse. Lebergott corrects some of the biases of the BLS data to find that for manufacturing production workers, hourly money "wages fell at least 31% from 1929 to 1932 for workers of given ability" [1990: 11]. With nonfarm output per worker-hour trending upward roughly 2 per cent per year (a conservative estimate) due to technical change [cf. USDC, 1975, p. 948: series W-3], this means that unit labor costs fell more than 37 per cent over this period. On the other hand, consumer prices fell only about 20 per cent during these years [USDC, 1975, p. 211: series E-135]. Due to obvious data problems, Lebergott was unable to correct for the effects of speed-up, which was quite common during these years. Such correction would make the wage and unit labor cost cuts effectively even larger. These falls relative to prices help explain the fact that falling "consumption accounted for a much larger fraction of the decline in real GNP in 1930 [and 1931-33] than in most previous or subsequent recessions" [Romer, 1993: 30, table 2; cf. Temin, 1976: 62-83].

Once the downturn occurs, domestic and international debts (and reparations) represent an imbalance discouraging recovery, as does the existing stock of luxury [[p. 167]] goods. Such debts, as mentioned above, may encourage debt-deflation: debt makes deflation destructive [Fisher, 1933]. Calomiris [1993] summarizes recent research indicating that such a debt-deflation occurred during the 1930s. Once deflation begins, outstanding debts and debt-service obligations become greater in real terms, encouraging yet more cutbacks in spending and world aggregate demand. Falling prices hurt even those sectors that had relatively low debt ratios during the previous prosperity, such as the U.S. corporate sector.

Borrower default and bankruptcy then hurts the creditors. Falling demand also hurts the value of real assets, causing a "balance-sheet crunch." All of this, plus the competitive wage-cutting discussed above, encourages further collapse and deflation.85

 [introduction] [section I] [section II] [top of this section]

[next section] [appendices] [references] [notes] [short version] [a new analysis of the future]