Working Papers

Meritocracy in a Bureaucracy [latest version]

Abstract: Bureaucracies often design rules and constrain discretion to avoid nepotism. Yet such rules may not be necessary in cases where the interests of the decision-maker and the bureaucracy are aligned. I examine discretionary promotions of junior Pakistan Administrative Services (PAS) bureaucrats, in a setting where corruption and nepotism are viewed as the norm. I compile unique data on the abilities of junior officers, including both publicly available recruitment exam rank and information on job performance that is private to senior officials. Results show that seniors use both public and private information meritocratically in making these fast-track promotions. Despite no explicit incentives, seniors are equally meritocratic when choosing and promoting juniors for other teams as for their own teams. This is consistent with implicit incentives aligning incentives.

Corruption as informal fiscal policy (with Clement Minaudier (University of Vienna) and Sandip Sukhtankar (UVA)) [email for latest version]

Abstract: Corruption is widespread and persistent, with governments often turning a blind eye to its existence. We explore a hitherto underappreciated reason explaining why governments may deliberately ignore corruption: its ability to serve as an informal, parallel fiscal system. We document that in settings with low state capacity and resources, governments expect low-level officials to informally extract rents from citizens in order to fund delivery of public goods and services. Using survey data and government accounts in Pakistan and India, we show that public officials use their own personal funds to complement official funding for public services, and part of these funds come from bribes. We propose a model of bureaucratic agency to explore when governments might want to sustain such systems. These systems are more likely to arise when bureaucrats face strong incentives to provide public services, the costs of monitoring corruption are high, and the government relies on the support of voters less affected by corruption.

Spillovers in State Capacity Formation (with Clement Minaudier (University of Vienna)) [email for latest version]

Abstract: State capacity is important for development, yet we know little about how different forms of state capacity interact. In this paper, we present evidence that legal capacity reforms can have positive spillover effects on improving fiscal capacity due to organizational reasons. We employ a difference-in-difference strategy and exploit the staggered roll-out of computerization of land records in Punjab, Pakistan as random variation in legal capacity. We analyze the performance of bureaucrats who were in charge of both land records and tax collection before the reform but only of tax collection following automation. We find that automating land records led to an improvement in tax collection by these bureaucrats. However, we find significant heterogeneity by honesty of bureaucrats. We provide evidence that taxes were not easier to collect from households, rather part of the positive effect on performance is driven by selection of honest bureaucrats into automated revenue circles. Our results are consistent with a multitasking career concern model where bureaucrats are heterogeneous in their alignment with the organization. These findings imply that the social benefits of investments in legal capacity are underestimated, particularly when they affect the incentive structures of bureaucracies. The paper highlights the importance of understanding state capacity development from an organizational perspective.


Gender, information exchange and choice over co-workers: experimental evidence (with Clement Minaudier (University of Vienna), Brais Álvarez Pereira (NOVAFRICA, Universidade NOVA SBE) and Shamyla Chaudry (Lahore School of Economics)) [new draft coming soon]

Abstract: We study whether the gender of co-workers matters for information exchange and performance in teams. We carried out a lab-in-the-field experiment where `test-takers’ were randomly assigned `helpers’. Helper allocation was randomised on two dimensions: (1) gender and (2) test-taker preference for that helper at baseline. To directly investigate frictions in information exchange, we cross-randomized the availability of hints to help with the task. We find that in categories that are not gender stereo-typically female,  randomly-assigned male helpers reduce performance of women test-takers, as compared to preferred male helpers or female helpers of either type. Moreover, less hints are accessed with them, which are not compensated by any informal knowledge spillovers.   This suggests that working with a partner of the opposite gender creates frictions, but that these frictions can be mitigated by allowing choice over co-workers. Organizations can benefit from asking women to select their co-workers in team formation.

Should diversity be introduced in large or small groups? Evidence from lab in the field (with Brais Álvarez Pereira (NOVAFRICA, Universidade NOVA SBE)) [latest version here]

Abstract: In this paper we analyse the relationship between group size and skills diversity on performance of workers. We first explore this link theoretically in a model with knowledge spillovers in production. The main prediction from the model is that the size of the group in which diversity is introduced matters. Having a diverse group-member increases performance of workers in a large group more than in a small one. We test this hypothesis in a lab in Guinea-Bissau with Nursing and Economics students. Students solve a test both individually and in homogeneous and heterogeneous groups of different sizes. The questions on the test are such that there is complementarity of skills required from Economics and Nursing disciplines to answer it. We study the outcomes of Nursing students that are randomly allocated to large and small groups. The Economics student provide a source of exogenous variation in skills diversity in these groups of different sizes of nursing students. We find that the nurses with an economist colleague in a large group perform 4.3 times higher than those in a small group. Our findings suggest that group size plays a central role in determining how skills diversity affects performance of workers.

Charitable donations and violence: Evidence from Pakistan 

Abstract: This paper suggests a new channel of violence: the charitable donations channel. I exploit the rules of religious donations coupled with variations in the international price of gold/silver to arrive at a source of exogenous variation for donations. Using district-year level data on average household donations and terrorist attacks in Pakistan for the years 2001-2013, I find that donations increase the probability of a terrorist attack by 79% and the number of terrorist attacks by 30. All the effect of donations appears to work through an increase in suicide attacks as a specific terrorism tactic. All other tactics appear unaffected. 

  • Based on the MRes paper:  The Economic Causes of Terror: Evidence from Rainfall Variation and Terrorist Attacks in Pakistan [click here]

Selected Work in Progress

Welfare Effects of Improving Public Distribution System in India (with Sheetal Seekhri (UVA) and Gaurav Chiplunkar (UVA))

Efficiency of discretionary allocations: Evidence from PAS bureaucracy in Pakistan (with Gaurab Aryal (UVA), Clement Minaudier (University of Vienna) and Zahid Habib Bhutta (Additional Accountant General, AGP))

Why innovation happens in the public sector? The case of National Cancer Institute (with Natalie Aviles (UVA))