«BEYOND SHOCKS: WHAT CAUSES BUSINESS CYCLES? AN OVERVIEW Jeffrey C. Fuhrer and Scott Schuh* In the summer of 1997, when the Federal Reserve Bank of ...»
First, “most downturns are domestic in origin, and are not powerfully transmitted to the other important trading nations.” Second, if one were interested in international transmission, one would focus on 1876, a year in which the Continent and Canada experienced declines in GDP, and on 1879 and 1908, years in which several countries experienced output declines. Third, Belgium exhibits only one downturn during these
BEYOND SHOCKS: WHAT CAUSES BUSINESS CYCLES? AN OVERVIEW 11periods, a suspicious ﬁnding given the 12 downturns in neighboring Netherlands and 14 in France. As a result, Cooper calls into question the reliability of the annual data for any of these countries prior to 1914.
For the period 1957 to 1994, Cooper notes that the few recessions have been concentrated in ﬁve years: 1958, 1975, 1981-82, and 1993. This suggests strong international transmission, in contrast to the earlier period. All of the recessions in the United States were accompanied by recessions elsewhere. The greater coherence may be attributed to the importance of the oil price shocks in these recessions, Cooper notes.
Cooper goes on to question the detrending method used by Bergman and his coauthors. Only 60 percent of their recessions match NBER reference dates. The issue of appropriate ﬁltering is important when considering the welfare implications of business cycles, Cooper suggests.
A departure of output below its (rising) trend may imply relatively little lost income or underutilized resources, whereas an absolute decline in output would almost surely entail signiﬁcant welfare losses.
Cooper outlines a number of broad changes in industrial economies that would lead one to question Bergman, Bordo, and Jonung’s conclusion about the stability of the business cycle over long spans of time. He suggests that “the most dramatic by far... is the reduction in the fraction of the labor force required for food production.” The decline in this number from about one-half in 1880 to below 5 percent by 1995 for all of these countries is likely to have altered the dynamics of the business cycle signiﬁcantly, according to Cooper. Other important secular changes include the increased participation of women in the paid work force, the growth in the importance of government expenditures, and major technological innovations, including electricity, automobiles, and aircraft.
“A relatively unchanged economic cycle that survived these dramatic secular changes in modern economies would be robust indeed,” Cooper suggests.
INTEREST RATE POLICY BUSINESS CYCLESAND Christopher Sims examines one of the most contentious questions in macroeconomics: the role of monetary policy in twentieth-century business cycles. Sims points out that one cannot determine the inﬂuence of monetary policy simply from observed changes in interest rates and output. The observation that a rise in interest rates precedes each postwar recession does not show that policy-induced interest rate movements caused the recession. If, for example, rapid expansion of private demand for credit systematically causes all interest rates to rise near the end of an expansion, this rise in interest rates should not be interpreted as the cause of a subsequent slowdown; it is a consequence of previous strong demand. Because such “eyeball” interpretations of the data can lead to confusion about the role of monetary policy, Sims advocates examining 12 Jeffrey C. Fuhrer and Scott Schuh the interactions among many economic variables in order to obtain a clear picture of the role of any one of them in economic ﬂuctuations.
Sims employs a methodology that allows each of six variables (industrial production, consumer prices, currency, a monetary aggregate, the discount rate, and commodity prices) to respond to lags of the other variables, and to the contemporaneous values of some of the other variables. The restrictions on the contemporaneous interactions among variables reﬂect common-sense notions about policy, goods market, and ﬁnancial market behavior. Monetary policy-induced interest rate changes affect prices, output, and monetary aggregates only with a one-month lag;
monetary policy responds to output and prices only with a lag, reﬂecting data availability; and commodity prices respond to everything contemporaneously, reﬂecting their auction-market, ﬂexible nature.
This simple model is estimated on monthly data for the postwar years 1948 to 1997. Sims uses the model to show that most of the variation in the Fed’s discount rate represents systematic policy responses rather than unanticipated shifts in policy. The discount rate responds primarily to movements in production, commodity prices, and M1. These three determinants of interest rate movements in turn cause the largest increases in CPI inﬂation, suggesting that the Fed responds to these as signals of future inﬂationary pressures.
When Sims estimates this same model on the interwar period from 1919 to 1939, he ﬁnds similarities but also some important differences in monetary policy responses and inﬂuences. One key difference is that the effect of interest rate changes in the early period is roughly double the effect in the later period. On the other hand, monetary policy in the early period appears to be more accommodative toward unanticipated increases in output, raising the discount rate less in response to output and thereby allowing greater inﬂation in commodity and in ﬁnal goods prices.
Interestingly, the model shows that when depositors’ worries caused a rush into currency in the interwar period, the Fed typically raised the discount rate, accelerating the shrinkage of money.
This ﬁrst set of exercises establishes that the systematic responses of policy to output and prices represent the dominant source of interest rate ﬂuctuations in Sims’s model, and that these interest rate movements are likely the most important source of policy’s effects on the rest of the economy. Noting that economic ﬂuctuations have been smaller in the
postwar period, Sims proposes using his model to answer a key question:
whether better systematic monetary policy is responsible for the improved economic performance of the postwar period.
To answer this question, Sims transplants the estimated monetary policy equation for one period into the other period, then observes the estimated behavior of output, prices, and monetary aggregates under this counterfactual monetary regime. The results from these exercises are remarkable. In the ﬁrst variant, the (estimated average) policy judgment
BEYOND SHOCKS: WHAT CAUSES BUSINESS CYCLES? AN OVERVIEW 13of Burns, Volcker, and Greenspan is imposed on the 1920s and 1930s.
