Understanding Real Business Cycles

Plosser argues the Keynesian framework’s “essential flaw” was “the absence of a consistent foundation based on the choice theoretic framework of microeconomics,” and credits Friedman (1968) and Lucas (1976) with demonstrating this: Friedman showed basic microeconomic homogeneity implies the long-run Phillips curve must be vertical, directly conflicting with the Keynesian inflation-output tradeoff, while Lucas argued that Keynesian behavioral relations could not correctly evaluate policy changes because they ignored how expectations about future policy alter current decisions (Introduction). Plosser extends this critique to the dynamic elements of Keynesian models, arguing their accelerator and partial-adjustment mechanisms “did not arise from any choice theoretic framework of maximization” and so amounted to arbitrary behavioral rules rather than principled dynamics.

Quoting Hicks (1933), Plosser frames the research program’s logical necessity: understanding how a perfectly functioning dynamic economy would behave is a precondition for attributing any portion of observed fluctuations to market failure, since “it is logically impossible to attribute an important portion of fluctuations to market failure without an understanding of the sorts of fluctuations that would be observed in the absence of the hypothesized market failure.” He argues Keynesian models, by simply positing market failures (unexploited gains from trade) from the outset, could never supply this understanding, and that RBC models – built from growth theory – represent the first attempt to actually characterize that idealized equilibrium state.

The model has infinitely lived, identical agents choosing consumption, leisure, and investment to maximize expected lifetime utility subject to a constant-returns production technology (using capital, which depreciates, and labor) that is subject to technology shocks known with certainty (perfect foresight), with the usual nonnegativity and resource constraints (The Basic Real Business Cycle Framework). Citing Debreu (1954) and Prescott and Lucas (1972), Plosser notes the competitive equilibrium of this many-agent economy is identical to the planning problem of a single representative agent (“Robinson Crusoe”), whose first-order conditions pin down the competitive real wage and real interest rate and yield a standard consumption-leisure tradeoff and intertemporal (Euler) consumption condition.

Plosser stresses that the basic model, without productivity disturbances, predicts convergence to a constant steady state, which cannot explain the U.S. economy’s roughly 1.5% average annual per capita growth from 1954-1985; following Solow (1957), who attributed about 85% of early-20th-century U.S. per capita growth to technological change rather than capital per worker, Plosser treats technology shocks as doing double duty – driving both long-run growth and short-run fluctuations. Citing Nelson and Plosser’s (1982) finding that productivity (and output) behave like a random walk, he argues this “nonstationary” property means shocks to productivity are permanent, so detrending with a deterministic trend “would be econometrically incorrect and may be quite misleading,” and that King, Plosser, and Rebelo (1988b) and King, Plosser, Stock, and Watson (1987) show the random-walk-technology model implies output, consumption, and investment share one common stochastic trend, consistent with the data. Quoting Hicks (1965) directly, he rejects the idea that trend and cycle must be driven by different economic forces just because they are a “statistical distinction.”

Plosser shows that a one-time, one percent permanent increase in labor productivity eventually raises capital, consumption, output, and investment each by one percent in the new steady state, but reaching that steady state requires a transition path in which investment rises sharply and then tapers off, consumption and output rise gradually, and work effort is “temporarily high along the transition path” because higher productivity (from the still-rising desired capital stock) and higher real interest rates (inducing intertemporal substitution toward current work) outweigh the wealth effect that would otherwise discourage work (Economic Growth and Business Cycles). Plosser’s key point is that “the responses, and thus the fluctuations that are present in the model, are the result of the same factors that generate economic growth,” so the model provides, in his words, “an integrated approach to the theory of growth and fluctuations.”

Using a logarithmic-utility, Cobb-Douglas-technology calibration (labor share 0.58, depreciation 10% per year, discount factor chosen to yield a 6.5% average real return to capital) driven by a Solow-residual-based productivity series, Plosser reports the model’s simulated series correlate with the actual U.S. series at between 0.52 (real wages) and 0.87 (output), and that the model “generates sample means that are very close to the data for output, consumption and hours,” with investment somewhat too low and wages somewhat too high in the model. He calls the fit “very surprising” to many economists, “given that most macroeconomic research over the past 50 years stressed the importance of” government, money, market failures, adjustment costs, or heterogeneous agents, none of which this bare-bones model includes.

