Relevance of Blp Model for The Medical Markets Issues for Hahn, Hausman, Lustig Specification Test Application
Abstract:
This paper continues a research project on physicians' choice models, started in 2017 at MIT. Professor Hut tin and Professor Hausman published a first experimental study to compare simulated data with an original dataset, extracted from the US National Ambulatory Care Survey, on Type II diabetes [1, 2]. This paper aims to discuss the specification issues of a second type of model, called the BLP model.Berry's approach is ideal for differentiated markets, such as pharmaceuticals, where oligopolies or duopolies are common. Empirical studies, often related to merger cases in markets like AIDS, already exist where misspecification tests can be empirically implemented. The medical markets have many cases of oligopolies, where competition authorities often use demand estimation in merger cases, to check ex ante and ex post cross price elasticities of market segments, impacted by the merger. Additional choice sets are then identified in major chronic conditions, or in procedures for genomic medicine (such as NGS) to compute misspecification tests at the individual level, as proposed by Hahn et ales (2020) and already tested in a simulated ML model on diabetic type II, in Huttin and Hausman (2021).
