Nested or random coefficients logit? A comparison between alternative discrete choice models of product differentiation

© 2014 by the President and Fellows of Harvard College and the Massachusetts Institute of Technology. We propose a random coefficients nested logit (RCNL) model to compare the tractable nested logit (NL) model with the more complex random coefficients logit (RC) model. After a simulation study, we use data on the European automobile market. Both the NL and RC models are rejected against the RCNL model. The RC model results in different substitution patterns and a wider market definition than the NL and RCNL models. Nevertheless, the predicted price effects from mergers are robust across models. Our findings stress the importance of accounting for discrete sources of market segmentation not captured by continuous product characteristics.