Research Interests

Substantive: Healthcare, Information Disclosure, Public Policy, Technology, Targeting, Loyalty Programs Methodological: Causal Inference, Quasi-Experiments, Machine Learning, Applied Econometrics, Empirical IO

Research Overview

Tong's research interest lies in the intersection of marketing and healthcare. Specifically, she studies the effect of mandated disclosure of pharmaceutical firm-physician financial relationship on how physicians prescribe, and how pharmaceutical firms market products to physicians. Methodology-wise, she uses the state-of-the-art machine learning models to facilitate causal inference in quasi-experimental data.

U.S. pharmaceutical companies frequently pay doctors to promote their medicine. This creates conflict of interest issues that policy-makers often address by introducing payment disclosure laws. However, it is unclear if such information disclosure has an effect on physician prescription behavior. Tong and coauthors use individual-level claims data from a major provider of health insurance in the U.S. and employ a diff-in-diff research design to study the effect of the payment disclosure law introduced in Massachusetts in June 2009. The research design exploits the fact that while physicians operating in Massachusetts were impacted by the legislation, their counterparts in the neighboring states of Connecticut and New York were not. In order to keep the groups of physicians comparable, the authors restrict the analysis to the physicians in the counties that are on the border of these states. To deal with the concern of parallel pre-trends, they further match these border physicians using the generalized synthetic control method. They find that the Massachusetts disclosure law resulted in a decline in prescriptions in all three drug classes studied: statins, antidepressants, and antipsychotics. The findings are robust under alternative controls, time periods, and variable transformations. Tong and coauthors show that the effect is highly heterogeneous across brands and physician groups, and that the decrease in prescription is likely a consequence of increased self-monitoring among physicians to curb over-diagnosis. The paper is forthcoming at Marketing Science .

On the firm side, is more information about competitors' strategies good or bad? Will this information change the subsequent competition for physicians? While prior theories are ambiguous on this issue, Tong empirically measures how information disclosure changes firms' marketing payments to physicians using a 29-month national panel covering $133 million-dollar payments between 20 anti-diabetics brands and 50,000 physicians. Leveraging the state-of-the-art machine learning technique, Tong estimates the heterogeneous treatment effects for each drug-physician pair based on 300 features non-parametrically. Overall, disclosure of competitor payments leads to an increase in within-physician payment disparity across brands. This indicates further differentiation and reduced competition in physician targeting, which is associated with lower payment per physician-brand: the monthly payments declined by 2% on average. Moreover, physicians who were paid more heavily pre-disclosure and who have larger prescription volumes are shielded from the pay cut. These findings warn policy makers about the mixed consequences from payment disclosure.

In addition to her research on healthcare marketing, Tong has been working on a research project with Prof. Orhun on how frequent flyer program tiers increase customer loyalty to the airline. The authors find that as travelers stretch to attain status with the airline, they become more likely to choose it even when it is less appealing than its competitors. They also become more willing to pay higher prices than they otherwise would. Consumers are sophisticated about when to make such tradeoffs. If their progress falls significantly behind the pace required to attain status by the end of the year, such that their chances of attaining status seem low, they are less likely to sacrifice current utility. If their progress is substantially ahead of the target pace, they are also less likely to sacrifice utility. We document a stronger willingness to pay response among business travelers than among leisure travelers. Moral hazard explains a substantial part of the response differences between business and leisure travelers. Across all members, it accounts for one-third of the increase in willingness to pay in response to making progress towards attaining status. We estimate that companies would save at least 7% on their travel costs if their employees did not exhibit moral hazard. Overall, our results suggest a significant role of moral hazard in the success of frequent-flyer status incentives.