I don’t really know what caused the crash, but I think I have as good an explanation as anyone. For those who don’t know, I am taking a two month break from blogging in order to finish revising my manuscript. During that time I will post material from chapters 4, 5, and 6. Today I have the first 9 pages from chapter 4, and I will do 3 more posts from that chapter over the next week or two. Chapter 4 covers the first year of the Depression.
Believe me, I’d rather be blogging. Just yesterday I saw the following two stories juxtaposed:
1. Stocks fall and the dollar rises on Bernanke warnings
2. Obama pushes new jobs bill
Policy coordination just doesn’t get any more uncoordinated than that. But I held back.
The following excerpt does not include graphs, as I haven’t figured out how to post graphs from my Word files. But you’ll get the general idea. The next post will provide a narrative of how the stock market responded to policy news in 1928-29. The third post will look at the events of 1930, and the last one will provide my take on the first year of the Depression. In subsequent weeks we will look at chapters that cover 1931 and 1932; much more interesting years than 1929-30. Don’t worry if you don’t find my arguments totally convincing–I don’t either. Consider these ideas to be seeds that will bear fruit in later chapters. And I don’t think we’ll ever fully understand the stock crash, my goal is to partly understand it. (Footnotes are handled oddly, you’ll need to get used to it.)
Chapter 4: From the Wall Street Crash to the First Banking Panic
“Partly, no doubt, the stock market crash was a symptom of the underlying forces making for a severe contraction in economic activity. But partly also, its occurrence must have helped to deepen the contraction.” (Friedman and Schwartz, 1963, p. 306.)
In mid-1929 the macroeconomic performance of the U.S. economy seemed close to ideal. There was strong growth in output, stable prices, low unemployment, a federal budget surplus, a trade surplus, and a booming stock market. Then after September 1929, both prices and output began a precipitous decline that would continue for nearly 3 years. There is now widespread agreement that the 1929-32 contraction was caused by a decline in aggregate demand, which may have been triggered by tight money or by an autonomous drop in private spending. But no one has yet explained why demand fell so sharply in the year after the crash.
Over the next 9 chapters we will be looking at the response of financial markets, and especially the U.S. stock market, to economic policy-related news. An obvious place to begin this narrative is with famous stock market crash of 1929, which might have been linked to the subsequent Depression in one of two ways. Most historical accounts have assumed that the crash helped trigger the Depression. Alternatively, the stock market crash could have been caused by (expectations of) an oncoming Depression. This is the interpretation most compatible with the efficient markets hypothesis.
It will be useful to begin by briefly comparing the 1929 crash with a strikingly similar decline that occurred in 1987. In 1929, the Dow peaked in early September, fell at an increasing rate during late September and early October, and finally plunged 23 percent on October 28 and 29. The total eight-week decline was 39.7 percent. In 1987, the Dow peaked in late August, fell at an increasing rate during September and early October, and then plunged 22.6 percent on
 Of course it is natural to compare the U.S. economy of the late 1920s with the late 1990s, which featured many of the same characteristics. But even with that comparison the 1920s would probably come out ahead. In retrospect, it is obvious that in the late 1990s there was over-investment in IT and telecommunications, even assuming there had been no recession in 2001. Although over-investment may have occurred in the 1920s as well, it is not at all obvious that booming industries such as automobiles would have been overextended had the U.S. economy continued to grow at a healthy rate during the early 1930s.
 This was by far the largest single day decline in U.S. stock market history.
October 19th. The total eight-week decline was 36.1 percent. And both occurred after the U.S. economy had experienced a sustained period of economic growth under conservative Republican tax-cutting administrations. One difference is that further sharp declines occurred over the weeks following ‘Black Tuesday’ and thus the total market decline in 1929 was nearly 48 percent. The more important differences, however, relate to macroeconomic events that occurred after each crash.
The 1987 stock market crash was followed by three more years of strong economic growth and healthy corporate profits. Because of the lack of obvious ‘news’ surrounding the 1987 crash it was widely viewed as an example of investor irrationality, an interpretation that is incompatible with efficient markets theory. Popular historical accounts also tend to portray the 1929 crash as an episode in mass hysteria, which is odd given that the 1929 crash was followed by the most severe depression in U.S. history. The 1929 crash might just as well be viewed as a striking confirmation of the extraordinary sophistication of market expectations—as investors were able to perceive the onset of a depression, even as many so-called experts remained optimistic about the economy. Before considering evidence for this hypothesis we need to take a closer look at the relationship between the stock market and the business cycle.
