Recent events such as the “deflationary boom” in China have led a few mainstream macroeconomists to re-examine and revise their views on the phenomenon of deflation, conventionally defined as a general and persistent decline in prices. The long-held view that a general fall in prices, or increase in the value of money, whatever its origin spells unmitigated disaster for overall economic activity and social welfare has begun slowly to give way to attempts to distinguish between “good” and “bad” deflation. The distinction between “corrosive” deflation and deflation that is compatible with healthy economic growth has even penetrated the publications of some of the more enlightened regional Federal Reserve Banks.1
These developments, while gratifying to Austrian economists, are hardly sufficient to undo nearly a century of myths about the pernicious effects of deflation that have been systematically perpetrated by professional economists beginning with the proto-monetarist Irving Fisher. But now comes a simple and straightforward empirical study published in the leading academic economics journal by two economists with impeccable mainstream credentials and affiliations successfully challenging the most widespread and deeply ingrained belief about deflation: that there is a well established empirical relationship between deflation and depression.2
Atkeson and Kehoe utilize panel data on inflation and real output growth for seventeen countries including the United States, the United Kingdom, France, and Germany. The data set for each country encompasses at least 100 years. The authors focus on medium-term fluctuations by breaking the time series on inflation and on economic growth for each country into periods of five years and calculating the average annual rates of real output growth and inflation for each such period or “episode.”
“Deflation” is then defined for each episode as “a negative average inflation rate” and “depression” as “a negative average real output growth rate.” The five-year episodes are selected so as to begin and end with years ending in “9” or “4” so that, for example, the years of the Great Depression (1929–1933) and the depression of 1921–22 are grouped together in single episodes, 1929–1934 and 1919–1924, respectively.
The Great Depression Episode
Isolating the Great Depression episode, the authors do find a loose link between deflation and depression. All 16 countries for which data were available experienced deflation during this episode, while only 8 of the 16 experienced depression. Output growth was regressed on a constant and the inflation rate, and the estimated slope coefficient was .40 while the standard error was .28.
In other words, a one-percentage point reduction in inflation is associated with a .40 percentage point decline in real output growth during the Great Depression, although even during this episode the probability that there is no relationship between deflation and depression (the level of significance) exceeds 10 percent. In the jargon of statistical inference this means that the relationship between deflation and depression is not “statistically significant.”
All Episodes Exclusive of the Great Depression Episode
When the authors leave the Great Depression aside, and plot average inflation and output growth rates for all countries for all five year episodes—which begin in 1820 for some countries in the sample—except 1929–1934 it turns out that 65 of 73 deflation episodes involved no depression while 21of 29 depression episodes were not associated with deflation. In other words, 90 percent of deflation episodes did not culminate in depression. From this the authors conclude, “In a broader historical context, beyond the Great Depression, the notion that deflation and depression are linked virtually disappears.” This conclusion is also supported by the slope coefficient and the standard error for the data excluding the Great Depression, which are 0.04 and 0.03 respectively.
All Episodes
When the regression is run for all episodes including the Great Depression, the result is that a 1-percentage point drop in inflation is associated with a piddling decline in the average growth of real output of .08 percentage points with a standard error of .03. While this result is statistically significant it is certainly not economically significant.
Thus, for example, assuming the value of money was initially constant and then began to appreciate by 1 percent per year, real output growth in the economy would fall from, say, 3.00 to 2.92 percent per year. This means that even a massive deflation of 30 percent per year visited upon an economy that was growing at 3.00 percent per year would not cause a depression—defined as negative growth of real output—since it would only lower the real growth rate by 2.4 percentage points to 0.60 percent per year. From a strictly empirical standpoint, then, the Great Depression can hardly be explained by a price level that declined by about 5 percent per year between 1929 and 1933.3
Pre- and Post-World War II Episodes
The study also finds that the relation between deflation and depression differs markedly before and after World War II. The regression using the prewar data yields a slope coefficient of 0.11 with a standard error of 0.04 indicating a weak link between deflation and depression. In contrast, the regression run with postwar data suggests no link between deflation and depression with the slope coefficient of –0.03 and a standard error of 0.04. Indeed the negative slope coefficient suggests the possibility of a link between inflation and depression. Given this empirical result, you might hope that the large and growing contingent of mainstream economists who are clamoring for the Fed to implement “inflation targeting” of 2 to 3 percent per year—i.e., to deliberately dilute the purchasing power of the dollar by a fixed percentage every year—would now switch to prescribing deflation targeting of a few percent per year just to be on the safe side.4
Conclusion and Implications
As Atkeson and Kehoe conclude: “The data suggest that deflation is not closely related to depressions. A broad historical look finds more periods of deflation with reasonable growth than with depression, and many more periods of depression with inflation than with deflation. Overall, the data show virtually no link between deflation and depression.” (Emphases added.) The authors caution, however, that their study “characterizes the relation in the raw data between deflation and output growth, with no attempt to control for anything” and “perhaps a link between deflation and depression could be teased out of the data with a well-motivated set of controls.”5
From the Austrian standpoint, it is precisely the virtue of the Atkeson-Kehoe study that it uses raw data that have not been subjected to arbitrary statistical manipulations. For it is unaveraged, unsmoothed, unadjusted data that are the direct and immediate outcome of unique and non-repeatable human choices in the marketplace. As such, these data are the most meaningful in applied theoretical analysis and for the interpretation of economic history.
