Review of Milton Friedman and Anna J. Schwartz's 'A Monetary History of the United States, 1867-1960'

Lucas argues the book’s real contribution goes beyond its data: “I think it is clear that A Monetary History is much more than a collection of useful time series… it organized nearly a century of U.S. macroeconomic evidence in a way that has had great influence on subsequent statistical and theoretical research” (Sec. 1, p. 5). He credits it with a decisive role in the 1960s debates between Keynesians and monetarists over stabilization policy, and states the purpose any narrative history must serve is “to organize a coherent story of important events” – which he judges the book to have done “well” (Sec. 1, p. 5).

"A Monetary History constructs this vision through the consistent application to specific historical events of two simple principles" (Sec. 2, p. 6). The first is long-run monetary neutrality: “there is a trend path of real output, governed by forces that are not examined in the book, which has the property that neither its level nor its growth rate is affected by monetary policies… the economy returns to its trend behavior after displacements.” The second is a short-run non-neutrality: “fluctuations in M2 induce spending fluctuations and these, in the face of nominal price rigidities, induce real output fluctuations,” with “no effort… made to elucidate or explain the nature of these price rigidities, except to say that they are transient” (Sec. 2, p. 6).

“Given this account of observed depressions, the normative analysis is straightforward: the monetary authority has always had the ability to eliminate M2 instability, and it should have done so. In every instance, Friedman and Schwartz provide a detailed, operational account of how and when actions could have been taken that would have achieved this outcome” (Sec. 2, p. 7). Lucas summarizes the resulting picture: “in place of a ninety-year period that in fact included many depressions and episodes of both deflation and inflation, one is given a vision of the way this portion of our history might have evolved, with stable prices and smoothly growing real output” (Sec. 2, p. 6).

Yes, explicitly and strongly: “I will say that I find the argument of A Monetary History wholly convincing. I think Friedman and Schwartz are right to focus on the avoidance of the really major macroeconomic disasters of the past as the main responsibility of current monetary policy. I find their diagnosis of the 1929-33 downturn persuasive and indeed, uncontested by serious alternative diagnoses” (Sec. 3, pp. 7-8). He adds a memorable aside: “If I ever go to Washington for some reason other than viewing cherry blossoms, I will pack my copy of A Monetary History and leave the rest of my library – well, most of it – at home” (Sec. 3, p. 8).

"A Monetary History is full of numbers, but there are many quantitative questions to which its model-free approach cannot provide answers" (Sec. 3, p. 8). He quotes Friedman and Schwartz’s own qualitative conclusion about the Great Contraction – that preventing the money-stock decline “would have reduced the contraction’s severity and almost as certainly its duration,” with output and prices unlikely to have fallen as far absent the monetary collapse (p. 301 of the book, quoted Sec. 3, p. 8) – and observes: “This is not a verbal summary of tables describing the results of a numerical simulation; it is the simulation.” He presses the natural follow-up questions the book cannot answer: “by how much would the decline in real output to 1933 have been reduced had such a monetary policy been pursued? In general, what would the variance in real output growth have been over the 90-year period under study had money growth been smooth?” (Sec. 3, pp. 8-9).

A number of explicit models (Lucas 1972; Fischer 1977; Phelps and Taylor 1977; Taylor 1979; Mankiw 1985) “were designed to reconcile the two neutrality principles on which Friedman and Schwartz built… using the principle of rational expectations, neutrality in the long run was preserved,” while short-run non-neutrality arose from some form of nominal price rigidity generating real effects from unanticipated (but not anticipated) monetary changes (Sec. 5, pp. 10-11). Crucially, Lucas notes these models do not agree on policy implications: “though it is now clear that the two neutrality principles used by Friedman and Schwartz can be reconciled, the question of the appropriate conduct of monetary policy remains unresolved… This conclusion depends critically on the details of the way price rigidities are modeled” – his own 1972 model with competitive markets implies smooth money is efficient even facing real shocks, but contract-based models (Fischer 1977, Phelps and Taylor 1977, Taylor 1979, Mankiw 1985) “there is no presumption that simply removing monetary variability will result in a system that responds efficiently to other shocks” (Sec. 5, p. 11).

