Quarterly Journal of Austrian Economics
The Gold Standard Did Not Cause the Great Depression
The Midas Paradox is an impressive piece of scholarship, representing the magnum opus of economist Scott Sumner. What makes the book so unique is Sumner’s use of real-time financial data and press accounts in order to explain not just broad issues—such as, “What caused the Great Depression?”—but to offer commentary on the precise zigs-and-zags of the economy during the 1930s.
Sumner rejects the standard Friedmanite monetarist “long and variable lags” approach, and argues that financial markets respond virtually instantly to new information, including announcements and events that would change expectations about the future path of monetary policy. Both because of his methodological innovations and his painstaking research, Sumner’s book is an invaluable resource to economists and historians interested in the Great Depression and the operation of the classical gold standard.
Although I admire much of the book, I must reject its central thesis. Indeed, the very title The Midas Paradox is an allusion to the disaster that comes from an obsession with gold. Sumner agrees with standard Austrian critiques of the New Deal and its crippling effects on labor markets, but he also thinks a large portion of the blame for the Great Depression lies with the unfortunate fact that policymakers’ hands (and currencies) were tied to gold. Even though economists back in the 1930s thought that central banks were “pushing on a string” with their low interest rate policies, Sumner thinks it is now well established that it was unwittingly tight money that made this depression “Great.”
Furthermore, Sumner draws lessons for today, believing that economists are wrong to focus on low nominal interest rates and even the huge expansions in monetary bases that the world’s major central banks have delivered since the 2008 crash. Instead, with his “Market Monetarist” framework, Sumner believes that central banks have foisted enormously tight monetary policy on the world, and that this largely explains the horrible crash and then sluggish recoveries of Western nations in the last decade.
In Sumner’s view, only by adopting a more useful criterion for assessing monetary policy can economists explain past crises and help policymakers avoid future ones. As Sumner concludes his introductory chapter: “The events of the past five years should make us all a bit more forgiving of those interwar policy experts who failed to correctly diagnose the problem in real time. When aggregate demand collapses, it looks to almost everyone as if the symptoms of the fall in aggregate demand are the causes. That was true in the 1930s and it is equally true today” (p. 32).
Although I could spend the rest of this review noting the areas on which I agree with Sumner, the best contribution I can make is to point out why I think his thesis ultimately fails. To that end, I will first show that the single most important relationship he charts in the book—and it is Sumner himself who christens it as such—is just as consistent with the Rothbardian (1963) explanation of the Depression as it is with a Market Monetarist one. Then I will show that Sumner’s emphasis on gold—which is the reason for the book’s title, after all—is misplaced; it cannot fulfill the criterion that Sumner himself says it must.
I will conclude that Sumner’s book, excellent though it is in many respects, fails in its purpose. Austrians who subscribe to the Rothbardian explanation (which in turn was an elaboration of the Misesian theory of the business cycle) may collect some interesting nuances and a wealth of data from Sumner’s book, but they have no reason to abandon their basic framework.
Evidence that Fits Both Frameworks: The Connection between Real Wages and Output
In his introductory chapter Sumner declares, “If I were asked to give a talk on the Great Depression and allowed just one slide, it would undoubtedly be Figure 1.2” (p. 20). We have reproduced that crucial chart below.
Figure 1.2: The Relationship between Detrended Industrial Production and Detrended (Inverted) Real Wages, 1929–1939, Monthly
In Sumner’s figure, the gray line shows the logarithm of industrial production, meaning that straight lines indicate steady percentage rates of growth (or shrinkage). The dark black line is the logarithm of the inverse of the real (i.e. price-level-adjusted) wage rate.
The figure shows quite clearly that during the 1930s, as real wages increased, industrial production fell. On the other hand, increases in industrial production went hand-in-hand with declines in real wages.
As it happens, I am perfectly happy with Sumner’s graph. In fact, I will go further and enthusiastically endorse just about all of Sumner’s interpretation of it as well:
[A] sharp fall in output could be caused by either a rise in nominal wages or a fall in the price level. It so happens that both factors played an important role in the Great Depression….