Schwert (1990, p. 1237) showed that between 1889 and 1988 “future [industrial] production growth rates explain a large fraction of the variation in stock returns.” Dwyer and Robotti (2004, p. 11) observed that the “stock market does not necessarily decline before a recession, but the onset of a recession is invariably associated with a substantial decline in stock prices.” And McQueen and Roley (1993, p. 705) noted that “news of higher-than-expected real activity when the economy is already strong results in lower stock prices, whereas the same surprise in a weak economy is associated with higher stock prices,” a result of particular relevance to this study. In the following narrative I will show that during the Depression stock prices responded positively to news of policy initiatives that were expected to boost output, and vice versa.
It seems unlikely that the stock market crash and the Great Depression were entirely unrelated. But is it plausible that the 1929 crash could have been triggered by policy-related news? The crash might have merely been a reaction to signs that the economy was slowing in the autumn of 1929; industrial production had already peaked in late summer. Yet why would a relatively modest decline in production over a period of just a few weeks reduce equity values by 48 percent? On the other hand if bad policy was the cause, then what were those policies?
If the Depression did cause the stock market crash then the very scale of the crash suggests that any plausible explanation must involve a forward-looking mechanism whereby investors foresaw at least a part of the economic calamity to come. Did the stock market receive such bearish information in October 1929? It is unlikely that we will ever find an answer that is completely consistent with the efficient market hypothesis. But if the scale of the 1929 crash is destined to remain something of a mystery, we will at least find some tantalizing hints that can be developed much further in subsequent chapters.
This chapter focuses on two key questions: Did an increase in the world gold reserve ratio begin reducing aggregate demand in late 1929? And if so, did it also contribute to the 1929 stock market crash? Before applying the gold market approach to the onset of the Depression, however, it will be useful to first review some previous studies of the 1929 crash.
4.a Previous Explanations of the 1929 Stock Market Crash
The monetarist view of the Depression generally begins with the Fed’s move toward a tighter monetary policy in mid-1928, a policy switch that also shows up clearly in the U.S. gold reserve ratio, but not in the world gold reserve ratio. Despite the quotation that opens this chapter, in Friedman and Schwartz’s account “the underlying forces” behind the October 1929 stock market crash are never really explained. Instead, they basically treat the crash as an aggravating factor that depressed velocity. The problem is that the monetarists’ “long and variable lags” might explain why the Depression began more then a year after the Fed adopted a tight money policy, but it cannot account for the strong performance of U.S. equity markets during the intervening period. Although monetarists tend to believe in market efficiency, their policy narratives often overlook market responses to monetary policy actions, or even imply highly irrational behavior by investors. Nor are they alone in this regard.
In the Austrian view, inflationary monetary policies during the 1920s led to an unsustainable investment boom. Both the stock market crash and the ensuing Depression were a consequence of those policy errors. The assumption of market inefficiency is even more central to the Austrian view; as rational investors presumably would not have bid prices up to such lofty levels in mid-1929 if they understood that Fed policy would inevitably produce a bust. And whereas monetarists can at least point to the fact that monetary tightening has often been followed by economic downturns, the Austrian view is hard to reconcile with post-war U.S. monetary policy. Several post-war decades saw far more inflationary booms than the 1920s, and stock market booms of a nearly comparable magnitude. Yet none were followed by major depressions. Thus modern Austrians often distinguish between the initial shock, and what’s called a “secondary depression.”
 Friedman and Schwartz focus more on the forces that worsened the slump after the onset of the banking panics, but Schwartz (1981) clearly attributes the initial downturn to a tight money policy adopted in 1928.
See Rothbard (1963) and Palyi (1972.)
As with the monetarists, Keynesians viewed the 1929 crash as both an exogenous event and a causal factor in the ensuing contraction. Indeed, because the initial stages of the Depression are difficult to explain within an IS-LM framework, the importance of the crash is typically even greater than in monetarist accounts of the Depression. And Keynesians are even more likely to view the spectacular 1928-29 bull market as a “bubble”, the bursting of which depressed aggregate demand. For instance, Romer (1990) argued that 1929 stock market crash sharply reduced consumer confidence, and that this was a major factor depressing aggregate demand. But the quite similar stock market crash in 1987 seemed to have no impact at all on economic growth, suggesting that the impact of stock prices on real output is almost certainly very small. If Friedman and Schwartz had written their Monetary History 25 years later, they probably wouldn’t have even mentioned the crash as a causal factor.