Lucas credits Sargent’s study of the disinflations that ended the European hyperinflations and the 1920s French inflation with showing that “one can interpret them as anticipated, even though sudden and drastic, and hence reconcile their magnitude with the modesty of the real effects they induced” (Sec. 5, p. 11) – resolving what would otherwise be a puzzle, since “an unqualified association between monetary contractions… and real activity would lead one to expect these disinflations to have been associated with major depressions.”

“Sims (1972) took a very different approach to the study of monetary influences on real activity… Rather than attempting to construct an economic model consistent with the principles applied in A Monetary History, Sims developed a purely statistical definition of cause, related to Granger (1969), in terms of lead-lag relations among variables” (Sec. 6, p. 12), providing a test of whether money causes (in his sense) real output movements and estimates of the fraction of output variance attributable to monetary instability, by frequency. Lucas notes Sims’s lead-lag findings play “a very similar role, though not formalized in the same way, in Friedman and Schwartz’s discussion” of monetary independence.

Romer and Romer (1989) drew on the independence idea to argue for using historical ’natural experiments’ where money movements did not occur in response to real events; for them “exogeneity is a property of a particular realization,” while for Sims “it is a property of a distribution” – Lucas states “the two approaches are not the same” (Sec. 6, p. 12). He identifies a third sense, “which I prefer,” closer to Friedman and Schwartz’s own usage: independence not as statistical exogeneity but as meaning “that whatever the sources of monetary contractions may have been, on average or in particular instances, the monetary authorities could have maintained M2 growth had they chosen to do so.” Lucas judges this the sense “conclusively defended by Friedman and Schwartz in detailed analysis of episode after episode” (Sec. 6, p. 12).

“I do not see any possibility of obtaining answers to normative questions of economic policy by atheoretical, purely statistical means. But the attempt to estimate the fraction of real variability… that can be attributed to monetary instability by atheoretical (Sims) or similar methods… is certainly worth pursuing, and success in this effort would obviously be immensely useful in guiding future theorizing” (Sec. 6, p. 12).

He presents a direct quantitative objection: “The Solow (1957) residuals for the years 1928 through 1933 were: 0.020, -0.043, 0.024, 0.023, 0.011, 0.072! There is no real business cycle model that can map shocks into anything like the 40% decline in real output and employment that occured between 1929 and 1933” (Sec. 6, p. 13). He asks rhetorically what technological or psychological shocks of this magnitude could have gone “unremarked at the time, and remain invisible even to hindsight,” and concludes “it is surely no accident that no one has attempted to apply real business cycle theory to the 90-year period Friedman and Schwartz studied” (Sec. 6, p. 13).

He proposes reading Kydland-Prescott-style models as a normative benchmark rather than a universal positive theory: “One may thus think of the model not as a positive theory… to all historical time periods but as a normative benchmark providing a good approximation to events when monetary policy is conducted well and a bad approximation when it is not” (Sec. 6, p. 13). On this reading, the comparatively small role of money in accounting for postwar U.S. fluctuations should be read “not as evidence that money doesn’t matter” but “as evidence that postwar monetary policy has resulted in near-efficient behavior” (Sec. 6, p. 13) – i.e., the very success of RBC models postwar is itself indirect testimony to competent monetary management, consistent with rather than contradicting Friedman and Schwartz’s normative thesis.

He calls it “a remarkable and durable achievement of historical and economic scholarship,” crediting Friedman and Schwartz with using “a few basic economic principles to organize nine decades of tremendously varied economic history into a coherent picture… in which the effects of identifiable causes can follow” – one consistent with the instinct “that the depression of the 1930s was an event that should not have happened, a preventable disaster” (Sec. 7, p. 14). He concludes that subsequent research has mainly worked “to sharpen” this picture rather than overturn it, and states: “I find myself relieved to agree with Friedman and Schwartz that we already know enough, and knew enough in 1963, to avoid the major policy mistakes of the interwar period. Whatever may be the influence of A Monetary History on future research, it will stand as the classic statement of these important lessons from our past” (Sec. 7, p. 15).