Overall, Sims ﬁnds the outcomes—particularly the Great Depression— would have been little changed by this more responsive postwar policy.
The drop in production from 1929 to 1933 is “completely unaffected by the altered monetary policy.” Postwar policy would have made the 1920-21 and 1929-33 deﬂations less severe, but not by much. The upheaval of the 1920s and 1930s would have been the same, even if modern monetary policymakers had been at the reins. Sims notes that his methodology leaves the banking runs, panics, and currency speculations that plagued the Depression era as unexplained non-monetary shocks. To the extent that a persistent commitment to monetary ease would have alleviated such disruptions, the drop in output might have been less severe, he suggests.
The effects of substituting interwar monetary policy into the postwar economy are qualitatively the same. Even though the discount rate responds much more slowly to the postwar economic ﬂuctuations, resulting in a markedly different interest rate pattern, the inﬂuence of this altered policy on industrial production and consumer prices is quite small at business cycle frequencies. The implications for output and inﬂation at longer horizons are what one would expect with a more accommodative policy: Output and inﬂation both rise higher in the 1970s, resulting in a larger recession in the 1980s, although Sims is careful to point out that these ﬁndings may well be statistically unreliable. Overall, he reaches the startling conclusion that “the size and timing of postwar U.S. recessions had little to do with either shocks to monetary policy or its systematic component.” Lawrence Christiano focuses on Sims’s surprising conclusion that monetary policy played little or no role in the Great Depression. He disagrees with the methodology that Sims uses to reach this conclusion, but upon employing what he considers a superior method, he conﬁrms Sims’s results.
One criticism of Sims’s methodology revolves around the assumption that private agents behaved the same in the postwar period after the creation of the Federal Deposit Insurance Corporation (FDIC) as they did during the interwar period prior to the FDIC. Christiano suggests that the frequency with which interwar depositors converted deposits to currency at the slightest sign of bad news, in contrast to the virtual absence of such bank runs in the postwar period, suggests that the presence of the FDIC fundamentally changed private agents’ behavior. In particular, they may have viewed the commitment of Federal Reserve policy to maintain banking system liquidity quite differently in the postwar period, and in a way that cannot be captured by the simple “reaction functions” or interest rate equations in Sims’s analysis.
The more important ﬂaw in Sims’s analysis, according to Christiano, is the characterization of the postwar monetary policy rule. Under this 14 Jeffrey C. Fuhrer and Scott Schuh rule, after all, the Fed would have contracted the money supply by 30 percent in the 1930s. Christiano cannot conceive of a sensible policymaker who would pursue a contractionary monetary policy during a widely recognized, worldwide depression. So Christiano proposes instead to use a monetary policy equation that keeps money (M1) from falling during the episode.
Using this more plausible counterfactual policy in Sims’s model for the interwar period, Christiano ﬁnds that a stable M1 path for the early 1930s would have prevented the dramatic price declines that actually occurred. Surprisingly, however, even under the more realistic policy response, which implies a more realistic path of money growth, “the basic course of the Great Depression would not have been much different,” as shown by the similarity between the path of output in Christiano’s simulation and the actual path of output.
Benjamin Friedman is also skeptical of the empirical results developed in Sims’s paper, stating: “If the model he presents has succeeded in identifying Federal Reserve actions and measuring their economic effects, these ﬁndings should force us to reconsider many aspects of economics and economic policy.” Friedman ﬁnds troubling Sims’s result that postwar monetary policy would not have signiﬁcantly altered the course of the Great Depression, and he views as even more problematic the ﬁnding that Depression-era monetary policy would have worked just the same in the postwar period as did actual policy. Friedman notes that the general price level was approximately the same at the onset of World War II as at the onset of the Civil War, while prices since that time have risen approximately tenfold. That the monetary policy that delivered the interwar deﬂation is the same one that delivered the “historically unprecedented phenomenon of a half century of sustained inﬂation” would make inﬂation, even over periods of several decades, never and nowhere a monetary phenomenon.
Friedman suggests that Sims’s model delivers its surprising results because it fails to adequately identify the Fed’s monetary policy actions or the effects of those actions on the macroeconomy. If so, then the model’s “implied irrelevance of monetary policy” for the postwar inﬂation translates further into irrelevance for assessing monetary policy’s role in causing or cushioning business cycles. One indication that Sims’s postwar policy rule does not accurately represent Fed actions, Friedman argues, is the difference between the Sims model’s policy prescriptions for the Depression era and John Taylor’s policy rule prescriptions for the same period. Friedman ﬁnds that Taylor’s rule would imply nominal interest rates “an order of magnitude more negative than what Sims reports,” casting some doubt on how well Sims’s policy rule reﬂects all of postwar Fed behavior.
Finally, Friedman notes that the assumption that Fed policy can be characterized by one unchanging rule over the entire postwar period is
BEYOND SHOCKS: WHAT CAUSES BUSINESS CYCLES? AN OVERVIEW 15implausible. He asks, “Are we really to equate Paul Volcker’s tough stance against inﬂation with the see-no-evil regime of Arthur Burns?” While Friedman recognizes that Sims tests for a shift in monetary policy in 1979, Sims does so by testing for a shift in all 279 of his model’s parameters. Friedman notes that Sims could have more narrowly focused this test to detect only shifts in the parameters that summarize monetary policy.
FINANCIAL MARKETS AND BUSINESS CYCLES: