Friedman and Schwartz's Monetary Explanation of the Great Depression
Total Page:16
File Type:pdf, Size:1020Kb
PRELIMINARY DRAFT, DECEMBER 31, 2012 Friedman and Schwartz’s Monetary Explanation of the Great Depression: Old Challenges and New Evidence By CHRISTINA D. ROMER AND DAVID H. ROMER* * C. Romer: University of California, Berkeley, Berkeley, transmission mechanism. In Section I, we dis- CA 94720-3880 (email: [email protected]); D. Romer: University of California, Berkeley, Berkeley, CA cuss the challenge to Friedman and 94720-3880 (email: [email protected]). Schwartz’s explanation provided by the anomalous behavior of nominal interest rates. For many, Friedman and Schwartz’s Mone- Previous work has shown that expectations of tary History of the United States (1963) is deflation can explain how falling nominal in- synonymous with the notion that monetary terest rates could be consistent with a high real contraction and errors by the Federal Reserve cost of borrowing. But we argue that the mon- caused the Great Depression. Though that is etary explanation requires not just that there one of the book’s conclusions, this quick were expectations of deflation, but that those summary both sells the book short in im- expectations were the result of monetary con- portant ways and oversells its findings about traction. Because previous work has been the 1930s. largely silent on this issue, we are left without The crucial way it sells the book short is evidence about a necessary link from mone- that the contributions of the Monetary History tary shocks to the Depression. extend far beyond the Depression. Friedman In Section II, we analyze one component of and Schwartz use an extensive reading of the the business press in detail to see if there was historical record over nearly a century to iden- a link between monetary shocks and expecta- tify times when the money supply moved for tions of deflation in the central years of the reasons unrelated to current or prospective downturn—1930 and 1931. We find evidence macroeconomic conditions. That output that these professional observers did indeed moved in the same direction as money follow- expect deflation in substantial part because of ing these “crucial experiments” remains some Federal Reserve behavior and monetary con- of the strongest evidence we have that mone- traction. This suggests that monetary shocks in tary shocks have real effects. the Depression may have affected output and At the same time, saying that the book employment by raising real interest rates. proves that monetary shocks caused the Great Depression is a stretch. Of the monetary I. Challenges to Friedman and Schwartz’s shocks Friedman and Schwartz identify, those Explanation of the Great Depression for the early years of the Depression are argu- ably the most tenuous. And crucially, the book It is useful to begin with a review of Fried- provides scant discussion about the mecha- man and Schwartz’s monetary explanation of nism by which monetary shocks affected the the Depression and the literature that has de- economy. This weakness is most pressing in veloped both challenging and supporting it. the discussion of the Depression. Because nominal interest rates fell precipitously early A. Friedman and Schwartz’s Explanation in the Depression and remained low through- out, it is hard to appeal to the standard trans- The core of Friedman and Schwartz’s treat- mission mechanism operating through the ment of the Depression (Chapter 7 of the nominal cost of credit. Yet Friedman and Monetary History) is a careful historical anal- Schwartz provide little guidance as to how ysis of the underlying reasons for the decline they believe monetary contraction nonetheless in the money stock in the early 1930s. This is led to deep output declines in this period. where Friedman and Schwartz make their case This paper seeks to fill in the gap in the that the decline had a significant exogenous Friedman and Schwartz explanation of the component. Depression by investigating this missing 2 Friedman and Schwartz’s implicit definition 7 of a monetary shock is very broad: any 6 movement in money that is unusual given the 5 economic circumstances. One shock they identify in the early 1930s that is easy to de- 4 fend on these grounds is the Federal Reserve’s Percent 3 decision to raise the discount rate by 200 basis points in October 1931following Britain’s de- 2 parture from the gold standard. Quite clearly, 1 the resulting fall in the money stock was not an endogenous reaction to the fall in output. It 0 was the result of a conscious decision moti- vated by another consideration—defending Jan-25 Jan-26 Jan-27 Jan-28 Jan-29 Jan-30 Jan-31 Jan-32 Jan-33 the gold standard. FIGURE 1: COMMERCIAL PAPER RATE Friedman and Schwartz also suggest that there were other exogenous monetary shocks