The 1920-1921 Depression and Recovery

Let’s recall the story. Some austrian economists (Woods, 2009; Powell, 2009; Murphy, 2009) claimed that Warren Harding cut the taxes, and by this has promoted the economic recovery. But Kuehn (2010) challenges this view. Kuehn (2012) believes that it was the reduction in interest rates by the Fed that has helped the economy to recover. And Selgin (2014) answered that it was not the Fed’s monetary easing but gold flows that has contributed to the recovery.

For the description of the 1920-1921 crisis, Vernon (1991) reports that the industrial production has fallen 25.6% below its January 1920 peak and bottomed out at 32.6% below its January 1920 level in July 1921 while the unemployment rate was 1.4% for both 1918 and 1919, 5.2% for 1920, and 11.7% for 1921. These figures make the 1920-1921 crisis a serious one, indeed. On the other hand, Vernon indicate that the causes of the dramatic deflation were due to both a decline in aggregate demand and a positive aggregate supply shock.

The ratio of the percentage decline in the GNP deflator for 1920-21 to the percentage decline in real GNP is 2.6 using the Department of Commerce figures, 3.7 using the Balke and Gordon data, and 6.3 using the Romer data. By contrast, during 1929-30, the first year of the Great Depression, the GNP deflator declined by 2.7 percent and real GNP by 9.4 percent, for a ratio of 0.3. The ratios of the percentage decline in GNP prices to the percentage decIine in real GNP for 1930-31, 1931-32, 1932-33, and 1937-38, the other Great Depression years in which real GNP declined, were 1.0, 0.9, 1.2, and 0.3, respectively, all well below the 1920-21 figures.

Romer (1988, Figure 1) argues that Commerce series data were less reliable than the revised Kendrick series. The first shows a 15% decline in GNP and the second shows a 3% decline between 1919 and 1921 (Romer, 1988, pp. 108-109). Although Romer admits the sharp decline in aggregate demand, the GNP fell by little and so the aggregate demand is unlikely to move the economy down. Furthermore, the real GNP rose 5% between 1917 and 1918 and the GNP deflator rose 15%. In contrast, between 1920 and 1921 a fall in real GNP of only 2% was associated with a price decline of 16%. Romer suspects that there may have been some type of aggregate supply shock either during the war or in 1921. What this tells us is that the Keynesian prescription (i.e., fiscal stimulus) wouldn’t apply here because there seemed to be no expectation of a further decline in aggregate demand.

In response to the Woods, Powell and Murphy articles, Kuehn (2010) points out that the Harding administration did cut tax rates for higher income families in 1922 (the highest bracket’s rates were reduced from 73% to 58%) and implemented an across the board rate reduction in 1923 (from 58% to 43.5% for the highest bracket and from 4% to 3% for the lowest bracket). However, these rate cuts were accompanied by a considerable expansion of the income taxable at any given rate. While the top bracket’s rate was reduced by 15% points from 1921 to 1922 in Harding’s Revenue Act of 1921, the income taxable at that rate was expanded from all income over $1,000,000 to all income over $200,000. The net effect was that the percent of individual income collected as revenue through the income tax has actually increased from 3.67% to 3.95%. Kuehn (2010, pp. 12-13) argues that the decline in the tax burden came too late to be considered as a factor in the recovery from the 1920-21 downturn. Harding entered office as the contours of the new “normalcy” were emerging.

With regard to the fall in the general price level after the Federal Reserve began increasing the discount rate, it was a dramatic one. In January 1921, a year after the rate increase began, prices had already fallen by 23.7% to a level last seen in November 1918. The deflation was also accompanied by a sharp decline in wages. The NBER index of average weekly earnings across 12 manufacturing industries declined by 34% from June 1920 to its lowest point in January 1922, a considerably steeper drop than the decline of 19% recorded in the BLS’s consumer price index over the same period. Other wage indices showed a more measured decline, such as the New York Federal Reserve’s Composite Wage Index, which fell by only 12%. King (1923) reports a drop in wages of 8.9% from the fourth quarter of 1920 to the first quarter of 1922. King (1923) also reports that factory wages declined by 14.5% over the same period, with wages in Metals and Metal Products falling 20.5%. Other figures are available from the NICB 1922 report; Wages and hours in American industry, July 1914-July 1921. The nominal wage declines did not occur before the beginning of 1921. See pages 8, 10-11, 16-17, 21, 26, 31.

