Risk-Smoothing across Time and the Demand for Inventories: A Mean-Variance Approach

The standard production smoothing model of inventory demand cannot represent the added incentives for smoothing risks or explain the impact of market shocks that independently affect expectations and uncertainty. Those limitations are overcome by modeling inventory demand as a problem in deterministic optimal control, with the risk-averse firm maximizing utility that is a separable function of the mean and variance of returns and the firm controlling on two decision variables, production and inventory investment. Support for the mean-variance approach comes from regressions using Survey of Professional Forecasters data to show how changes in the mean forecasts of the GDP price deflator and changes in the disagreement among deflator forecasts can explain changes in aggregate inventory investment over time. Further support comes from the ability of the model to explain the excess volatility of industry output over sales—a fact at odds with the production smoothing theory.