I am a Princeton Economics Ph.D. candidate. I specialize in empirical industrial organization, especially e-commerce and multi-sided markets.
My research concentrates on the empirical analysis of e-commerce platforms using tools from structural industrial organization and causal inference techniques from applied microeconomics. During my Ph.D., I have developed in-depth knowledge and hands-on experience in economic modeling of multi-sided markets, demand estimation, machine learning, and causal inference.
In a series of papers, I explore how online platforms can design their marketplaces to shape algorithmic competition between merchants, allow for detection of collusive bidding rings, and tradeoff entry incentives against the need to surface competitively priced products in search and recommendation algorithms. More recently, I partnered with a major e-commerce company to evaluate the potential for personalizing ad exposure using a combination of tools from causal inference and machine learning such as causal forests.
I have 7+ years of experience in programming (focus on Python) and 8+ years of experience in economic modeling and data analysis.
Throughout my Ph.D., I have enjoyed collaborating with researchers at Princeton and other institutions, including partners in industry. I also had the privilege to teach students with varying levels of experience and preparation both at and prior to Princeton.
Prof. Jakub Kastl
Department of Economics
Prof. Kate Ho
Department of Economics
Prof. Adam Kapor
Department of Economics
Entry into Two-Sided Markets Shaped By Platform-Guided Search
with Kwok-Hao Lee
Job Market Paper
We evaluate the problem of firms that operate platforms matching buyers and sellers, while also selling goods on these same platforms. By being able to guide consumer search through algorithmic recommendations, these firms can influence market outcomes, a finding that has worried regulators. To analyze this phenomenon, we combine rich novel data about sales and recommendations on Amazon Marketplace with a structural model of intermediation power. In contrast to prior literature, we explicitly model seller entry. This feature enables us to assess the most plausible theory of harm from self-preferencing, i.e. that it is a barrier to entry. We find that recommendations are highly price elastic but favor Amazon. A substantial fraction of customers only consider recommended offers, and recommendations hence noticeably raise the price elasticity of demand. By preferring Amazon's offer, the recommendation algorithm raises consumer welfare by approximately $4.5 billion (since consumers also prefer these offers). However, consumers are made worse off if self-preferencing makes the company raise prices by more than 7.8%. By contrast, we find no evidence of consumer harm from self-preferencing through the entry channel. Nevertheless, entry matters. The algorithm raises consumer welfare in the short and medium run by increasing the purchase rate and intensifying price competition. However, these gains are mostly offset by reduced entry in the long run.
Job Market PAper
with Charles Louis-Sidois
R&R at Theoretical Economics
We analyze a vote-buying setup where a committee votes on a proposal important to the vote buyer. We characterize the cheapest combination of bribes that guarantees the proposal's passing in different voting environments. We find that the vote buyer publicly offers small bribes to a large supermajority of members for both simultaneous and sequential votes. Each member accepts because he anticipates that the proposal will pass regardless of his vote. We discuss the committee design that maximizes capture cost: combining demanding majority requirements with diversity among members makes the committee more expensive. In small committees, sequential voting increases cost, but the opposite is true for large committees. On the other hand, additional members and transparent voting rules lower the cost.
Algorithmic Pricing Facilitates Tacit Collusion
Marimar & Christina Torres Prize
As the economy digitizes, menu costs fall, and firms can more easily monitor prices. These trends have led to the rise of automatic pricing tools. We employ a novel e-commerce dataset to examine the potential implications of these developments on price competition. We provide evidence from an RDD that the immediate impact of automatic pricing is a significant decline in prices. However, repricers have developed strategies to avoid the stark competitive realities of Bertrand-Nash competition. By employing plausibly exogenous variation in the execution of repricing strategies, we find that 'resetting' strategies (which regularly raise prices, e.g., at night) effectively coax competitors to raise their prices. While the resulting patterns of cycling prices are reminiscent of Maskin-Tirole's Edgeworth cycles, a model of equilibrium in delegated strategies fits the data better. This model suggests that if the available repricing technology remains fixed, cycling will increase, and prices could rise significantly in the future.
Weiss Foundation Grant
Collusion between government suppliers likely has significant adverse welfare effects. In this paper, we study an e-procurement market in Ukraine. After motivating our interest by documenting suspicious bidding patterns in our data, we build a model of competitive equilibrium. Frequently, we observe bids that are inconsistent with this equilibrium. In particular, when initial bids are close, suppliers should have similar costs and usually be willing to undercut each other if allowed to update their bids. To the extent that firms only engage in the predicted amount of undercutting when facing some opponents (but not when facing others), we conclude that these firms are part of a collusive ring. Finally, we successfully validate the soundness of this novel structural test of collusion on a sample of 863 prosecuted collusive firms that participated in 23,515 tenders.
The Returns to Targeted Sponsored Search
with Sarah Moshary
A key selling-point of digital advertising technology is its scope for targeting relative to traditional media, such as TV or print. While information about users allows internet platforms to decide which ads to show them, it also allows personalized decisions as to who should be shown ads in the first place. Exploiting data from an experiment on a large e-commerce platform that disabled ads for a fraction of visitors, we employ causal forests to search for high-dimensional and interactive heterogeneity in the effect of being exposed to advertising. Due to our data use agreement, we cannot yet discuss our results publicly but hope to do so soon.
I have greatly enjoyed teaching both before and during my graduate studies at Princeton University. In 2011, I taught high-school pupils economics in Guangzhou (China); in 2015, I instructed Oxford MPhil students in Real Analysis; in 2016, 2017, and 2018, I tutored gap year students at the John Locke Institute; and, most importantly, in 2018, 2019 and 2020, I taught 159 undergraduates microeconomics at Princeton. To be exact, at Princeton I was a teaching assistant for ECO 310 ("Microeconomic Theory: A Mathematical Approach") with Professors Can Urgun and Sofia Moroni. My average instructor rating was 4.4/5; detailed course evaluations are available upon request.