We reconsider the paradox that the variance of production often exceeds the variance of shipments, a result that casts doubt on the production smoothing model of inventory behavior. Most studies of production smoothing have analyzed two-digit SIC dollar-value data from Commerce Department surveys. In contrast, this paper examines disaggregated physical-product data and finds that production is smoother than shipments in about two-thirds of the 38 industry or product groupings considered. However, doubts about the model persist; most notably, Euler equation estimates of the model's cost parameters are usually imprecise and often do not have the signs postulated by production smoothing models.

We reconsider the paradox that the variance of production often exceeds the variance of shipments, a result that casts doubt on the production smoothing model of inventory behavior. Most studies of production smoothing have analyzed two-digit SIC dollar-value data from Commerce Department surveys. In contrast, this paper examines disaggregated physical-product data and finds that production is smoother than shipments in about two-thirds of the 38 industry or product groupings considered. However, doubts about the model persist; most notably, Euler equation estimates of the model's cost parameters are usually imprecise and often do not have the signs postulated by production smoothing models.