Why Do Firms Hold Inventories?

A wide range of motives for holding inventories has been identified in the literature. These include the desire to smooth production in the face of fluctuating demand, the desire to avoid stockouts, and the desire to shift production into periods when costs are relatively low. Most of the inventory literature has focused on whether the 2nd moment properties of models with these and other motives are consistent with stylized 2nd moment facts in the data. That literature has largely abstracted from the implications of these motives for the level of inventory holdings. The objective of this paper is to determine if the motives listed above are strong enough to quantitatively account for the observed aggregate level of inventories held in the economy. To address this issue we construct a general equilibrium model of heterogeneous firms. We calibrate the shocks in our model by requiring that the model be consistent with the key 2nd moment properties of the microeconomic data that have been emphasized in the inventory literature. These include the volatility of sales, the covariance between production and inventory investment, and the volatility and serial correlation of production, all at the industry level. The model is also calibrated to be consistent with standard growth observations. We then ask the question, “how much of aggregate inventories holdings can be explained by each of the different motives for holding inventories?†The data analysis will be of independent interest to many macroeconomists. We estimate the degree of idiosyncratic variation experienced at a fine level of disaggregation. We document that volatility in production increases consistently as data is disaggregated, and that there is considerable diversity in volatility across industries. We find that: i) the average standard deviation of output at the finest (4-6 digit SIC codes) level of aggregation lies in the range of 8 to 20 percent; and ii) the average fraction of production variance due to idiosyncratic shocks lies in the range of 78 to 97 percent.