Romer suggested that the 1929 crash may have had a greater impact on consumer confidence than the 1987 crash because the earlier crash was followed by a higher level of stock market instability, but this hypothesis has two serious flaws. First, the fact that the stock market was modestly more unstable in 1930 than 1988 could explain a small difference in economic growth, but it can hardly explain the difference between a severe recession in 1930, and an economic boom in 1988. And even worse, the extra market volatility didn’t begin until mid-April 1930, by which time the economy was already deep in recession. Indeed, the Dow actually performed much better in the six months after the 1929 crash than during the six months following the 1987 crash. Of course the striking lack of impact from the 1987 crash might be an anomaly. But several recent studies looking at both time series and cross sectional data, found little or no evidence of a significant wealth effect from changes in stock prices.
Temin (1976) also argued that the first year of the Depression could be explained by an autonomous drop in consumption, but was uncertain as to what caused consumption to fall.
 It might be argued that the Fed eased policy much more aggressively after the 1987 crash, and that otherwise that crash would also have been followed by a depression. Even if one accepts this argument (and recall that the Fed also cut rates after the 1929 crash) it seems more of an argument for the importance of monetary policy as a determinant of business cycles, rather than stock prices.
Case, Quigley and Shiller (2005, p. 26) find “at best weak evidence of a stock market wealth effect.” Dwyer and Robotti (2004) also find little evidence of a “wealth effect”. And note that measured correlations probably overstate the impact of stocks on consumption, as both variables will, at least to some extent, respond to (unobservable) changes in expectations of future economic conditions.
In fairness, even most Keynesians don’t see the crash as “the cause” of the Great Depression, but rather as simply one of many unfortunate events that contributed to the disaster.
What evidence do we have that stocks were overpriced in the late 1920s? Some finance models suggest that stocks were grossly undervalued throughout most of the 20th century. If the investment community expected economic growth to continue right on into the 1930s, then would investor expectations really have been so irrational? Perhaps, economists are split on this issue. But Field (2003) showed that the 1930s were “the most technologically progressive decade of the twentieth century”, so there really were a lot of new developments for investors to get excited about. Before we throw up our hands and accept the “bubble” explanation, we should first see whether there is an alternative explanation that allows for sensible investors to have been highly optimistic in September 1929 and much more pessimistic in November 1929.
4.b Gold Market Indicators at the Onset of the Depression
It is not surprising that many observers would blame the stock market crash for the sharp decline in aggregate demand after October 1929; there were no other obvious culprits. For instance, there was no dramatic break in the growth rate of U.S. monetary aggregates in the year following the crash. But as we saw in the previous chapter, domestic monetary aggregates may not be a reliable indicator of monetary policy under an international gold standard. Rather it is the world gold reserve ratio that is the theoretically appropriate indicator of policy.
To see whether the gold market approach can help us understand the collapse in aggregate demand after September 1929, it will be helpful to take a closer look at the data in table 3.4 (at bottom of post.) If we focus on the gold reserve ratio, there are two statistics that really stand out. The first was the French gold reserve ratio, which between December 1926 and December 1932 increased by enough to reduce the entire world price level by 17.3 percent! And as noted earlier, there was a 9.6 percent increase in the world gold reserve ratio during the 12 months after October 1929 stock crash. If the gold reserve ratio is a useful policy indicator, then we ought to be able to see links between changes in the ratio and policy actions by major central banks.
Temin (1989, p. 44) argues that the crash was not a major independent shock, and also cites the 1987 crash.
 During the 20th century, long term rates of return on U.S. stocks have been much higher than on risk-free Treasury securities. The undervaluation hypothesis is based on the view that this differential is much too large to be explained by any plausible risk premium on equity investments
During the Great Depression Irving Fisher was widely ridiculed for having called the stock market undervalued in October 1929. And studies by DeLong and Shleifer (1991) and Rappoport and White (1993) provide indirect evidence of bubble-like behavior. But McGrattan and Prescott (2004) suggest that “Irving Fisher was right” about stocks being undervalued in 1929 if one accounts for the value of intangible corporate assets.
 Although the French franc was fixed to gold in December 1926, France did not officially return to the gold standard until June 1928.
The data certainly supports studies by Johnson and Eichengreen that showed the deflationary impact of French policy. Eichengreen (1986) cites France’s Monetary Law, which prohibited purchases of foreign exchange, as an important constraint on the Bank of France. This law mandated 100 percent gold backing for any increase in the currency stock. Contemporaneous observers noted that almost all of the increased circulation was occurring in the larger denomination notes, and attributed this increase to widespread currency hoarding by French peasants. But the French gold reserve ratio increased at a fairly steady rate, at least until uncertainty surrounding the devaluation of the British pound in September 1931 led France to sharply accelerate the rate at which it was replacing foreign exchange reserves with gold. Thus although French policy may well have contributed to the worldwide deflation that occurred between 1926 and 1932, it doesn’t tell us anything about why the U.S. price level was stable in the late 1920s, and then suddenly began a sharp decline after October 1929.
