Department of Economics, UCSD

In this paper we re-examine the recent evidence that technology shocks do not produce business cycle patterns in the data. We first extend Gali's (1999) work, which uses long-run restrictions to identify technology shocks, by examining whether the identified shocks can be plausibly interpreted as technology shocks. We do this in three ways. First, we derive additional long-run restrictions and use them as tests of overidentification. Second, we compare the qualitative implications from the model with the impulse responses of variables such as wages and consumption. Third, we test whether some standard "exogenous" variables predict the shock variables. We find that ilshocks, military build-ups, and Romer dates do not predict the sholck labeled "technology." We then show ways in which a standard DGE model can be modified to fit Gali's finding that a positive technology shock leads to lower labor input. Finally, we re-examine the properties of the other key shock to the system.