Inertia and Learning in Physician Prescribing Behavior
thesisposted on 2020-05-01, 00:00 authored by Caitlyn Rebecca Fleming
Pharmaceutical spending in the United States represents 10 percent of national health expenditure. There is a vigorous policy debate on lowering pharmaceutical spending, in which generics play a large role. Many institutional features encourage generic drugs as the default choice once they are available. However, the switch to generic drugs is slow, which is costly to patients and insurers. I estimate the effects of physician and patient familiarity with brand-name drugs on generic utilization for atypical antipsychotics. Using claims data from a large, national insurer, I show that physicians familiar with the drug class before generic entry are 16 percent more likely to prescribe the brand-name drug after generic entry. I then address two possible causes of choice persistence in brand-name drug prescriptions—switching costs and learning—using a structural demand model. Results show that physicians incur large switching costs, which are slightly reduced as learning occurs through additional prescribing of the drug. Counterfactual simulations show that policies that reduce physician switching costs, such as removing the availability of brand-name drugs once the generic enters, can increase welfare by $31 per prescription and decrease average patient copays.