top of page


Research: Welcome

Selected Working Papers

“How to Use Recognition Rewards to Improve Quality of Bureaucrats? Experimental Evidence from a Public Teacher Training Programme in Pakistan"

On-the-job training remains the primary vehicle for upgrading the quality of bureaucrats across many low-and middle-income countries. This paper reports experimental evidence on the impact of recognition rewards on teacher knowledge and skills in a mandatory teacher training program in Punjab, Pakistan. Public school head teachers attending the training were randomized into four different recognition rewards tied to knowledge acquisition as measured by training test scores, and a control group. The first recognition reward made peer-esteem from recognition salient, the second made potential career benefits from recognition salient, and the third and fourth treatments combined the first two treatments with a motivational framing to improve the design of the rewards. Results show the following: first, recognition rewards that make career benefits salient lead to higher training test scores. Second, adding a motivational framing makes teachers overconfident reducing their performance in the training. These findings suggest that recognition rewards can be a cost-effective way to improve the effectiveness of public teacher training programs, and civil service trainings more broadly. However, their framing is nontrivial and can have significant effects.

“Rules versus Discretion: Experimental Evidence on Incentives for Public Extension Staff “ (with Garance Genicot and Ghazala Mansuri) New version coming soon

This study provides evidence from a randomized evaluation of a province-wide implementation of three pay-for-performance schemes in partnership with the Agriculture Extension Department in Punjab, Pakistan. The performance schemes leverage the departmental digital monitoring system, AgriSmart, and link incentives to objective metrics available on AgriSmart (Objective arm), supervisors’ own subjective evaluation of staff (Subjective arm), or supervisors’ own subjective evaluation with an additional element of top-down monitoring. We show that all treatment arms improve extension outreach on measures such as days worked and village visits, but only the Subjective Plus arm shows positive treatment effects on farmer feedback on extension outreach and quality. Further analysis reveals that these effects in Subjective Plus are driven by supervisors scheduling more Farmer Training Programmes, which are completed in less frequently visited villages, leading to positive impacts on extension quality for marginalized farmers. These findings have implications for the design of incentive programs in government, emphasizing how to leverage the discretion of supervisors effectively in service delivery contexts where the different dimensions of performance are hard to measure.  



“Systems Approaches to Public Service Delivery: Methods and Frameworks” (with Martin Williams) 


Researchers and practitioners are increasingly embracing systems approaches to deal with the complexity of public service delivery. However, the diversity of these methods and their lack of common theoretical grounding has limited constructive engagement between those working within the systems tradition and those working outside it. We address this by reviewing and critically synthesizing systems literature from the fields of health, education, and infrastructure. We argue that the common theoretical core of systems approaches is the idea that multidimensional complementarities between a policy and other aspects of the policy context are the first-order problem of policy design and evaluation. What differentiates systems approaches from other research traditions is thus not so much a specific method as a general difference in question prioritization, and consequently greater methodological pluralism. We distinguish between macro-systems approaches, which focus on the collective coherence of a set of policies or institutions, and micro-systems approaches, which focus on how a single policy interacts with the context in which it operates. We develop a typology of micro-systems approaches and their relationship to standard impact evaluation methods, and discuss their relationship to adjacent concepts such as external validity, implementation science, and complexity theory.

"Approaches to implementation and delivery outcomes: Evidence from Ghanaian Education Districts” (with Clare Leaver, Pia Iocco Barrais, Michael Boakye-Yiadom) New version coming soon


This paper studies mid-level bureaucracies: organizations operating between national and street levels that are commonly tasked with implementation of sector plans. We examine how these bureaucracies approach policy implementation, how they perform, and the relationship between the two. Our data come from a nationally representative survey of 174 Ghanaian district education offices (and six schools within each of these districts for triangulation), together with a more detailed survey of 1,261 schools in the north of Ghana. We document substantial variation in how district education offices approach policy implementation and their performance (conceptualized in terms of effort by office staff, direct office outputs, and indirect outputs at the frontline). We then show that our management index—a measure of organizational approach to implementation across four delivery functions—is positively associated with performance. Not all of the delivery functions matter equally. Our descriptive evidence points toward benefits from problem-solving practices over top-down accountability, raising policy considerations for Ghana’s current delivery approach and beyond.

"Understanding and Mitigating Risk-Taking in Bureaucracies: The Role of Prosocial Motivation” (with Shehryar Banuri and Adnan Qadir) New version coming soon


Decisions of policymakers can have important implications for the quality of public service delivery and the populations they serve at large. While individuals often make decisions based on their individual biases and heuristics, the context of policy decision-making can also play a role. We study the latter dimension and shed light on the question of whether policymakers make decisions differently in policy versus personal contexts, and whether these decisions can change based on deliberation. We test this using a vignette in a lab-in-the-field setting involving two stages. In the first stage, policymakers are randomly allocated to two conditions - first, a scenario where a decision needs to be made about how to spend personal money (control); and second, a scenario where a decision needs to be made about how to spend public money (treatment). In the second stage, policymakers are randomly allocated to make the same decision in a two person team or individually. To operationalize decision-making under each of these scenarios, we measure risk appetite to understand differences in decision-making. We show three key results. First, policymakers exhibit a greater risk appetite when they take decisions with public money as opposed to their own money. Second, the greater risk appetite mainly stems from individuals who are the least prosocially motivated. Finally, we show exploratory evidence that team-based decision-making can have an impact on decision-making - in particular, having at least one member of the team to be prosocially motivated reduces risk appetite in policy decisions

"Intrinsic Motivation and Service Delivery: Evidence from the Public Education Sector in Pakistan”. 

"A Global Mapping of Delivery Approaches” (with Martin Williams, Dana Qarout, Celeste Carano, Kate Anderson, Liah Yecalo-Tecle, and Veronika Dvorakova)

Research: About

Selected Work in Progress

  • “Preferences and Performance Among Bureaucrats: Evidence from the Pakistan Civil Service" (with Adnan Qadir and Gabriel Tourek). Funded by IGC. Data collection in progress

  • “Delivery Models and Service Delivery: Experimental Evidence on Outcomes-based Contracts (with Clare Leaver, Noam Angrist, and Jennifer-Opare Kumi). Funded by FCDO. Starting field work.

Thanks for submitting!

bottom of page