News and Events

2nd Building an Inclusive Democracy Discussion Series: Patronage in Government Service Delivery

Posted: 2014-07-22




Assistance programs for poor communities and individuals are an important component of public sector activities in developing countries. Yet funders of these programs, whether donors or national taxpayers, may be concerned about potential misuse of the money. Donor agencies often try to influence program design to reduce mistargeting. But there is limited evidence on whether this is effective. In this paper, we investigate whether having a link with a politician increases households' access to government assistance programs, including social safety net and education programs. We also test whether that effect differs in programs with different design features. We collected a unique primary household dataset in rural Punjab, Pakistan, which allows us to identify households from politicians' home villages and clans. We exploit close elections to compare households connected to close winner and close runner-up politicians. We use a combination Regression Discontinuity - Difference-in-differences approach, comparing the connections of just-winners and just-losers, before and after an election. This provides stronger identification than either the RD or DD strategy used separately. We find that having a politician from one's village elected approximately doubles the amount of government assistance received. The effect is concentrated in a few programs, including programs with formal targeting criteria, suggesting that the formalization of these programs still leaves much leeway for political influences in targeting. We also exploit a major reform to the targeting mechanism of a cash transfer program as a natural experiment. We find evidence that the reform resulted in many more unconnected households receiving the cash transfer program, although connected households are still significantly more likely to receive it. We explore both how and why politicians influence targeting. We also document three methodological measures we have taken to ensure our results are robust against potential concerns about data mining. Our results have implications for donor agencies and civil society organizations who aim to engage or pressure governments to reduce corruption and improve public spending.


Kate Vyborny is a Postdoctoral Associate in the Department of Economics at Duke University, USA. She studied for her D.Phil. in Economics at the University of Oxford, England, where she held the Rhodes Scholarship. Kate's research focuses on Pakistan; her doctoral work examines issues of public service delivery and political economy. Her postdoctoral research also includes work on community organizations, gender, and urban development and public transportation. Kate is also a visiting faculty member at the Lahore School of Economics and the Lahore University of Management Sciences. Kate previously worked on development assistance effectiveness at the Center for Global Development, and on trade and development at the Carnegie Endowment for International Peace. She holds bachelor's degrees in Economics and International Affairs from the University of Georgia, USA.


<<More News and Announcements






Manuel D. Cantos


Is this by special invitation or open to the public? Thank you.

Post a comment