Table 4.1 is identical to 3.1, except that the percentage changes for each sub-period have been annualized to make it easier to identify policy changes. If we look at the worldgold reserve ratio, we can see that what had been a mildly contractionary policy between 1926 and 1929 (with a world gold reserve ratio rising at roughly 2.5% per annum) turn sharply contractionary after October 1929.
This policy shift is particularly interesting when contrasted with Friedman and Schwartz’s data on U.S. monetary aggregates. They had difficulty identifying any significant policy shift in the fall of 1929, but found strong evidence of a highly contractionary U.S. policy after banking panics began in late 1930. If the gold market approach is to provide a superior explanation for the onset of the Depression, and by implication, the stock market crash, it will almost certainly involve the sharp increase in the world gold reserve ratio during 1929-30. Any gold-based explanation must be able to explain why this policy shock occurred in late 1929.
By 1932 France held almost twice as large a proportion of the world’s monetary gold stock as during 1914. This led Nurske (1944), Hawtrey (1948), and Cassel (1936) to assert that French policies resulted in a “maldistribution” of gold stocks which contributed to the 1929-32 deflation.
 See the Commercial and Financial Chronicle (May 3, 1930, p. 3089).
TABLE 4.1. The impact of changes in the world gold-reserve ratio, real demand for currency, real demand for gold, and monetary gold stock, on the world price level, 1926-1932.
Dec 1926 Jun 1928 Oct 1929 Oct 1930 Aug 1931 Dec 1926
To to to to to to
Jun 1928 Oct 1929 Oct 1930 Aug 1931 Dec 1932 Dec 1932
D(ln 1/r) -2.65 -2.38 -9.62 +1.86 -4.35 -3.64
D(ln 1/md) -2.70 -2.21 -4.97 -19.42 -10.17 -6.80
D(ln 1/g) -5.35 -4.59 -14.59 -17.56 -14.52 -10.44
D(ln G) +3.88 +4.06 +5.25 +4.72 +3.88 +4.27
D(ln P) -1.47 -0.53 -9.34 -12.84 -10.64 -6.18
D(ln 1/r) = change in the log of (the inverse of) the gold reserve ratio
D(ln 1/md) = change in the log of (the inverse of) real money demand
D(ln g) = change in the log of (the inverse of) the real demand for monetary gold
D(ln G) = change in the log of the world monetary gold stock
D(ln P) = change in the log of the world price level
(The percentage changes now represent annualized first differences of logs.)
Notes: Outside the U.S., the currency stock was used as a proxy for the monetary base. The change in the real demand for monetary gold is equal to the sum of the changes in the gold reserve ratio and the real demand for currency. The change in P reflects changes in the monetary gold stock and the (inverse of the) real demand for monetary gold. The changes are not seasonally adjusted.
Although French policy was undoubtedly important throughout the entire period of 1926-32, if one looks at smaller time periods then shifts in U.S. and British policies take on greater significance. Table 3.4 shows that between December 1926 and June 1928 expansionary monetary policy in the U.S. helped offset the contractionary effects of French policy. Note that the terms ‘expansionary’ and ‘contractionary’ refer to the gold reserve ratio, not the (endogenous) money supply. The U.S. currency stock actually declined slightly over this period. The expansionary policy reflected the (activist) views of New York Fed Governor Benjamin Strong, who was supporting his friend Norman Montegu Montagu Norman at the Bank of England during this period. Because England had returned to gold at the pre-war parity, which overvalued its currency, whereas France had sharply depreciated its currency, England had the more difficult time in readjusting to the gold standard. Strong’s decision to adopt an expansionary policy during 1927 allowed Britain to attract gold without being forced to adopt deflationary policies.
Governor Strong’s policy was criticized at the time for being highly “inflationary.” This characterization may seem puzzling given that the U.S. currency stock, and price level, actually declined during this period. Yet the decrease in the gold reserve ratio confirms the expansionary nature of the policy. By mid-1928 the U.S. had exported almost $500 million in gold and there was a growing perception of excessive speculation in the stock market. During mid-1928 the Fed switched to a contractionary policy aimed at restraining Wall Street’s “irrational exuberance”.
This shift in U.S. policy was partially offset by a move towards a more expansionary policy in England. After the death of Governor Strong, the Bank of England was the only major central bank that saw the need for international cooperation to maintain price stability. However England’s resources were severely limited. Between July 1928 and October 1929, their monetary gold stock fell by nearly a fourth. By the summer of 1929 England’s gold stock had fallen below the £150 million level recommended in the Cunliffe Committee report and thus, on September 26th the Bank of England was forced to raise its discount rate. By late October the pound had risen to the gold import point and England’s gold reserve ratio began to increase rapidly.
