Job Market Paper (linked here)

“The Sorry Clause”

This paper shows the existence of a sorry equilibrium in a game of imperfect public monitoring. In this  equilibrium, a self-imposed costly apology tendered after an accidental defection allows for continued cooperation between players. The cost of such and apology cannot be too high or too low. Efficiency of this sorry equilibrium is evaluated and its welfare outcomes compared to other informal governance mechanisms and the formal legal system. With the possibility of accidental defections, it is shown that informal mechanisms have limitations, while formal legal systems can generate perverse incentives. The analysis shows that apologies serve as a useful economic governance institution. 

Keywords: Apology, Sorry, Imperfect Public Monitoring, Uncertainty, Social Norms, Economic Governance, Legal Institutions, Courts, Incentives.
JEL Classification: D02, D80, K40, Z13.


Work in Progress

“Understanding and Distinguishing Classifications” (with Jens Prüfer and Gillian Hadfield)

Classification institutions - such as social norms, cultural traditions, laws, or regulations - assign a normative label, acceptable or wrongful, to human behavior. Thereby they shape the expectations about other people’s behavior, reduce uncertainty, and create trust in other’s actions. This paper constructs a game theoretic model of infinitely repeated interactions between N players of different types to illustrate and compare classifications emerging from social norms with those from a state backed by formal legal system. We show that a strict classification might yield better outcomes when people comply, but will make enforcement more difficult. The central result illustrates how for a given degree of expected compliance, classifications used by social norms (don’t eat meat) with decentralized enforcement cannot be as strict as the one used by a police-action based court system (don’t eat peacock meat).

“A Clash of Classification Institutions” (with Jens Prüfer and Gillian Hadfield)

This paper investigates a situation in which two classification institutions do not conform, for instance, because a country with established norms is colonized and new laws are imposed? We construct a dynamic model where social norms clash with legal order. We show when and how norms decay gradually, where more and more players first stop enforcing and then stop complying with the norm as time proceeds. We also show that the existence of legal order can undermine norms, even if legal order cannot enforce its own laws very effectively. In such a case, players may rationally ignore the classification of both norms and laws and engage in novel behavior, implying the breakdown of both governance mechanisms. This might help explain the poor state of contemporary governance in many modern countries, including former colonies and Soviet republics.

“Yes Minister: The Problem of Monitoring the Bureaucracy” (with Ashutosh Thakur)

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. In federal system, a politician at the federal and provincial level might only be able to contract with bureaucrats at their respective levels. In such a multi-principal, multi-agent setting if both principals care about a single outcome that reflects combined efforts of both agents, the best possible contract between the principals and their respective agents will be sub-optimal. This contract would make monitoring agents harder by obscuring measurement of bureaucratic productivity by econometricians, exposing the bureaucracy to political turnover. The central argument made in the paper is that the problem of measuring bureaucratic productivity need not arise only due to noise, but also due to the institutional structure. We argue that in the Indian bureaucracy, this makes it possible for Indian Administrative Services (IAS) officers to better avoid accountability compared to officers of the Indian Police Services (IPS).