We develop an equilibrium business cycle model in which the producers of final goods pursue generalized (S,s) inventory policies with respect to intermediate goods, a consequence of nonconvex factor adjustment costs. Calibrating our model to reproduce the average inventory-to-sales ratio in postwar U.S. data, we find that it explains over half of the cyclical variability of inventory investment. Moreover, inventory accumulation is strongly procyclical, and production is more volatile than sales, as in the data. ; The comovement between inventory investment and final sales is often interpreted as evidence that inventories amplify aggregate fluctuations. In contrast, our model economy exhibits a business cycle similar to that of a comparable benchmark without inventories, though we do observe somewhat higher variability in employment, and lower variability in consumption and investment. Thus, equilibrium analysis, which necessarily endogenizes final sales, alters our understanding of the role of inventory accumulation for cyclical movements in GDP. The presence of inventories does not substantially raise the variability of production, because it dampens movements in final sales. Similarly, when reductions in adjustment costs lower, but do not eliminate, average inventory holdings, the variability of GDP is essentially unchanged, because the reduced costs cause an offsetting rise in the variability of final sales.