During the 1930s, the biggest supply shocks were New Deal programs aimed at artificially raising nominal wages. There were five big wage shocks, each of which tended to abort otherwise promising recoveries in industrial production. These wage shocks thus tended to make real wages more countercyclical—higher wages led to lower output.
But what about the demand shocks, which were the major cause of the Great Contraction? Recall that the real wage is the nominal wage divided by the price level…. Wholesale prices fell sharply during the 1929–1933 and 1937–38 contractions and rose sharply after the dollar was devalued in April 1933. Because nominal wages tend to be sticky, or slow to adjust, sudden changes in the WPI tend to show up inversely as changes in the real wage rate…. If prices fall much faster than wages, then profits decline and companies lay off workers. Real wages actually rose sharply during the early 1930s for those lucky enough to maintain full-time jobs. (Sumner, pp. 20–22, emphasis added.)
Perhaps surprisingly, in the above quotation, Sumner has provided the same basic explanation of the high (and persistent) unemployment rate that I myself gave, in my decidedly Rothbardian treatment in Murphy (2009). Sumner and I agree that during the 1930s, unemployment shot up whenever real wages were increasing and (perversely) made labor more expensive relative to other commodities.
However, where Sumner and I disagree concerns the blame for this state of affairs. If the general price level falls, while nominal wage rates do not fall nearly as much, then Sumner ultimately blames the monetary authorities for letting the purchasing power of money increase so rapidly. In contrast, I blame the other interventions of the federal government (in conjunction with labor unions) for making wages so much “stickier” than they had been in previous depressions.
In particular, we can compare the behavior of nominal wages and prices of the early 1930s with the experience from the 1920–1921 depression. Here we rely on the statistics and analysis from Gallaway and Vedder (1987). First we reproduce one of their tables:
Table 4: Rate and Indexes of Consumer Prices, Money Wages, Productivity, and Productivity-Adjusted Real Wages
As the final column from the table shows, real wages for hourly workers—especially if we further factor in productivity—grew substantially over the years of the Great Contraction, reaching almost 20 percent higher by 1933 (when the unemployment rate was almost 25 percent). For another amazing fact, note that nominal (money) wage rates for hourly workers in 1931 were only 5.6 percent lower than they had been in 1929, even though consumer prices by that point had fallen 11.4 percent. During this year, unemployment was already at a devastating 15.9 percent.
Even the table above does not shed light on the policies that might have contributed to the problem. After all, Sumner could take these data from Gallaway and Vedder in stride, showing the disastrous consequences of the Fed’s (allegedly) tight monetary stance in the early 1930s amidst “sticky nominal wages.”
Yet here is where the comparison with the 1920–1921 episode is decisive. After producing the above table, Gallaway and Vedder explain:
The issue is whether the Hoover recipe delayed the onset of money wage adjustments sufficiently to exacerbate the disequlibrium and increase the severity of the Great Depression. The evidence is persuasive that this is the case…. [A] monthly wage index compiled by the Federal Reserve Bank of New York (reported by Lionel Robbins) shows almost no movement in money wage rates from the fourth quarter of 1929 through the second quarter of 1930.
Contrast this pattern with that of the 1920–21 downturn. In both cycles, industrial production peaked at midsummer before the onset of the decline. In both cycles, the decline was precipitous, 27.5 percent from July 1920 to July 1921 and 21.3 percent from June 1929 to July 1930. However, as noted earlier, in the 1920-21 case, money wage rates fell by 13 percent, setting the stage for the sharp recovery that began in August 1921. One of the factors cited by Benjamin Anderson in explaining this recovery is “a drastic reduction in the costs of production.” How these costs were reduced is clear—money wage rates were cut, something that did not occur in the early days of the Great Depression. For example, according to data compiled by the National Industrial Conference Board, hourly wage rates for unskilled male labor fell more between 1920 and 1921 than they declined throughout the Great Depression.
The clear implication seems to be that the money wage rate adjustment process was distinctly different during the Great Depression compared to the 1920–21 decline in business activity. Apparently, Herbert Hoover’s goal of maintaining levels of money wage rates was achieved, at least temporarily. (Gallaway and Vedder, 1987, p. 46, emphasis added, endnotes removed.)
