My research uses tools from market design, matching theory, and game theory to study institutional design in development, political economy, labor, and finance.
- Market Design
- Political Economy
Job Market Paper
Using a matching theory perspective, I analyze the design and the impact of Indian Civil Service state assignment mechanisms used to allocate elite civil servants to different parts of the country. I find that a recent change in the matching mechanism in 2008 has systematically skewed assignments by assigning relatively poor quality bureaucrats to disadvantaged states: regions with external foreign conflict, states with internal political strife, and newly-formed states. This paper i) analyzes the causes of these imbalances, ii) assesses the impact of this mechanism change on state capacity, development outcomes, and bureaucratic performance, and iii) highlights trade-offs in implementing alternate mechanisms. Global balance in quality across state cadres is a unique constraint that arises when applying matching to political economy settings, as the mechanism designer is a paternalistic central planner. Thus, less is left to the market compared to most canonical matching applications. On the other hand, the use of matching in political economy is also novel, and careful understanding of how different matching mechanisms address underlying correlations in the data has far-reaching consequences for bureaucratic performance and development outcomes.
I analyze the assignment mechanisms used by the political parties in the US Senate to match their members to legislative committees from a matching theory perspective: one-sided, many-to-many matching with existing tenants. Understanding the data-generating process through detailed analysis of committee assignment procedures is crucial, but under-appreciated in the literature. I find that Republicans and Democrats use two distinct matching mechanisms creating very different strategic incentives, and analyze the theoretical and empirical implications of these organizational procedures. First, foundational theories of committee assignments must be re-evaluated in light of the constraints and induced strategic considerations of the two mechanisms. Second, my findings inform how to parse data for well-grounded empirical analysis by exploiting marked differences in strategyproofness which these mechanisms induce, both across parties and across seniorities within party. Lastly, I derive testable predictions from matching theory and present suggestive empirical tests highlighting the strategic intricacies of the assignment process.
In many institutional settings, members need to be assigned to certain positions, whether this is legislators to committees, executives to roles, or workers to teams. These are known as assignment problems and in this paper I show that they lead to novel questions about institutional stability. Will the set of agents being assigned prefer or vote in favor of some alternative allocation over their current allocations? I explore questions of institutional stability where the choice of the institution (i.e., the matching mechanism) is chosen and agreed upon by the very people who are assigned by the allocation procedure. I endogenize an institution's choice of allocation procedures by generalizing social choice to what I call a social allocation choice problem. I discuss a variety of voting rules (plurality, majority, and unanimity) and their stability counterparts in matching theory (popular matching, majority stability, and pareto efficiency). The novel property of majority stability is introduced and its existence and robustness to correlated preferences and interdependent preferences are analyzed. I find that certain mechanisms commonly used in practice, such as Serial Dictatorship, are surprisingly robust to highly correlated preferences under majority stability, especially in comparison to popular matching which places highly restrictive requirements on preferences of individuals and details of the assignment procedures. Unlike traditional social choice, where stability is undermined by cycles in preferences, in this setting, I show that chains of envy undermine majority stability. Chains of envy are necessary to overcome the packing problem that arises in reallocating a majority to a new set of assignments under an alternative allocation.
We model the dynamics of endogenous organizational restructuring, where those being assigned positions in an organization can themselves lobby for who gets which position. Internal labor market changes depend on how much individuals value their own status in the organization, the organizational output, their friends' welfare, and the quality of their own departmental colleagues. Meritocratic assignments are reached with probability one when agents value organizational output even with epsilon weight, provided friend networks and departments are not too large. We also characterize the effects of various voting rules, agendas, and specializations on the paths and the stability of organizational structures.
We explore how the signaling by waiting mechanism functions to overcome adverse selection in a multi-period market, where asset qualities differ in their per-period probability of default. In this novel dynamic adverse selection environment, we begin to consider the impact of partitioning the market by quality with stress tests or rating agencies, and evaluate the relative costs, dynamics, and efficacy of various government intervention policies: changing interest rates, quantitative easing, bailouts, and asset purchases. Although focusing solely on the market with uniformly distributed qualities, we characterize the optimal method and timing of government interventions in response to a market freeze, and discuss the trade-offs and driving forces of the model which we hope to generalize.
We analyze continuous time optimal contracting in principal multi-agent moral hazard settings; particularly, what the implications are as the number of agents increases in the model. Using tools from game theory and stochastic optimal control, we derive the optimal, history-contingent contract for the generalized principal n-agent dynamic problem; in the process showing how continuous time framework makes dynamic contracting and analysis tractable. Efficiency gain from specialization and rising disincentives from increased moral hazard counteract one another as we increase the number of agents in our model, thus, we derive the optimal size of a firm/team (microeconomics application) or of a fiscal union (political economy application). Furthermore, our model suggests that development of strong political and economic institutions goes hand-in-hand with greater delegation of responsibility, decentralization, and federalism.
Existing literature shows that it is possible for rational players to establish one-sided and two-sided reputation in a bilateral bargaining environment by mimicking irrational, “r-insistent” players, and that this reputation build-up can drastically change the Rubinstein (1982) outcome, by causing delay in reaching agreement. Furthermore, the literature on outside options in bilateral bargaining suggests that unless outside options are very large, they do not affect these bargaining outcomes. We are interested in the impact on the bargaining outcome from endogenizing outside options---namely, bargaining in markets. In the presence of competition in decentralized search markets, can rational agents mimic irrationality to build reputation, when can irrational types on both sides of the market trade in equilibrium, and what are the consequences for delay and efficiency? We develop discrete, hybrid and continuous-time models of bargaining in markets with and without irrational types by combining reputational bargaining of Abreu and Gul (2000) with a continuous-time version of the Rubinstein and Wolinsky (1985) model of bargaining in markets. Applications include decentralized search markets for labor, exotic assets, over-the-counter securities, venture capital funding, and repurchase agreements.
Work in Progress
We propose a rating system for two-sided dating markets where i) men and women ratings lie on the same rating scale and ii) rating of an individual takes into account both the liking behavior and the ratings of the counterparties who are recommended their profile for evaluation. We find four uses of this dimension reduction exercise using this ELO-style rating system. First, incorporating ratings of counterparties in adjusting one's rating allows the platform to more efficiently use information and is an improvement over simple measures such as percentage likes amongst recommendations the platform has suggested. Second, assortative matching by recommending man-woman pairs closest in rating maximizes the expected number of matches on the platform. Third, for any individual i, using data from all N\i to compute i's estimated rating, and then using maximum-likelihood techniques to use i's liking behavior holding N\i ratings fixed to calculate his behavioral rating, allows detection of behavioral biases such as over-confidence. Lastly, the single-dimension rating system interacted with other covariates allows the platform to better identify which dimensions an individual behaves differently relative to their estimated rating: i.e., whether individual i's preference criterion are different for counterparties with same race, religion, hobbies, etc.
We show how differences in institutional and contractual design might explain the differences in ease of accountability and monitoring of public servants across various civil service agencies. We find that relative to the optimal contract in a principal-agent, optimal contract in a federalist system with multiple principals who can only contract with their agent, but care about a single outcome which reflects the combined effort of multiple agents (as either substitutes or complements) leads to suboptimal outcomes, makes monitoring and bureaucratic accountability of agents harder, obscures outcome measurement of bureaucratic productivity by econometricians, and exposes bureaucratic monitoring to political turnover. The result is driven not by noise but by the institutional structure, and highlights how IAS---but not IPS---officers are able to hide behind and blame others, and hence cannot be easily monitored.