Working Papers

“Flavorants and Addiction: An Empirical Analysis of Tobacco Product Bans and Taxation.”

with Jiawei Chen, revise and resubmit at the International Journal of Industrial Organization

Awarded: Charles A. Lave Paper Prize (UCI)

  • We evaluate the proposed FDA menthol cigarette ban using aggregate-level retail data and micro-level household data. The model incorporates addiction and household heterogeneity, with a focus on low-income households and the Black community, who consume menthol cigarettes the most. The ban reduces cigarette usage by 12.6% and the Black smoking rate by 35%, while demand for e-cigarettes and cessation products increases by 4.9% and 1.7%, respectively. A $1.02-per-pack cigarette sales tax is as effective as the menthol cigarette ban, with a smaller reduction in consumer surplus across most demographic groups, especially Black Americans. Including non-tobacco flavored e-cigarettes in the ban reduces cigarette consumption similarly, while e-cigarette usage decreases by 46%.

“Substitution Patterns and Welfare Implications of Local Taxation: Empirical Analysis of a Soda Tax.”

with Jiawei Chen and Saad Andalib Syed Shah

  • We present a structural choice model that incorporates households' geographic and product substitution for studying the effects of localized taxation policies. Using detailed retail and household data pertaining to Philadelphia's soda tax, we estimate the choice model linking households' demographic characteristics and proximity to the city border to their tax avoidance behavior—switching from taxed to untaxed products or from Philadelphia to non-Philadelphia stores. We find that the inclusion of travel time is vital for modeling households' heterogeneous responses, with an extra minute of travel time to reach the untaxed region equivalent to adding 47¢ to the product price. Taking into account travel costs and the switch to less preferred products, Philadelphia households on average incur a loss in consumer surplus more than twice the amount of tax paid, with low-income households bearing the largest burden.

Work in Progress

“Money and Politics: A Double-Debiased Examination of 2018 House Election Campaign Contributions.”

  • The 2018 House election in the United States marked a significant juncture in the nation's political landscape, with campaign contributions playing a pivotal role in shaping electoral outcomes. Drawing on a structural model of consumer choice, I leverage a unique dataset that integrates both aggregate and individual-level data. In doing so I account for heterogeneous preferences that exist among the electorate. Further, I employ a double-debiased machine learning method to identify the causal relationship between consumer choice and political campaign contributions.

“A Choice-Based Examination of Healthy Food Labeling.”

  • In an era where health-conscious consumer choices are driving significant shifts in the food industry, understanding the economic dimensions of "healthy" food labeling is paramount. This paper investigates consumer willingness to pay (WTP) for food products bearing the "Healthy" label and conducts a counterfactual simulation to explore the potential consequences of the FDA's proposed update to the criteria governing the use of this term in food labeling.

“Two-Stage Structured Probit Demand Estimation for Application to Large Choice Sets.”

  • An often unappreciated feature in discrete choice literature is the application of probit choice models when in the presence of large alternative sets. The difficulty of both estimating multinomial probit models and identifying high-dimensional covariance matrices leads most researchers to employ alternative methods or restrictive assumptions that disregard realistic substitution patterns. However, when unobservables are correlated across elements of the choice set, ignoring the presence of these substitution patterns can bias estimation results. In this paper, a structured covariance matrix is proposed through which substitution patterns are modeled as a function of product similarity - allowing for feasible estimation in the presence of large alternative sets. In addition, individual parameter heterogeneity is combined with a two stage consumer decision process allowing for dynamic and individual level behavior. To estimate the model, this paper develops a Bayesian MCMC process that employs the a Tailored Random Block Metropolis Hastings algorithm. Finally, through a simulation experiment this paper demonstrates that the proposed model out preforms alternative methods of estimation, suggesting that restrictive substitution patterns may inhibit proper estimation of parameter values.