Some Skeptical Observations on Real Business Cycle Theory

Questions & answers

Q1. What is the occasion for this piece, and what is its overall thesis?

Summers states his thesis bluntly: “My view is that real business cycle models of the type urged on us by Prescott have nothing to do with the business cycle phenomena observed in the United States or other capitalist economies. Nothing in Prescott’s papers or those he references is convincing evidence to the contrary” (p. 23-24). The piece is explicitly a response to “the appearance of Ed Prescott’s stimulating paper, ‘Theory Ahead of Business Cycle Measurement,’” which Summers calls “brilliant in highlighting the appeal of real business cycle theories and making clear the assumptions they require,” while objecting that Prescott “does not make much effort at caution in judging the potential of the real business cycle paradigm” (p. 23). A footnote notes an earlier version of the remarks was presented at the July 25, 1986 NBER Economic Fluctuations Group meeting (p. 23).

Q2. What does Summers say real business cycle theories claim, and why does he regard the stakes as high for the profession?

Summers characterizes the common view of these theories as modeling “the economy… as a floating Walrasian equilibrium, buffeted by productivity shocks,” which he says implies “that monetary policies have no effect on real activity, that fiscal policies influence the economy only through their incentive effects, and that economic fluctuations are caused entirely by supply rather than demand shocks” (p. 23). He frames the stakes sharply, writing that if these theories are correct they imply “the macroeconomics developed in the wake of the Keynesian Revolution is well confined to the ashbin of history,” and that, per “enthusiastic proponents,” Keynesian economics and astrology “are in many ways similar” in lacking scientific support (p. 23) — a comparison Summers uses rhetorically rather than as an argument in itself, cautioning separately (via the Ptolemaic astronomy and Lamarckian biology examples) that “extremely bad theories can predict remarkably well,” so a model’s ability to mimic facts does not establish that it is “close to right” (p. 24).

Q3. What is Summers’s first objection, concerning the model’s parameter values?

Summers’s first fundamental objection is that “Prescott’s claim to have parameterized the model on the basis of well-established microeconomic and long-run information is not sustainable.” He gives two examples: Prescott’s one-third figure for the household time share devoted to market activity versus the one-sixth average value “over the past 30 years” that Eichenbaum, Hansen, and Singleton (1986) report, which Summers says he cannot find support for in Prescott’s own cited source (Ghez and Becker 1975); and Prescott’s 4 percent average real interest rate assumption versus the roughly 1 percent that Summers says the 30-year period Prescott studies “in fact averaged” (p. 24). He adds that Prescott “cites no direct microeconomic evidence” for the intertemporal elasticity of substitution in labor supply central to the model, relying instead on “a rather selective reading of” consumption-substitution evidence, when “essentially all the available evidence” he is aware of (citing Altonji 1982; Mankiw, Rotemberg, and Summers 1985; Eichenbaum, Hansen, and Singleton 1986) “suggests only a minimal response of labor to transitory wage changes” (p. 24). He also notes Japan’s roughly 8 percent real wage growth over 30 years has not produced the large labor-supply decline the model’s utility function would predict (p. 24), concluding with the image that “the image of a big loose tent flapping in the wind comes to mind” (p. 24).

Q4. What is Summers’s second objection, concerning the existence and source of “technology shocks”?

Summers’s second objection is “the absence of any independent corroborating evidence for the existence of what [Prescott] calls technological shocks,” which he calls “obviously crucial since Prescott treats technological shocks as the only driving force behind cyclical fluctuations” (p. 24). He argues it is “hard to find direct evidence” of large technological shocks, citing Berndt’s (1980, p. 85) finding that energy price/quantity variation after 1973 “do[es] not appear to have had a significant direct role” in the 1973-77 U.S. manufacturing productivity slowdown, and notes Prescott cannot point to a specific technology shock accounting for the 1982 recession, “the most serious downturn of the postwar period” (p. 24-25). Summers suggests instead that “the vast majority of what Prescott labels technology shocks are in fact the observable concomitants of labor hoarding and other behavior which Prescott does not allow in his model,” citing Fay and Medoff’s (1985) survey of roughly 170 manufacturing plants, which found the typical plant paid for 8 percent more blue-collar hours than needed in the trough quarter of its most recent downturn, of which “4 percent of the blue-collar hours paid for were hoarded” after accounting for other useful work performed (p. 25).

Q5. What is Summers’s third objection, about the model being “price-free”?

Summers’s third objection is that Prescott “does price-free economic analysis,” illustrated by the analogy of an analyst who ignores the price of ketchup when explaining fluctuations in ketchup purchases — a choice Summers says “would considerably increase the analyst’s freedom in accounting for fluctuations” and make it “impossible to distinguish supply shocks from demand shocks” (p. 25). He writes that although “real wages, interest rates, and returns to capital are central variables in [Prescott’s] model, he never looks at any data on them except for his misconstrual of the average real interest rate,” and that where others have confronted RBC-type models with price data, the results have been “dismal,” with “no evidence to support any of the price effects predicted by the model” (p. 25). He specifically notes the model does not address the Barro and King (1982) finding that consumption and leisure move in opposite directions over the cycle with “no apparent procyclicality of real wages,” and observes that Prescott’s own work with Mehra (1985) shows the asset-pricing implications of this class of model are “decisively rejected by nearly 100 years of historical experience” (p. 25).

Q6. What is Summers’s fourth objection, about breakdowns of the exchange mechanism?

Summers’s fourth objection is that Prescott’s framework “ignores the fact that partial breakdowns in the exchange mechanism are almost surely dominant factors in cyclical fluctuations,” illustrated by the 1929-33 U.S. GNP decline of 50 percent and the post-1970 rise in European unemployment to “more than fivefold in many countries” even as European total factor productivity grew “more than twice as rapidly” as in the U.S. (p. 25-26). He calls it “simply absurd” to describe the 1929-33 period as “constrained Pareto optimal given the technological decline that took place,” arguing instead that “what happened was a failure of the exchange mechanism” that “no model… of a long-lived Robinson Crusoe” can confront (p. 26). He explicitly rejects the traditional Keynesian story that sticky prices alone explain this failure, calling it “far from being a satisfactory story,” but maintains that “a model that embodies exchange is a minimum prerequisite for a serious theory of economic downturns” (p. 26). As a candidate mechanism, he points to credit-market breakdowns, citing Eckstein and Sinai’s (1986) account of each postwar recession as caused by “a credit crunch in an effort to control inflation” as “compelling” (p. 25-26).

Q7. How does the paper conclude, and what research direction does Summers recommend instead?

Summers concludes that “economists are much better at analyzing the optimal response of a single economic agent to changing conditions than they are at analyzing the equilibria that will result when diverse agents interact,” which he says helps explain the profession’s difficulty “controlling, predicting, or even explaining economic fluctuations” (p. 26). He states that “improvement in the track record of macroeconomics will require the development of theories that can explain why exchange sometimes works well and other times breaks down,” and warns that “nothing could be more counterproductive in this regard than a lengthy professional detour into the analysis of stochastic Robinson Crusoes” (p. 26) — a direct dismissal of the single-agent, no-exchange-failure structure of the Prescott-style RBC model that is the object of his critique throughout.