An alternative date for the recession of 1920-1921

Ok, it's been awhile since I wrote a post in English so let’s start. In order to downplay the possibility of self-recovery from the recession of 1920-1921, some people from Keynesian and monetarist camp have claimed that, despite the fact that fiscal policy played a contractionary role, expansive monetary policy by the Fed had a significant if not fundamental role for the beginning of the recovery.

The main shortcoming I find with this story is that it only uses the dates of National Bureau of Economic Research (NBER) for the beginning and end of the contraction of early 20s. Why is that a “shortcoming”? Because NBER recession dates have been severely questioned several years ago mainly by, among others, Christina Romer (Romer, 1994). As Romer (1999) explains: “[T]he NBER’s dating procedures have not been entirely consistent over time.” The post-World War II recession dates are derived from aggregate indicators in levels while, on other hand, the prewar and interwar dates are derived from detrended (long-term trend removed) series. When you detrend data series which are generally upward sloping, like real GDP, the result is a serie which tends to peak earlier and trough later compared to the series in levels. The consequences of this inconsistent procedure are not trivial for Romer (1999): “As a result, the earlier procedure of using detrended data is likely to make pre-World War II expansions look shorter and pre-World War II recessions look longer than they would if postwar procedures had been used.” So, as we can see, the pre-World War II NBER recession dates are not as reliable as you can expect. It is true that NBER recession dates are, by far, the most widely used, but that fact does not increase their accuracy at all. Once we use the much better and modern Romer’s dates, we find some interesting things about the “Depression of 1920-1921”. As we will see, this alternative and better dating casts serious doubts about the alleged “monetary policy exit” from recession.

According to NBER the peak was reached in January 1920 and the trough in July 1921. On other hand, while the peak was also January 1920, for Romer (1999) the trough was reached in March 1921. Now we can put this new recession date and monetary policy data together and check whether or not the slump was over because of the actions of the FED.

Let’s start with Fed’s discount rate data. In the next graph, the monthly discount rate of four Federal Reserve banks are plotted for the period 1919-1922. The recession interval is taken from Romer (1994; 1999) in the usual manner as starting the immediate period following the peak. The leading bank in those days was Federal Reserve bank of New York under the leadership of Benjamin “the greatest central banker ever” Strong, so its discount rate is an important index of the ease or tightness of monetary policy. You can also see the discount rate from Federal Reserve banks of Minneapolis, San Francisco and Dallas. 

Monthly discount rates of Federal Reserve banks
and recession, 1919-1922 (Y axis do not start at zero).
Source: FRED and Romer (1994; 1999)

As we can see, if we use the modern recession dates of Romer, the U.S. was already out of the slump even when Federal Reserve discount rate was at its historical maximum of 7 %. Even when every single discount rate was still at its highest, the economy was able to manage an escape from the recession.

But monetary policy is not only about interest rates, it is also important to investigate what happened to money supply. 

Year over year growth of monetary base, M1, M2 and M3 and recession, 1919-1922.
Source: FRED, Friedman and Schwartz (1970: 16-21) and Romer (1994; 1999)

The year over year rate of growth of the monthly (seasonly adjusted) data of the monetary base and of the broad money supply M1, M2 and M3 (Friedman and Schwartz, 1970: 16-21, columns 8, 9 and 11 respectively) clearly shows that the recession ended long before the peak of the monetary contraction was reached.

Monthly monetary base, M1, M2 and M3
with January 1919 = 100 and recession, 1919-1922.
Source: FRED, Friedman and Schwartz (1970: 16-21) and Romer (1994; 1999)

Even if we plot every measure of the monthly money supply, we clearly see that the recession ended several months before the contraction of the stock of money was over. The recession lasted until March of 1921 but the broad money supply (M1, M2 and M3) contracted until September. After that the level of broad money supply stagnated and it did not start to grow in a considerable way until April 1922. Meanwhile the monetary base contracted until February 1922. In other words, according to these data, the end of the recession was not preceded, much less helped, by a monetary easing. On the contrary, the slump ended despite the fact that the money supply was decreasing continuously.

Thus, according Romer’s dates, it is highly doubtful that the termination of the 1920-21 recession was helped by the FED’s expansionary actions. Mainly because the recession was already over when, according to discount rate and money supply data, the Federal Reserve was still being contractionary.

Friedman, Milton and Schwartz, Anna J. (1970) Monetary Statistics of the United States: Estimates, Sources, Methods. National Bureau of Economic Research. New York: Columbia University Press.

Romer, Christina D. (1994) “Remeasuring Business Cycles.” The Journal of Economic History. Vol. 54, No. 3, pp. 573-609.

Romer, Christina D. (1999) “Changes in Business Cycles: Evidence and Explanations.” The Journal of Economic Perspectives. Vol. 13, No. 2, pp. 23-44.