between 1929 and 1933, though they are real and nominal interest rates. Such an ac- vague about the number and their timing. count fits what happened following many of Their argument is that the Federal Reserve Friedman and Schwartz’s contractionary mon- stood by as banking panics caused large de- etary shocks—such as 1920 and the panics clines in the money multiplier, and hence in under the National Banking System. But it the supply of money, and that under either the provides a terrible description of what hap- previous clearinghouse system or a better pened during most of 1929–33. Figure 1 shows the monthly prime commer- functioning central bank, such drops in the 1 money supply would not have occurred. cial paper rate for 1925 to 1933. One thing In Chapter 7, the role of the monetary con- that stands out is the sharp rise in nominal in- traction in causing the real decline is largely terest rates following the October 1931 in- implicit. Only after Friedman and Schwartz crease in the discount rate. Interest rates also bring in other episodes—1920, 1936–37, and rose noticeably beginning in early 1928. As perhaps the panics under the National Banking described by both Friedman and Schwartz and System—do they have a clear case that exog- Hamilton (1987), the Federal Reserve tight- enous monetary contractions cause output ened policy moderately in this period to try to contraction. Armed with that relationship, one rein in speculation in the stock market. Since can then go back to the 1930s and say that policy was not responding to current or pro- since there were large exogenous monetary spective economic growth, but rather to asset contractions in this period, they likely caused prices, this episode appears to fit Friedman much of the real decline. and Schwartz’s definition of a monetary shock. However, other than these two times, the B. The Anomalous Behavior of Nominal course of nominal interest rates during the Interest Rates Depression was strongly downward. Rates plummeted following the Great Crash of the A striking feature of the Monetary History stock market in October 1929. They fell fur- is that the approach is almost entirely reduced ther during the first two waves of panics (Oc- form. Friedman and Schwartz focus on the tober 1930 and March 1931). After the brief evidence that output and prices fell after mon- rise in late 1931, they fell continuously until etary shocks. They provide little discussion February 1933. and virtually no evidence on the possible transmission mechanism of monetary shocks to spending and output. Most scholars have assumed that Friedman and Schwartz had in mind a conventional in- terest-rate channel. In the textbook IS-LM 1 model, an exogenous reduction in the money The data are from Board of Governors of the Federal Reserve supply shifts the LM curve back and raises System (1943, Table No. 120, pp. 450-451). 3 C. Challenges to Friedman and Schwartz’s panics proposed by Bernanke (1983): the Emphasis on Monetary Shocks banking panics had effects on spending not just through a decline in the money supply, The literature has responded to this anoma- but also by causing a reduction in credit sup- lous behavior of nominal rates in two ways. ply at a given level of the safe interest rate. One is to suggest that Friedman and Schwartz There is certainly some evidence that such ef- are simply wrong about the importance of fects were present, but evidence that they were monetary shocks, at least in some time peri- large enough to account for the enormous out- ods. put decline is lacking. Temin (1982) argues that the behavior of There were also contractionary fiscal chang- nominal rates implies that the IS curve must es in the early 1930s. In particular, the passage have shifted back more than the LM curve. in June 1932 of the Revenue Act of 1932, Temin posits an unusual drop in consumer which raised taxes by roughly 2 percent of spending as the culprit, but cannot identify a GDP, was a surely a substantial IS shock. But cause. Romer (1989) suggests that a rise in it came too late to explain the dramatic fall in income uncertainty associated with the stock output and interest rates in 1931 and early market crash may have been the source. Olney 1932. (1999) argues that particulars of the household Thus, the hypothesis that nominal rates fell bankruptcy law led consumers to cut spending because monetary shocks were not the main to ensure their ability to pay consumer credit source of the downturn has considerable sup- charges. port for the period immediately following the We suspect that even Friedman and stock market crash. But scholars have not per- Schwartz would agree that nonmonetary forc- suasively identified large nonmonetary shocks es were important in the first year of the De- for the critical period from roughly late 1930 pression.