When:
Thursday, October 9, 2025
11:00 AM - 12:00 PM CT
Where: Kellogg Global Hub, 3301, 2211 Campus Drive, Evanston, IL 60208 map it
Audience: Faculty/Staff - Post Docs/Docs - Graduate Students
Contact:
Maggie Hendrix
(847) 467-7263
margaret.hendrix@northwestern.edu
Group: Department of Economics: HELP Workshop
Category: Academic
Speaker: Lydia Cao
Title: When Payers Intervene: Insurer Restrictions, Cost Control, and Physician Expertise
Abstract: In this project, I examine the trade-offs involved when insurers intervene in physicians’ prescribing decisions, focusing on two widely used utilization management (UM) protocols: step therapy and prior authorization. These restrictions are typically applied to expensive brand-name drugs, requiring prescribers to first try a lower-cost alternative or complete an administrative approval process. Within a principal-agent framework, insurers employ these non-financial restrictions to (1) align physician behavior with their own preferences for cost containment and (2) reduce inefficient drug experimentation, especially when physicians lack a strong informational advantage. Using data on antidiabetic drugs in Medicare Part D, I show that UM reduces the use of restricted drugs and delays their initiation. These effects are concentrated in early prescribing decisions, consistent with UM disrupting physician learning through initial choice set constraints. To quantify the mechanisms of incentive alignment and limited experimentation underlying these patterns, I estimate a structural model of physician drug choice. This allows me to recover the distribution of drug match values and the joint distribution of physician preferences and ability. Preliminary results indicate that, on average, physicians are relatively insensitive to out-of-pocket costs, and their uncertainty about drug effectiveness resolves substantially over time. This suggests that the incentive gap plays a crucial role in insurers’ use of UM. At the same time, there is substantial dispersion across physicians in both cost sensitivity and signal precision, pointing to potential gains from more targeted, non-uniform restrictions. I plan to use the model estimates to assess the welfare implications of alternative insurance restriction designs, and related policies that govern physician autonomy and insurer oversight.