Real Business Cycle Models
While Keynesians were trying to challenge Lucas, others were trying to implement the research programme he had initiated. Kydland and Prescott’s “Time to build and aggregate fluctuations” (1982) and John Long and Charles Plosser’s “Real business cycles” (1983) are the two papers which started the real business cycle line of research.
Both tried to model business fluctuations as the result of real shocks to the economy (rather than monetary shocks, as in Lucas’s model). Kydland and Prescott’s paper had the additional feature of wanting to move from the model to the facts, so inaugurating a new methodology by taking the neoclassical growth model to the computer.Kydland and Prescott’s model is, like Lucas’s, neo-Walrasian. The equilibrium discipline, rational expectations, a dynamic-stochastic environment, and a central role for intertemporal substitution are all present in both types of model. But there are also striking differences. First, Kydland and Prescott shifted towards real technology shocks. Second, they abandoned the imperfect information line of research. Third, and most important, Kydland and Prescott’s work was quantitative. In Michael Woodford’s words:
The real business cycle literature offered a new methodology, both for theoretical analysis and for empirical testing... It showed how such models [of the Lucas type] could be made quantitative, emphasising the assignment of realistic numerical parameter values and the computation of numerical solutions to the equations of the model, rather than being content with merely qualitative conclusions derived from more general assumptions. (Woodford 1999: 25-6, original emphasis)
Woodford was right. However, merely asserting that a qualitative model was transformed into a quantitative one may fail to convey the full measure of the change. While models a la Lucas could recruit only a tiny fraction of the macroeconomic profession, Kydland and Prescott were able to devise a research programme that became the bread and butter approach for legions of macroeconomists, both top-notch and average, for decades to come.
This is the sign of a successful revolution.The aim of Kydland and Prescott’s 1982 model was to show that economic fluctuations could be explained as a consequence of economic agents’ optimising adjustment to exogenous technological shocks. Their starting point was Ramsey’s (1928) and Cass’s (1965) models of optimal growth, which were extended to a stochastic economy by Brock and Mirman (1972). Ramsey studied the intertemporal optimising programme of a representative agent over an infinite horizon, subject to a budget and a technology constraint calculated by a benevolent and omniscient planner.
To the outside observer, what is striking in Kydland and Prescott‘s endeavour is the contrast between the model they build and its avowed purpose, to shed light on the development of the US economy from 1950 to 1975. Their model economy is summarised in one utility function and one production function. The production function is subject to stochastic technology shocks. The variables considered are, for production, capital, the level of employment (number of hours worked; not the number of people employed as opposed to those who are unemployed) and productivity, and for household preferences, consumption and investment. Two additional variables are involved: the hourly real wage and the real interest rate. Kydland and Prescott used two sources to parameterise the functional forms of the models: first, steady state conditions and, second, calibration. Calibration, a technique borrowed from computational general equilibrium analysis, consists of assigning values to the model’s parameters by using information from panels, national accounts and other data banks. If such data are unavailable, the model-builder ascribes values based on theoretical reasoning.
The validation of the model occurs by comparing the moments (volatilities, correlations and auto-correlations) that summarise the actual experience of the US economy with the equivalent moments from the model economy.
The model succeeds if the simulation mimics the empirical observations. To a large extent, this is true for Kydland and Prescott’s model, which satisfactorily reproduces both the low variability of consumption and the high variability of investment. The same is true for the pro-cyclical character and persistence of most of the variables considered. However, as readily admitted by the authors, the model is wanting on two scores. It is unable to account for the variation in hours worked. In the real-world data, these hours are closely correlated with output, but they vary significantly less in the model. Another weakness concerns changes in the wage rate and the interest rate; in the model, these are pro-cyclical, but in reality wages are only weakly pro-cyclical (almost a-cyclical) while the interest rate is anti-cyclical.Kydland and Prescott proved able to achieve their aim of constructing a model mimicking several important empirical traits of the fluctuations in the US economy over a quarter of a century, on the basis of the most rudimentary possible model. Before their paper appeared, the general opinion was that such an enterprise was impossible. Nevertheless, a large number of criticisms were levelled at Kydland and Prescott’s model. Answering these led to a series of wide-ranging improvements, which we cannot enter into here. With time, Kydland and Prescott’s initial real business cycle model grew into a simplified canonical model, the twin advantages of which were its parsimony and the purposes which it can serve.
New developments resulted from attempts to reply to the early criticisms, which pointed out insufficiencies and inconsistencies. New stylised facts were integrated into its successors. This led to a growth in the number and type of shocks considered. For example, in order to improve upon the anomalous correlation between productivity and hours worked, Christiano and Eichenbaum (1992) introduced a shock related to government consumption expenditures, which had a negative wealth effects on households. Another striking defect of the early real business cycle models was their lack of consideration of money. Kydland and Prescott had argued that monetary shocks played only a minor role in explaining business fluctuations. Accepting this conclusion was one thing, but the nagging stylized fact of the inverse evolution of the interest rate, on the one hand, and of inflation and output, on the other, was another. Monetary policy had thus to re-enter the picture. Woodford’s (2003) book, Interest and Prices, blazed the trail.