Consumption rivalry generates variation in the choice sets decision-makers face. Not taking into account such variation may generate biased demand estimates. It remains unclear how this impacts estimation accuracy because researchers often lack information on temporal variation in product availability. This paper uses information on the exact set of available alternatives at the time of choosing to formulate time-variant deterministic constraints. In an application to the market for public charging infrastructure for electric vehicles, I show that incorporating this information significantly improves the out-of-sample forecasting accuracy of individual choice and hence the aggregate demand estimates for local charging facilities.
Discrete choice Preference estimation Consumption rivalry Electric Vehicles