Objections

The real-business-cycle approach to analyzing economic fluctuations is controversial. Four objections have received particular attention.

The first criticism concerns the technology shocks. Real-business-cycle models posit technology shocks with a standard deviation of about 1 percent each quarter. It seems likely that such large technological innovations would often be readily apparent. Yet it is usually difficult to identify specific innovations associated with the large quarter-to-quarter swings in the Solow residual.

See Abraham and Katz (1986); Murphy and Topel (1987a); Lougani, Rush, and Tave (1990); Davis and Haltiwanger (1990); and Brainard and Cutler (1993).

Some of the other factors that have been incorporated into the models include: lags in the investment process, or time-to-build (Kydland and Prescott, 1992); non-time-separable utility (so that instantaneous utility at t does not depend just on c, and {,) (Kydland and Prescott, 1982); home production (Benhabib, Rogerson, and Wright, 1991); roles for government-provided goods and capital in utility and production (for example, Christiano and Eichenbaum, 1992a, and Baxter and King, 1993); multiple countries (for example, Baxter and Crucinl, 1993); embodied technological change and variable capital utilization (Greenwood, Hercowltz, and Huffman, 1988); and labor hoarding (Burnside, Eichenbaum, and Rebelo, 1993).

Most of these objections are raised by Summers (1986a) and Mankiw (1989).

Another important extension of real models of fluctuations is the inclusion of multiple sectors and sector-specific shocks. Long and Plosser (1983) develop a multisector model similar to the model of Section 4.5 and investigate its implications for the transmission of shocks among sectors. Lilien (1982), building on the theoretical work of Lucas and Prescott (1974), proposes a distinct mechanism through which sectoral technology or relative-demand shocks can cause employment fluctuations. The basic idea is that if the reallocation of labor u-ross sectors is time-consuming, employment falls more rapidly in the sectors suffering negative shocks than it rises in the sectors facing favorable shocks. As a result, sector-specific shocks cause temporary increases in unemployment. Lilien found that a simple measure of the size of sector-specific disturbances appeared to account for a large fraction of the variation in aggregate employment. Subsequent research, however, has shown that liliens original measure is flawed and that his results are almost stnely too strong. This work has not reached any firm conclusions concerning the contribution of sectoral shocks to fluctuations or to average unemployment, however.

These are only a few of a large number of extensions of real-business-cycle models. At this point, these models are an active and rapidly evolving subject of research.

"As Hall explains, oil-price movements should not affect productivity once oils role in production is accounted for.

More importantly, there is significant evidence that short-run variations in the Solow residual reflect more than changes in the pace of technological innovation. As described above, Bernanke and Parkinson find that the Solow residua] moves just as much with output in the Great Depression as it does m the postwar period, even though it seems unlikely that the depression was caused by technological regress. This is just one example of a broader pattern. Mankiw (1989) shows that the Solow residual behaves similarly in the World War II boom-which was also probably not due to technology shocks-as it does during other periods. Hall (1988a) demonstrates that movements in the Solow residual are correlated with the political party of the President, changes in military purchases, and oil-price movements; yet none of these variables seem likely to affect technology significantly in the short run.*" If true technology shocks are considerably smaller than the variation in the Solow residual suggests, real-business-cycle models ability to account for fluctuations is much smaller than the calibration exercise of the previous section implies.

The second criticism of these models concerns not their shocks but one of their central propagation mechanisms, intertemporal substitution in labor supply. Variations in the incentives to work in different periods drive employment fluctuations in the models. Thus a significant willingness to substitute labor supply between periods is needed for important employment fluctuations in the models. Microeconomic studies, however, have had little success in detecting significant intertemporal elasticities of substitution m labor supply. The results of Ball (1990) are typical (see also Altonji, 1986, and Card, 1991). Ball divides the workers in a panel data set into those who say that their labor-supply decisions are constrained by the availability of jobs or hours and those who say they are unconstrained. He then investigates the predictions of a model where fluctuations in work are driven by intertemporal optimization for each of the two groups. The results are consistent with what the workers report: the model is rejected for the ones who say they are constrained, and not rejected for the ones who say they are unconstrained. More tellingly, for the workers who say they are unconstrained, the estimated labor-supply elasticities in response to transitory wage changes are small. Thus Balls results suggest that fluctuations in overall labor supply are driven primarily by forces other than intertemporal substitution.

The third criticism concerns real-business-cycle models omission of monetary disturbances. A central feature of the models is that fluctuations are due to real rather than monetary shocks. Yet, as described at the beginning of the chapter, Keynesian macroeconomists argue that monetary disturbances are central to imderstanding aggregate fluctuations.

"Section 5.6 discusses some of the empirical evidence concerning the effects of monetary shocks.

The argument that follows is due to Matthew Shapiro.

If monetary shocks have important real effects, this would mean more than just that real-busmess-cycle models omit one source of output movements. As described in the next two chapters, the leading candidate explanations of real effects of monetary changes rest on incomplete adjustment of nominal prices or wages. But, as we will see there, incomplete nominal adjustment implies a new channel through which other disturbances, such as changes in government purchases, have real effects. We will also see that incomplete nominal adjustirent is most likely to arise when labor, credit, and goods markets depart significantly from the competitive assumptions of real-business-cycle theory. Thus if there is substantial monetary non-neutrality, many of the central features of the real-business-cycle approach might need to be abandoned or greatly modified.*

The final criticism of the approach concerns its empirical philosophy rather than the specifics of its models. As described in Section 4.9, the empirical fit of real-business-cycle models is evaluated mainly through calibration exercises. Although calibration has some advantages, it has disadvantages as well. First, as our discussion of extensions of the basic model suggests, there are now many potential ingredients for real-business-cyclc models, and a large number of potential ways of combining them. Moreover, not all of the functional forms and parameter values of these ingredients are pinned down by microeconomic evidence. Thus the models have some flexibility in matching characteristics of the data. How much flexibihty they have is, at this point, not known. Thus we do not know how informative it is that there are real-business-cycle models that can match important moments of the data relatively well. Nor, because the models are generally not tested against alternattves, do we know whether there are other, perhaps completely different, models that can match the moments just as well.

Second, given the state of economic knowledge, it is not clear that matching the major moments of the data should be viewed as a desirable feature of a model.* Even the most complicated real-business-cycle models are grossly simplified descriptions of reality. They generally omit such considerations as heterogeneity in goods, capital, and labor; adjustment costs; and departures from simple functional forms. And, as later chapters describe, consumption and investment behavior and the characteristics of financial, labor, and goods markets may depart significantly from the simple assumptions made in real-business-cycle models. It would be remarkable if none of these potential comphcations (or any others) have quantitatively important effects on the properties of fluctuations. But given this, it is hard to see how the fact that real-business-cycle models do or do not match aggregate data is informative about their overall usefulness.