Much more recently, Lee Ohanian (2009) develops a formal neoclassical model and concludes that Herbert Hoover’s policies—which asked large firms to maintain nominal wage rates in exchange for keeping out unions—are ultimately to blame for the Great Depression. He writes in his abstract: “The theory also can reconcile why deflation/low nominal spending apparently had such large real effects during the 1930s, but not during other periods of significant deflation.”
In summary, regarding the “one slide” that Sumner would use if he had to choose just one, he and I are in agreement: The key to understanding the massive unemployment of the 1930s is real wage rates. Sumner and I agree that during an economic downturn, the last thing in the world we want is for labor to become artificially more expensive as prices fall faster than wage rates.
Yet rather than ask (ask Sumner does) why policymakers at the Federal Reserve allowed such a deadly fall in prices, instead I would ask why policymakers in the federal government hindered the fall in (nominal) wages that had been the norm in previous depressions (or “panics”).
Sumner's Misplaced Emphasis on Gold
In the previous section, I argued that the Rothbardian interpretation of the Great Depression could easily incorporate the single most important graphical relationship of Sumner’s book. Namely, a Rothbardian could agree that the immediate driver of unemployment was the real wage rate, but the Rothbardian would lay the blame on government measures that interfered with nominal wage adjustments, rather than with deflationary monetary policy.
In this section, I question Sumner’s emphasis on money—and in particular, the operation of the gold standard—as a key component of the Great Depression. Here again we will reproduce a key chart from Sumner’s book, namely Figure 2.1 (p. 44), which plots the inverse of the “gold ratio” against industrial production:
Figure 2.1: Industrial Production and 12-Month Change in C/G Ratio
To understand the significance of this figure, we first must explain the “inverted gold ratio.” Sumner had earlier (p. 28) defined the gold reserve ratio as “the ratio of the monetary gold stock and the currency stock.” Now under the rules of the classical gold standard, “countries were supposed to adjust their currency stock in proportion to their changes in their monetary gold stock,” and thus if a country did not do so, then such “[v]ariations in the gold reserve ratio can be seen as an indicator of discretionary monetary policy” (p. 29).
Returning to the figure above, we now see how it apparently endorses the Sumnerian framework. If the currency/gold ratio (the dark black line) falls, it means that the outstanding stock of currency has fallen relative to the amount of gold held for monetary purposes. It is discretionary monetary policy tightening, in the context of the classical gold standard. And since the dark black line goes hand-in-hand with industrial production (the gray line), Sumner believes that this chart is consistent with his central thesis.
However, even at this stage, there are problems. First, note that from January 1929 up until the fateful month of October 1929, the 12-month change in the currency/gold ratio is (slightly) negative. Even so, industrial output rises through the summer. Moreover, the particular zigs and zags do not coincide with each other; there is a relative tightening (i.e. falling dark black line) from April through June, while industrial production rises during this stretch. Furthermore, there is a spike in the black line going into October 1929, which (to repeat) represents a relative loosening of monetary policy in Sumner’s framework.
To be sure, eventually both lines collapse, but it is hardly clear that the movements in the black line are causing reactions in the gray line. Indeed, consider that as of January 1930, the height of the black line has returned to the same position it held back in April 1929. That means that the (modest) 12-month decline in the inverted gold ratio by January 1930 was no larger than that same change had been in April 1929. And yet, this monetary tightening coincided with growing industrial output back in April, while by January industrial production was in free-fall.
Now, when it comes to explaining the stock market crash of October 1929, what really matters is not the mechanical policy of that moment but rather the expectations of investors. Perhaps the Federal Reserve signaled in some way the sharp tightening of monetary policy that would eventually come, and investors realized how much things had changed as fall 1929 unfolded.
As a staunch proponent of the Efficient Market Hypothesis (EMH), this is indeed the approach Sumner adopts. Space constraints do not allow me to summarize his case, but I think it is fair to say that he presents no smoking guns. In fact, Sumner himself implicitly admits that he has failed in the task he set for himself, when he (no doubt subconsciously) moves the goalposts.
