Incomplete product availability arising from stock-out events and capacity constraints is a common and important feature of many markets. Periods of unavailability censor the observed sales for the affected product, and potentially increase observed sales of available substitutes. As a result, failing to adjust for incomplete product availability can lead to biased demand estimates. Common applications of these demand estimates, such as computing welfare effects from mergers or new products, are therefore unreliable in such settings. These issues are likely to arise in many industries, from retail to sporting events to airlines. In this paper, we study a new and unique dataset from one of the first wireless inventory management systems in the country, which was installed on a set of 54 vending machines in order to track product availability at high frequency (roughly every four hours). These data allow us to account for product availability when estimating demand, and introduces a valuable source of variation for identifying substitution patterns. We also develop a simple procedure that allows for changes in product availability even when we only observe inventory (and thus availability) periodically. We find significant differences in the parameter estimates of demand, and as a result, the corrected model predicts significantly larger impacts of stock-outs on profitability.
You may purchase this paper on-line in .pdf format
from SSRN.com ($5) for electronic delivery.
This paper was revised on April 22, 2009
Machine-readable bibliographic record -
MARC,
RIS,
BibTeX
