Distributed lag inventory analyses
The objective of this study was to measure determinants of short-term inventory behavior for selected commodities: beef, pork, butter, cheese, department store stocks, manufacturers' nondurable inventories and manufacturers' durable inventories. The last two were studied using monthly data; the others, with quarterly data.
Dynamic considerations must enter into any adequate explanation of inventories. Distributed lag models were used in this study because they are one reasonable way of treating such dynamic phenomenon as expectations, frictions and lags. Such models are useful for study of inventory behavior. Nevertheless, there are some problems in using them. They commonly lead to equations to be estimated which are nonlinear in the parameters. Reduced equations containing exactly the same variables but different nonlinear combinations of parameters may be obtained from different models containing different behavioral assumptions. Whenever linear estimation is used, as in this study, we must be cautious about placing specific behavioral interpretations on the resulting coefficients.