Specifically, on page 40 Sumner tells us his strategy (consistent with the EMH):
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. (Sumner, p. 40, emphasis added.)
To reiterate, for Sumner’s book to “work,” he must now show us what tangible actions (which could have been in the form of remarks made to the press) the Federal Reserve made in a two-month window from September to October 1929, which involved the handling of the gold standard and which made both the stock market valuations of early September and late October 1929 “rational.” Were there any such actions that would have altered expectations in such a drastic way?
I submit that Sumner gives us nothing that fits the bill. He himself seems to acknowledge this when, twenty-one (unconvincing) pages later, Sumner writes:
At the beginning of this chapter, I suggested that in order to understand the October  crash, one needed to explain why it would have been sensible for investors to be highly optimistic in September 1929, and somewhat pessimistic in November 1929. Is there an explanation for such a dramatic change in sentiment? (Sumner, pp. 60–61, emphasis added.)
Note the subtle movement of the goalposts (again, I believe innocent enough); on page 40 he had sought something that would make investors “much more pessimistic” two months later, while on page 61 he has lowered the bar to “somewhat pessimistic.” (Would a mere change to “somewhat pessimistic” explain back-to-back drops of almost 13 percent and then 12 percent, which is what happened in the market on October 28 and 29?) Sumner knows he doesn’t have it. Indeed, later on this page Sumner writes, “This makes it almost impossible to establish a clear link between monetary policy and the 1929 crash” (p. 61).
Now in fairness, Sumner might respond that his book does not need to explain how monetary tightening—due to the constraints of the gold standard—led to the 1929 stock market crash. This is because one of the ways Sumner departs from conventional analyses is that he thinks market crashes do not necessarily coincide with “real” downturns; his best counterexample is the 1987 market crash, which was bigger than the 1929 one and obviously didn’t spawn a decade-long depression.
Even so, it sure seems as if the 1929 stock market crash had an awful lot to do with the onset of the Great Depression. Just look again at the final chart above, taken from Sumner: the big drop in industrial production clearly began with the market crash. The fact that Sumner admits his framework can’t really explain this sharp turnaround is (in my opinion) key evidence that his focus on gold—and denial of the existence of asset bubbles—is fundamentally mistaken.
In truth, no economic historian can explain the precise timing of every movement in the financial markets and broader economy, for the simple reason that humans have free will. Even so, using the very criteria Sumner himself embraces, we can conclude that his book—though superb in several dimensions—does not achieve its stated purpose.
Putting aside the detailed statistics, I will end this review with a simple question: How can it be that the classical gold standard is largely responsible for the Great Depression, when the classical gold standard was operating during several previous financial panics and depressions (small “d”)? To blame the Great Depression on the gold standard is akin to blaming a particular plane crash on gravity.
In contrast, the Rothbardian analysis at least has a shot at being satisfactory. After all, Herbert Hoover in his memoirs tried to defend his legacy by assuring his readers (truthfully) that his administration had taken unprecedented measures in battling the Depression, meddling in the economy in ways that no president during peacetime had done before. That’s the place to start, when we ponder why Herbert Hoover suffered from a worse downturn than any president before.
Gallaway, Lowell and Richard K. Vedder. 1987. “Wages, Prices, and Employment: von Mises and the Progressives,” Review of Austrian Economics 1: 33–80.
Murphy, Robert P. 2009. The Politically Incorrect Guide to the Great Depression and the New Deal. Washington, D.C.: Regnery.
Ohanian, Lee E. 2009. “What—or Who—Started the Great Depression?” Working Paper No. 15258 (August), National Bureau of Economic Research.
Rothbard, Murray N. 1963. America’s Great Depression. Princeton: D. Van Nostrand.
Cite This Article
Murphy, Robert P., "Review of The Midas Paradox: Financial Markets, Government Policy Shocks, and the Great Depression by Scott Sumner," Quarterly Journal of Austrian Economics 19, no. 1 (Spring 2016): 101–111