The Value of Patients: Heterogenous Physician Learning and Generic Drug Diffusion

This paper explores how the difference in the quantity and quality of information received by physicians shapes the learning process and subsequently the diffusion of generic drugs. By exploiting prescription level data, I find that both the volume of information and the difference in the composition of information signals received by a physician contributes to the heterogeneity in adoption rates. In particular, having more information signals from new patients who move from peers increases the adoption rate of generic drugs. To explain the findings, I develop a physician learning framework where the informativeness of signals differ across old patients and new patients from other doctors. The calibrated results suggest that new patient signals weigh more than own patient signals in directly raising physicians’ expectations on the true quality, whilst this effect does not act through reducing uncertainty around the expectation. The results on the compositional effect of information echoes with "the strength of weak ties" where new patients from peers, seen as weak ties, are more informative in raising physicians’ optimism of new drugs.