A Structural Model of Correlated Learning and Late-Mover Advantages: The Case of Statins

Abstract

We propose a structural model of correlated learning with indirect inference to explain late-mover advantages. Our model focuses on a class of products with the following two features: (i) products which build on a common fundamental technology (e.g., computer processor, car, smartphone, etc.); (ii) consumers can observe some product attributes of a product (e.g., CPU clock speed, horse power of a car engine, screen size of a smartphone, etc.), but when making their purchase decisions, consumers are not sure how efficient the product can translate its observed attributes to perform tasks which they care about. For products which base on a similar technology, it is plausible that consumers use the information signals of one product's technological efficiency to help them update their belief about another product's technological efficiency within the same product category. As a result, a late entrant could benefit from the information spillover generated by an early entrant. We apply our framework to the statin market in Canada, where statin drugs rely on a similar mechanism to reduce the bad cholesterol. Patients/doctors can observe a statin's efficacy in reducing the bad cholesterol, but they are uncertain about how effective it can convert its reducing cholesterol ability to reducing heart disease risks. Our estimation results show that the combination of correlated learning, informative and persuasive detailing explain the success of the late entrants: Lipitor and Crestor.

Keywords: Correlated Learning, Late-mover Advantage, Clinical Trials, Detailing, Efficiency Ratio

JEL Classification: D12, I11, L65, M30, M31, M37