Inertia, Learning, and Welfare in Physician Prescribing Behavior: The Case of Generic Entry

Prescription drug spending and increasing pharmaceutical prices have taken a pivotal role in health care policy debates. The use of generic drugs is seen as a key decision that could aid in bringing down costs and spending on prescription drugs. However, physician-prescribing behavior could be a key factor in how often and to whom generic drugs are prescribed.

Physician prescribing behavior determines how often and to whom specific drugs are prescribed. Despite their overwhelming market share, the use of generic drugs may be impeded by physician awareness of generic equivalents and uncertainty or risk-aversion towards generics. As drugs are an experience good, physicians and their patients may be uncertain about the quality of new generic drugs. While generic drugs have the identical active ingredients as their brand-name counterparts, the generic drugs may vary in inactive ingredients that can result in differential reactions for different patients.

In this paper, I explore physician prescribing behavior and drug choice in the face of generic entry into the market. My research focuses on five atypical antipsychotics that lose exclusivity in the 2012 to 2015 period and uses pharmaceutical and medical claims data from a national health insurance company. Descriptive and quasi-experimental evidence shows that physicians who are prescribing prior to generic entry exhibit learning: in the months following generic entry, the likelihood of prescribing a generic drug increases as I extend the time horizon. I create a model of physician demand that allows for unobserved preference heterogeneity in drug choices to identify inertia. To the extent that inertia exists, I isolate switching costs from an element of learning, which develops as physicians gain more experience with the generic drug. I identify and quantify the effect of physician inertia in prescribing behavior separate from persistent unobserved preference heterogeneity using a random coefficients choice model.

The type of choice persistence by physicians I observe has important welfare implications. Finding that physicians experience switching costs, if it is welfare reducing, may suggest that investments in interventions to make physicians less inertial will be beneficial. Such interventions might include increasing physician awareness of copay differentials or education on side effects of the generic due to non-molecular differences in the drug. However, if physicians' choice persistence is a function of learning, these investments may not be worthwhile if physicians make the welfare-enhancing choice eventually. I shine light on these possibilities through counterfactual simulations, which explore policies related to prescribing interventions and payments that pharmaceutical firms make to physicians for meals or speaking fees (physician detailing).