 Of course the concept of causation is difficult to define when the actions of several countries are independently affecting the world price level. For instance, if policies in England, France, and the U.S. each tend to reduce the world price level by 2 percent, and if the monetary gold stock rises by 4 percent, then the price level will decrease 2 percent. In this case each of the three countries independently, or all three jointly, could be said to have ’caused’ the 2 percent deflation.
The fact that the French franc was undervalued does not explain the huge inflows of gold into France. Hawtrey noted that France’s creditor position and favorable balance of payments could only affect her demand for gold by affecting the demand for currency notes. Instead, the undervaluation of the franc explains why in returning to the gold standard France did not have to deflate its price level by as much as England.
In retrospect, the period from October 1929 to October 1930 was decisive. U.S. monetary policy became even more restrictive than during the previous 16 months. Well into 1930 many Fed officials continued to emphasize the need to liquidate the excessive debts accumulated during the previous boom. Proposals that monetary policy be eased were rejected on the grounds that such a policy would simply repeat the mistakes that resulted in the crash. At the same time France continued to increase the gold backing of its currency. The simultaneous adoption of tight money policies in the U.S., France and Britain made worldwide deflation almost inevitable. The world gold reserve ratio, which had increased at an annual rate of 2.53 percent from December 1926 to October 1929, soared by 9.62 over the next 12 months. Despite a slight acceleration in the growth rate of the monetary gold stock, the price level fell almost as sharply.
The sudden upward surge in the world gold reserve ratio after October 1929 calls into to question the Keynesian view that monetary policy is unable to explain the first year of the Depression. Even previous researchers who focused on the contractionary role of the gold standard, such as Temin and Eichengreen, were not able to find the sort of dramatic shift in world monetary policy that could have plausibly caused both stocks and output to fall sharply in the fall of 1929. The next step is to see what specific policy actions might have contributed to the increase in central bank gold reserve ratios, and how markets reacted to those policy shifts. As we do so, we need to continually think about investor perceptions of monetary policy from a “what did they know, and when did they know it” perspective.
 See Nelson (1991.)
TABLE 3.4. The impact of changes in the gold-reserve ratio, the real demand for currency, the real demand for gold, and the total monetary gold stock, on the world price level, 1926-1932.
Time Dec 1926 Jun 1928 Oct 1929 Oct 1930 Aug 1931 Dec 1926
Period to to to to to to
Jun 1928 Oct 1929 Oct 1930 Aug 1931 Dec 1932 Dec 1932
U.S. 1. r +1.24 -2.59 -4.69 +1.55 +9.46 +5.15
2. m +1.55 -0.58 -0.46 -9.85 -9.66 -19.01
3. g +2.79 -3.17 -5.15 -8.30 -0.20 -13.86
4. r -1.07 +1.52 -1.38 +1.05 -1.37* -1.19
5. m -0.13 +0.30 -0.55 -0.59 +1.19 +0.29
6. g -1.20 +1.82 -1.93 +0.46 -0.18 -0.90
7. r -3.23 -2.73 -2.49 -1.80 -6.53 -17.27
8. m -1.47 -1.68 -3.08 -3.02 -5.17 -13.99
9. g -4.70 -4.41 -5.57 -4.82 -11.70 -31.26
10. r -0.92 +0.62 -1.06 +0.75 -7.36 -8.55
11. m -4.01 -0.98 -0.89 -2.73 +0.08 -8.09
12. g -4.93 -0.36 -1.95 -1.98 -7.28 -16.64
13. r -3.98 -3.18 -9.62 +1.55 -5.80 -21.86
14. m -4.05 -2.94 -4.97 -16.18 -13.56 -40.80
15. g -8.03 -6.12 -14.59 -14.63 -19.36 -62.66
16. G +5.82 +5.42 +5.25 +3.93 +5.18 +25.61
17. P -2.21 -0.70 -9.34 -10.70 -14.18 -37.06
r = direct impact of changes in gold-reserve ratio on price level
m = direct impact of changes in real currency demand on price level
g = direct impact of changes in real demand for gold on price level
G = change in the total world monetary gold stock
P = change in the world price level
Note: All numbers represent the first difference in the log of P (times 100); the numbers are not annualized. The sum of the impact of changes in r and m, should equal the impact of changes in g. For the world as a whole, the change in P reflects changes in G and g. For more information on rest of world (ROW), see appendix.
*The decomposition of the British demand for gold has little significance after Britain left the gold standard in September 1931.