Abstract

Endogenous Product Choices and Optimal Regulations: A Case of E-Cigarettes This paper studies firms' endogenous product-entry decisions in the U.S. e-cigarette market by combining a random-coefficients logit demand model with a multi-stage model of product entry. Big firms first choose which nicotine categories to serve; small firms then make ordered entry decisions in response. This structure generates a unique equilibrium prediction for the number of small-firm entrants in each product category and partially identifies fixed costs for big firms. Competition effects are heterogeneous: big-firm presence depresses small-firm profitability while small firms depress each other with-in and across categories. In the main counterfactual, a ban on high-nicotine e-cigarettes reduces total consumption by 34.8 to 64.2 percent across representative markets and lowers consumer surplus by 44.7 to 70.2 percent. The equilibrium response differs across markets: in some, the ban increases active firms by up to 15.9 percent as reduced big-firm competition creates room for small-firm entry; in others, active firms fall by 12.1 percent. Comparing the endogenous equilibrium with a benchmark that holds firm choices fixed shows that endogenous firm adjustment materially affects entry, product variety, and the magnitude of the consumer-surplus loss. The consumer-surplus cost of the ban establishes a lower bound on the health and externality benefits the policy must deliver to be welfare-improving. Strategic Interactions in Ambulance Diversion: Evidence from San Francisco County Ambulance diversion is the practice by which Emergency Departments (EDs) signal that they are unable to accept more ambulance patients, thereby discouraging incoming transports. While intended by EMS agencies solely as a response to internal congestion, this paper investigates whether diversion decisions are also driven by strategic interactions, specifically whether a hospital's likelihood of diverting is directly influenced by the diversion status of its peers. Using administrative data from San Francisco County, California, between 2015 and 2020, we first document a cascade of diversions where the likelihood of a hospital declaring diversion increases monotonically with the number of peers already on diversion. To determine whether this pattern is driven by strategic interactions rather than correlated demand shocks, we estimate a Cox Proportional Hazards model of the time to diversion, defined as the duration a hospital remains open before declaring diversion, while controlling for time-varying internal congestion measures including ambulance inflow, ED walk-in volume, and inpatient census. Our estimates confirm that peer diversion significantly influences the timing of diversion, as the hazard of declaring diversion increases monotonically with the number of diverting peers, peaking at a 125.0% increase when three peers are on diversion compared to none. We also observe substantial heterogeneity in diversion behavior across hospital types: while community hospitals exhibit high sensitivity to peer behavior, an insurer-owned hospital that internalizes medical costs remains mostly insulated from these strategic interactions. These findings provide empirical evidence that diversion is not solely a response to internal congestion but a structural feature of strategic interactions within the hospital network.

Committee Chair

Stephen Ryan

Committee Members

Gaurab Aryal; Guanyi Wang; Ian Fillmore; Marcus Berliant

Degree

Doctor of Philosophy (PhD)

Author's Department

Economics

Author's School

Graduate School of Arts and Sciences

Document Type

Dissertation

Date of Award

5-6-2026

Language

English (en)

Author's ORCID

https://orcid.org/0009-0005-6623-279X

Download

Available for download on Wednesday, April 14, 2027

Included in

Economics Commons

Share

COinS