Subsequently, Kuehn (2012) provides the following graph :

A note on America's 1920-21 depression as an argument for austerity (Kuehn 2012) Figure 1

Most of the federal spending declines occur before the beginning of the crisis, even though there is a slight decline in federal expenditures starting from the beginning of 1921. What Kuehn was pointing out is that the formidable slash in public spending occurred before Harding was even elected president (who was in office in March 1921). He then suggests that the recovery was accomplished by the Federal Reserve System, when it initiated discount rate reductions in May 1921 to keep the recovery on track. The graph below shows the discount rates (taken from Jon Catalan’s blog; the old blog, not the one in link).

discount rates 1920 depression

A mere reduction of 0.5% on interest rates may account for the recovery. Really ? My impression is that if the reduction leads to a quick recovery, the keynesians would claim that the reduction was sufficient, and if there is no recovery, the keynesians would claim that the reduction was not enough. With this kind of logic, they always win. Anyway, according to the Bank of Canada, the effect of a change of the interest rates on aggregate output takes several months :

Monetary Policy: How It Works, and What It Takes

The Bank of Canada’s policy actions relating to the overnight interest rate have almost immediate effects on the exchange rate and interest rates, but current estimates suggest that it takes between 12 and 18 months for most of the effect on aggregate output to be observed. Most of the effect on inflation is not apparent for between 18 and 24 months (Duguay 1994). And even these estimates are subject to considerable variation. … In particular, these long time lags mean that central banks must be forward-looking in their policy decisions. … If, on 1 January 2005, the Bank of Canada observes an event in the world economy that is likely to reduce aggregate demand beginning in June of the same year, there is nothing the Bank can do in January to fully offset that shock. Even if it responded immediately and lowered its policy rate in early January, there simply would not be enough time for its policy to stimulate aggregate demand sufficiently to offset the effects of the shock by June.

As noted by Selgin (2014) however, it is even worse in the case of the Fed of the 1920s, which policy rate was the discount rate, rather than the federal funds rate (which target rate is achieved through overnight loans of bank reserves). Thus, in the 20s, a lowering of the Fed’s policy rate (discount rate) might not even imply an increase in Fed lending or security purchases. In reducing its discount rate, the Fed merely allowed banks possessing the requisite commercial paper to discount that paper with it at the newly reduced rates. Whether they would do so, however, depended on whether the rates in question were low, not merely compared to previous rates, but relative to market rates or to natural rates. If not, the volume of discounting might not budge, and the lower rates would not imply any actual monetary expansion, except perhaps relative to the contraction that might have ensued had rates remained high.

Now, there is another interpretation of the 1921 crisis. Selgin at has left a comment on the book The Forgotten Depression of Jim Grant (2014). Monetary expansion helped for the post-1921 recovery but not owing to the Fed’s easing.


As you can see from the chart, although there was some increase in “bills discounted” in response to the Fed’s lowering of its discount rate, the increase was slight compared to the massive decline in total Fed non-gold assets since 1920. What’s more, it was more-or-less perfectly–and by implication quite intentionally–offset or “sterilized” by means of Fed sales of government securities. The Fed’s contribution to recovery, in short, consisted, not of any actual monetary stimulus, but of a mere cessation of what had been a precipitous decline in its interest-earning asset holdings.

This isn’t to say that monetary expansion played no part in the post-1921 recovery. In fact, it played a significant part. But the expansion that took place was due solely to gold inflows, which were themselves encouraged by relatively high interest rates as well as by falling prices–that is, by the normal working of the price mechanism rather than by activist Fed policy. (In the 30s as well, by the way, such recovery as took place was entirely the result not of Fed easing–or of fiscal stimulus–but of the dollar’s devaluation and subsequent gold inflows from Europe.) That gold flows (as opposed to Fed easing) contributed to the post-1921 recovery is itself a fact that Jim Grant readily acknowledges; his book’s 17th chapter is called “Gold Pours into America.”


  1. Kuehn, D. (2011). A critique of Powell, Woods, and Murphy on the 1920–1921 depression. The Review of Austrian Economics, 24(3), 273-291.
  2. Kuehn, D. (2012). A note on America’s 1920–21 depression as an argument for austerity. Cambridge journal of economics, 36(1), 155-160.
  3. Romer, C. D. (1988). World War I and the postwar depression: A reinterpretation based on alternative estimates of GNP. Journal of monetary Economics, 22(1), 91-115.
  4. Vernon, J. R. (1991). The 1920‐21 Deflation: The Role of Aggregate Supply. Economic Inquiry, 29(3), 572-580.
This entry was posted in Economics and tagged . Bookmark the permalink.