Estimation and inference in inventory models with integrated technology shocks

Mehmet Balcılar, Robert J. Rossana

Abstract


This paper shows that the presence of integrated technology shocks will prevent input prices and sales from being cointegrated with inventories. It is further shown that the commonly used cointegration tests are not applicable to the inventory equation arising from a dynamic linear‑quadratic (LQ) model, because of the long run endogeneity of the regressors and the moving‑average error term. The paper examines suitable estimation methods for the LQ inventory model in the presence of endogenous regressors and moving-average errors. Using appropriate estimation methods, cointegration is found among inventories, material prices, and sales for two-digit sectoral data, thus, providing evidence against the presence of integrated technology shocks. The results also indicate that the implausible parameter estimates obtained by previous researchers are the result of inappropriate estimation methods and ignored structural changes.

Full Text:

Full Text

Comments on this article