Research Interests: Matching, Incomplete Information
Assistant Professor: University of Notre Dame
Email: pingandrew[at]gmail[dot]com
My CV
Embedded Files
Working Papers
Working Papers
I introduce a framework for studying transient matching in decentralized markets where workers learn about their preferences through their experiences. Limits on the number of available positions force workers to compete over matches. Each capacity-constrained firm employs workers whose match value exceeds a threshold. Since employment offers both payoff and information benefits, workers effectively face a multi-armed bandit problem. To them each firm acts as a bandit where the probability of “success” at the firm is driven by market competition. In such markets, aggregate demand for firms satisfies the gross substitutes condition which ensures equilibrium existence. The resulting search patterns match a variety of stylized facts from labor market data. High-quality workers search less and tenure increases with age. In general, equilibria are inefficient because competition depresses the level of search. Natural interventions designed to improve efficiency are effective in uncongested markets, but can fail when congestion is severe. From a market design perspective, the utilization of headhunters has differential effects depending on workers’ quality, conclusively improving both outcomes for low-quality workers and overall efficiency. Reducing congestion through unemployment benefits, can depress search and may ultimately reduce match efficiency.
How should the supply of public housing be optimally designed? Although commonly used queuing mechanisms treat the supply of goods as exogenous, designers often control the inflow of goods in practice. We study a dynamic matching model where the designer minimizes a convex combination of mismatch and vacancies, motivated by the Singaporean housing allocation process, Build-To-Order. The optimal mechanism overproduces underdemanded housing relative to the proportional benchmark, and competition over housing improves match quality. Batching applications artificially generates competition and is optimal when the planner places a high weight on match quality. Following our dissemination, the Singaporean government increased batching.
We study strategic interactions in decentralized matching markets, where firms make directed offers to workers and agents' preferences are aligned. We show that implementing stable outcomes through decentralized interactions is possible, albeit under stringent conditions. Stable implementation in decentralized markets requires either complete information of preferences; no time frictions; or sufficient richness of plausible market realizations. Unique implementation occurs under even harsher restrictions on market interactions.
Public Housing at Scale, 2023, with Kwok Hao Lee and Luther Yap
We consider the design of a large-scale public housing program where consumers face dynamic tradeoffs over apartments rationed via lotteries and prices. We show, theoretically and empirically, that changing rules complements increasing supply. First, we present a motivating example in which supplying more housing leads households to strategically delay their applications. By waiting for “better” developments arriving tomorrow, households forgo mediocre developments available today, resulting in more vacancies. Turning to the data from the mechanism, we formulate a dynamic choice model over housing lotteries and estimate it. Under the existing mechanism, we find that increasing supply fails to lower wait times. However, when a strategyproof mechanism is implemented, vacancies and wait times fall, but prices on the secondary market rise. Under this new mechanism, building more apartments lowers wait times and reduces the upward pricing pressure on the secondary market.
Works in Progress
Works in Progress
Evident Competition, 2024, with Clara Nguyen, and Erez Yoeli
United States civil courts rely on an adversarial system where two parties in a lawsuit obtain and present evidence according to a process known as discovery. Two discovery regimes are predominantly used: voluntary disclosure, which does not require parties to reveal all evidence in their possession, and formal discovery, which does. How do these regimes influence the extent of the parties' search for evidence and the information available to the judge? We find that each regime has its advantages: Voluntary disclosure tends to provide a stronger incentive to search relative to formal discovery, but formal discovery ensures the judge is better informed conditional on the evidence found. Furthermore, the quality of evidence plays an important role, when evidence is decisive for the judge, parties are encouraged to search more and present more evidence. Our results can help explain the legal literature's inconclusive findings on the relationship between disclosure and settlement.
Completed Papers
DyPy, 2020, with Anjalika Nande, Eric Lubin, Erez Yoeli, and Martin Nowak
We've developed a python library for simulating matrix form games! DyPy is an open source Python software library that is hosted on Github at https://github.com/anjalika-nande/dynamics_sim. The package is designed to make it simple to run evolutionary game theory simulations to model populations undergoing biological and cultural evolution in a range of fields, from biology to economics to linguistics. Detailed documentation for each command in the library, sample code for exemplary simulations and a Wiki is provided in the Github repository. Improvements through pull requests and suggestions for additional functionality are encouraged.
Geometric Invariants of Numerical Semigroups, 2016, with Maksym Fedorchuk and Jian Zhou
A natural invariant of a unibranch curve singularity is the numerical semigroup of its valuations. In the case when the curve singularity admits a GGm-action, this semigroup also determines the singularity uniquely. A rational-valued function on curve singularities with GGm-action that leads to an ordering of singularities according to their geometric complexity was proposed. We explore this function and give a classification of those numerical semigroups for which the values of this function are above a certain threshold.
Page updated
Google Sites
Report abuse