Forward-Looking Behavior in Mobile Data Consumption and Targeted Promotion Design: A Dynamic Structural Model


Speaker


Abstract

This paper examines the dynamic consumption behavior of individual mobile data users by employing a unique dataset on individual-level daily usage over multiple months. Whether and which individual mobile data users are forward-looking by dynamically balancing present and future usage and how to design profitable promotions targeting these users are questions of both academic and managerial interest. By developing a dynamic structural model and formally proving its theoretical properties, we discover distinct temporal usage patterns that can distinguish forward-looking users from myopic ones. An empirical test is constructed to test for individual forward-looking behavior by matching the observed usage patterns with the theoretical results. We find a considerable proportion of users (about 40%) are indeed forward-looking and also find empirical evidence of individual consumer myopia. Our approach enables us to apply the dynamic model only to those exhibiting forward-looking behavior. It hence serves as a feasible option to control for consumer myopia in estimating dynamic structural models given the inherent limitation that individual discount factors are generally unidentifiable. Our structural model is shown to accurately capture the dynamic trends observed in the actual usage data. It enables sophisticated counterfactual simulations incorporating various factors (e.g., consumer anticipation, plan switch) to deliver rich implications for targeted promotion design. As we find, promotions targeting only forward-looking consumers could be significantly more profitable than blanket promotions uniformly applied to all. Properly designed end-of-month promotions targeting forward-looking users could help mobile carriers fully utilize the otherwise excess network bandwidth and increase revenue at little extra cost.