Uneasy Money

In 1978 Robert Lucas and Thomas Sargent launched a famous attack on Keynes and Keynesian economics, which they viewed as having been discredited by the confluence of high inflation and high unemployment in the 1970s. They also expressed optimistism that an equilibrium approach to business-cycle modeling would succeed in replicating reasonably well the observed time-series variables relating to output and employment. In particular they posited that a model subjected to an unexpected monetary shock causing an immediate downturn from an equilibrium time path would be followed by a gradual reversion to that time path, thereby capturing the main stylized facts of historical business cycles. Their optimism was disappointed, because the model that Lucas had developed, based on an informational imperfection preventing agents from distinguishing immediately between real and nominal price changes, could not account for downturns because the informational imperfection assumed by Lucas could not account for the typical multi-period duration of business-cycle downturns.

It was this empirical anomaly in Lucas’s monetary business-cycle model that prompted Kydland and Prescott to construct their real-business cycle model. Lucas warmly welcomed their contribution, the abandonment of the monetary-theoretical motivation that Lucas had inherited from his academic training at Chicago being a small price to pay for the advancement of the larger research agenda derived from his methodological imperatives.

The real-business cycle variant of the Lucasian research program rested on two empirical pillars: (1) the identification of technology shocks with deviations, as measured by the Solow residual, from the trend rate of increase in total factor productivity, positive residuals corresponding to positive shocks and negative residuals corresponding to negative shocks; and (2) estimates of elasticities of intertemporal rates of labor substitution.

Positive productivity shocks induce wage increases, and negative shocks induce wage decreases. Responding to the shifts in wages, presumed to be temporary, workers increase the amount of labor supplied in response to above-trend increases in wages and decrease the amount of labor supplied in response to below-trend increases in wages. The higher the elasticity of intertemporal labor substitution, the greater the supply response to a given deviation of actual wages from the expected trend rate of increase in wages. Real-business-cycle theorists used calibration techniques to obtain estimates labor-supply elasticities from microeconomic studies.

The real-business-cycle variant of the Lucasian research program embraced all the dubious methodological precepts of its parent while adding further dubious practices of its own. Most problematic, of course, is the methodological insistence that equilibrium is necessarily and continuously maintained, which is possible only if all agents correctly anticipate future prices and wages. If equilibrium is not continuously maintained, then Solow residuals may capture not productivity shocks, but, depending on their sign, either movements away from, or toward, equilibrium. In disequilibrium, labor and capital may be held idle by firms in anticipation of subsequent increases in output, so that measured productivity does not reflect the state of technology, but the inherent inefficiency of unemployment resulting from coordination failure, a contingency explicitly deemed by Lucasian methodology to be off limits.

Such ad hocery is generally frowned upon by scientists. Ad hoc assumptions are not always unscientific or unproductive, as famously exemplified by the discovery of Neptune. But in the latter case, the ad hoc assumption was subject to empirical testing; Neptune might not have been there waiting to be discovered. But no independent test of the presence or absence of a technology shock, aside from the Solow residual itself, is available. Even this situation might be tolerable, if Lucasian methodology permitted one to inquire whether the world or an economy might not be in an equilibrium state. But Lucasian methodology forbids such an inquiry.

The use of calibration to estimate intertemporal labor-supply elasticities from microeconomic studies are also extremely dubious, because microeconomic estimates of labor-supply elasticities are typically made under conditions approximating equilibrium, when workers have some flexibility in choosing whether to work more or less in the present or in the future. Those are not the conditions in which workers find themselves in periods of high aggregate unemployment, and are, therefore, not confident that they will retain their jobs in the present and near future, or, if they lose their jobs, that they will succeed in finding another job at an acceptable wage. The calibrated estimates of labor-supply elasticity are, for exactly the reasons identified in the Lucas Critique, unreliable for use in replicating time series.

An early real-business-cycle theorist Charles Plosser (“Understanding Real Business Cycles”) responded to criticisms of the RBC techniques as follows:

If the measured technological shocks are poor estimates (that is, they are confounded by other factors such as “demand” shocks, preference shocks or change in government policies, and so on) then feeding these values into our real business cycle model should result in poor predictions for the behavior of consumption, investment, hours worked, wages and output.

Plosser’s response ignores the question-begging nature of the RBC model; the supposed productivity shocks that cause cyclical fluctuations in the model are identified by the very time series that the model purports to explain. Nor does calibration provide clear and unambiguous estimates that the modeler can transfer without exercising discretion about which studies and which values to insert into an RBC model. Plosser’s defense of RBC is not so very different from the sort of defense made on behalf of the highly accurate epicyclical replications of observed planetary movements, replications that were based largely on the ingenuity and diligence of the epicyclist.

Eventually, the methodological prohibitions against heliocentrism were overcome. Perhaps, one day, the methodological prohibitions against non-reductionist macroeconomic theories will also be overcome.

Lucasian macroeconomics gained not only ascendance, but dominance, on the basis of  conceptual and methodological misunderstandings. The continued dominance of the offspring of the early Lucasian theories has been portrayed as a scientific advance by Lucas and his followers. In fact, the theories and the supposed methodological imperatives by which they have been justified are scientifically suspect because they rely on circular, question-begging arguments and reject alternative theories based on specious reductionist arguments.