A Comparison of Nonlinear Pricing Preference Models for Digital Services

We review and empirically compare a variety of hierarchical Bayes models that use the data obtained from discrete choice experiments to capture usage behavior under twopart tariffs, as are commonly used in service industries such as telecommunication, cloud computing, and online music downloads. The models enable the simultaneous prediction of consumers' service purchase-, tariff choice-, and usage quantity-decisions as well as service provider’s profit. We outline recent extensions, which differ with respect to the concepts applied to capture usage uncertainty, estimation space, and complexity of the parameter estimation. We illustrate and compare the models using data from an empirical study about Internet access services. The results show that willingness-to-pay differs significantly across quantity units, a simple estimation procedure leads to inferior results, and usage uncertainty as well as estimation space have only limited influence on the results. We also illustrate how to use these results to determine optimal tariffs.