Research Summary
Research Summary
Incorporating Price and Inventory Endogeneity in Firm-Level Sales Forecasting.
Description
Forecasting firm-level sales is a key activity in top-down planning in most organizations. In the retailing industry, firms can use inventory and price to stimulate demand. Hence, standard time series methods for sales forecasting can be improved by incorporating inventory and price as causal variables. However, the relationships among sales, inventory and prices can be complex due to mutual dependence on each other. In this paper, we propose a simultaneous equations model to represent these dependencies, and apply this model to forecast sales. We augment this model with lagged variables as well as several exogenous explanatory variables such as the proportion of new stores, the age of inventory, capital investment per store, and selling expenditure. We use publicly available firm-level financial data and data collected from annual reports to test the model and various hypothesized relationships. In numerical tests, our model provides similar or better forecast accuracy of sales compared to benchmarks based on time-series models that either ignore inventory and price or the simultaneity that exists among the three variables. Our paper extends previous empirical research on firm-level inventories by showing that firm-level sales, inventory and price have a triangular relationship with each variable affecting the other two variables.