As Murray Rothbard often emphasized: ”Austrians realize that empirical reality is unique, particularly raw statistical data. Let that data be massaged, averaged, seasonals taken out, etc. and then the data necessarily falsify reality.”6
Rothbard objected even to the seemingly innocuous practice of seasonally adjusting the data: ”In our view the further one gets from the raw data the further one goes from reality, and therefore the more erroneous any concentration upon that figure. Seasonal adjustments in data are not as harmless as they seem, for seasonal patterns, even for such products as fruits and vegetables, are not set in concrete. Seasonal patterns change, and they change in unpredictable ways, and hence seasonal adjustments are likely to add distortions to the data.”7
The Atkeson-Kehoe study has a number of important implications for competing schools of macroeconomics. First, from the point of view of “aprioristic” or logical-deductive Austrian economic theory, while it does not validate or “falsify” any particular theoretical approach to business cycle theory, it is certainly illustrative of the Mises-Rothbard argument that an increase in the value of money is neither a necessary cause of depression nor an impediment to healthy economic growth. Second, the monetary disequilibrium approach to depression, which has been embraced by many, but not all, of the “free banking” wing of Austrian macroeconomics, seems to now have a serious problem.8 According to this approach, which was initiated in the writings of the proto- monetarist, FDIC official Clark Warburton in the 1940’s, the primary cause of depression is the emergence of an excess demand for money in the economy whose effects are not instantaneously or rapidly neutralized by a corresponding increase in the supply of money.9 The result is a tendency toward a general fall in prices (deflation) and, at least in the short run, in real output (depression).
The lack of a historical relation between deflation and depression found by Atkeson and Kehoe, however, seems to indicate that the monetary disequilibrium theory of depression, which is also a logical-deductive theory, is inapplicable to most, if not all, of empirical reality. Along the same lines, the study also demolishes one of the main props of the argument in favor of unregulated fractional-reserve banking. If there is no link between deflation and depression then there is no need for banking institutions that putatively respond to every change in the demand for money with an offsetting change in the supply of money.
Finally, the study is potentially devastating to the now widely accepted Friedman-Schwartz explanation of the Great Depression. In a recent symposium celebrating the fortieth anniversary of their famous work, A Monetary History of the United States, Milton Friedman correctly noted, “The most controversial of [our major themes]—our attribution to the Federal Reserve of a major share of the responsibility for the 1929–1933 contraction—has become almost conventional wisdom.”10
Friedman and Schwartz ascribed culpability to the Fed for what they called the “Great Contraction” because it allegedly pursued deflationary policies in the early 1930’s.11 Unfortunately, for Friedman and Schwartz the causal connection they posited between the deflation and depression of the early 1930’s was purely empirical, based not on sound praxeological reasoning, but on statistical correlations using the data of a single country for the years 1857–1960.
With the validity of their correlations now called into serious question by a study using well over 100 years of data from seventeen different countries, we may yet see the deflation-depression link follow another supposedly ironclad empirical relation, the Phillips Curve, into well-deserved oblivion.
- 1“Deflation“ The Federal Reserve Bank of Cleveland 2002 Annual Report released May 9, 2003; and James B. Bullard and Charles M. Hokayem, “Deflation, Corrosive and Otherwise,” National Economic Trends (July 1, 2003), p. 1 available at research.stlouisfed.org.
- 2Andrew Atkeson and Patrick J. Kehoe, “Deflation and Depression: Is There an Empirical Link,” American Economic Review Papers and Proceedings 94 (May 2004): 99–103. The first co-author is on the faculty of the Department of Economics of UCLA and the second works in the Research Department of the Federal Reserve Bank of Minneapolis.
- 3This figure is computed from the data in Kenneth Weiher, America’s Search for Economic Stability: Monetary and Fiscal Policy since 1913 (New York: Twayne Publishers, 1992). Pp. 39, 57.
- 4On the current status of the inflation targeting debate, see the articles in Inflation Targeting: Prospects and Problems, Proceedings of the Twenty-Eighth Annual Economic Policy Conference of the Federal Reserve Bank of St. Louis, Federal Reserve Bank of St. Louis Review 86 (June/August 2004).
- 5Atkeson and Kehoe, p. 102.
- 6Murray N. Rothbard, Making Economic Sense (Auburn, AL: Ludwig von Mises Institute, 1995), pp. 233–34.
- 7Murray N. Rothbard, The Mystery of Banking (New York: Richardson & Snyder, 1983), p. 259.
- 8Two works which are representative of this branch of Austrian macroeconomics are: George A. Selgin, The Theory of Free Banking: Money Supply under Competitive Note Issue (Totowa, NJ: Rowman & Littlefield, Publishers 1988); and Steven Horwitz, Microfoundations and Macroeconomics:An Austrian Perspective (New York: Routledge, 2000).
- 9Clark Warburton, Depression, Inflation, and Monetary Policy: Selected Papers 1945–1953 (Baltimore: The Johns Hopkins Press, 1966). The most prominent contemporary proponent of this approach is Leland Yeager. (See Leland Yeager, TheFluttering Veil: Essays on Monetary Disequilibrium, ed. George Selgin (Indianapolis, IN: Liberty Fund, Inc., 1997))
- 10Milton Friedman, “Reflections on A Monetary History,” The Cato Journal 23 (Winter 2004), p. 349. For evidence that the Friedman-Schwartz view has indeed become the dominant explanation of the Great Depression among mainstream macroeconomists, see Ben S. Bernanke, Essays on the Great Depression (Princeton, NJ: Princeton University Press, 2000), pp. 6–8.
- 11For a defense of Murray Rothbard’s view that the Fed did not pursue a deflationary policy in the early 1930’s and, indeed, tried desperately to inflate the money stock, see Joseph T. Salerno, “Money and Gold in the 1920 and 1930s: An Austrian View,” Ideas on Liberty 49 (October 1999).