He states plainly that “the major discrepancy appears to be that in the model the growth rate of hours has a standard deviation that is less than one-half of that in the data” (Real Business Cycles and the 1954-1985 U.S. Economy) – the model cannot generate nearly enough volatility in hours worked relative to the volatility actually observed, even though it otherwise reproduces the broad comovement and relative-volatility rankings of output, consumption, and investment reasonably well.

Drawing on Barro (1981) and Hall (1980), Plosser describes two channels by which temporary increases in government purchases raise output in RBC models: a negative wealth effect (raising work effort to restore consumption) and intertemporal substitution (prompting more work and less consumption/investment today in anticipation of the temporary spending), and cites Baxter and King’s (1988) finding that more persistent government-spending increases have larger output “multipliers” than temporary ones, while temporary purchases hit investment harder. Summarizing King, Plosser, and Rebelo (1988b), he notes that under a period-by-period balanced budget financed by an output tax, a positive productivity shock requires the tax rate to fall to maintain balance, which reinforces the shock’s effect on after-tax productivity and makes work effort and investment more volatile than in the no-tax case (Government Policies and Suboptimal Equilibrium) – and stresses that once distortionary taxes are present, the model’s competitive equilibrium is generally no longer Pareto optimal, since removing the distortion would raise welfare.

Plosser concedes that “real business cycle research has focused almost exclusively on models with no role for money,” which “for some economists…borders on blasphemy,” and states his own view, shared by “many other real business cycle researchers,” that “the role of money in an equilibrium theory of growth and fluctuations is not well understood and thus remains an open issue,” citing early attempts by King and Plosser (1984), Kydland (1987), Eichenbaum and Singleton (1986), and Cooley and Hansen (1988) without endorsing any as settled. He argues, however, that understanding real fluctuations in the moneyless baseline model first is a precondition for ever measuring money’s incremental importance.

On hours, he surveys Kydland and Prescott’s (1982) non-time-separable preferences (letting past leisure affect current utility, raising effective intertemporal substitutability), Rogerson’s (1988) and Hansen’s (1985) labor-supply indivisibilities (forcing an all-or-nothing work/not-work margin, which raises aggregate hours volatility without requiring high individual labor-supply elasticities), and heterogeneous-agent extensions (Cho and Rogerson 1988; Kydland 1984; King, Plosser, and Rebelo 1988b; Rebelo 1987) that can reveal “important downward biases in estimates of aggregate labor supply elasticity” from pooling agents of different skill. On growth, he discusses endogenous-growth extensions (Uzawa 1965; Romer 1986; Lucas 1988; King and Rebelo 1986, 1988) in which eliminating diminishing returns to reproducible factors lets “purely temporary productivity shifts…have permanent effects on the level of economic activity,” since higher output raises resources devoted to human-capital production, altering the growth path itself – a structure he notes is also consistent with Nelson and Plosser’s (1982) stochastic-trend evidence.

He describes calibration, “made popular in this literature by Kydland and Prescott (1982),” as choosing many of a model’s parameters from other, independent empirical studies to limit the model’s remaining degrees of freedom, then checking whether the model’s implied moments (means, variances, autocorrelations, cross-correlations) match the data, with any remaining free parameters tuned to improve that match – a different strategy from traditional structural estimation and hypothesis testing. He is explicit that “the appropriate empirical strategy for investigating the class of models discussed in this paper remains an open area of research,” and that “real business cycle models will have to face and pass more stringent empirical tests than they have to date” (The Real Business Cycle Research Agenda).

He states candidly that “real business cycle theory is still in its infancy and thus remains an incomplete theory of the business cycle,” but argues its central achievement so far is to have “changed the standard by which macroeconomic theories are judged” by demonstrating that fully optimizing, market-clearing equilibrium models “are not necessarily inconsistent with many characteristics attributed to the business cycle” – dynamic behavior “heretofore thought to be incompatible with any notion of equilibrium.” His closing assessment is measured rather than triumphalist: “While the promise is great, much work remains before economists have a real understanding of business cycles.”