Erik Voeten: This is a guest post by Joel Selway, an Assistant Professor of Political Science at Brigham Young University and a member of the AidData Research Consortium.
AidData recently launched a new online platform and geographical information system (GIS) module, which makes it significantly easier for researchers to upload, download, and visualize high-resolution spatial data. The launch of “AidData 3.0” inspired me to take the GIS module for a test drive to explore the following question: How does foreign aid affect the competitiveness of elections in developing democracies? I suspect that aid projects are attractive to candidates because they can (1) take credit for the completion of a past project in their constituency that is highly palpable to voters (the credit-taking hypothesis), or (2) manipulate who benefits from future projects, either by awarding contracts to supporters or engaging in corruption in order to enrich themselves (the rent-seeking hypothesis). An alternative hypothesis is that aid reduces competitiveness because the incumbent takes all the credit for the project, thereby disadvantaging competitors (the incumbency-advantage hypothesis). The null hypothesis – that aid has no effect on electoral outcomes – is also plausible. Given the vast array of issues at stake in a typical election — such as national economic policies, security, etc. — one might not expect aid projects to have any discernible impact.
To do this, I combined data on the sub-national distribution of World Bank and African Development Bank projects (from AidData) and electoral competitiveness at the constituency level, drawing on information about constituency boundaries from the Global Mapping of Electoral Districts (GMED) project. In particular, I examined three developing countries: India, Ghana, and Zambia. I measured electoral competitiveness in each district by calculating the effective number of electoral parties from the Constituency Level Electoral Archive (CLEA) project. I then leveraged the capabilities of the platform to join together and visualize these data in the maps displayed below.
Map of Electoral Competition and Aid Projects in Zambia
Interactive version of map here
Map of Electoral Competition and Aid Projects in India
Interactive version of map here
Map of Electoral Competition and Aid Projects in Ghana
Interactive version of map here
There was mixed evidence about how foreign aid affects electoral competitiveness in these three countries. I found no evidence for incumbency advantage in any of the countries, and no evidence for any of these hypotheses in India, where aid did not seem connected to competitiveness.
In Ghana, I found evidence for both the credit-taking and rent-seeking hypotheses. For aid before elections, constituencies with some funding had a higher level of electoral competitiveness than those with no funding. For aid after the elections, the same pattern occurred.
For Zambia, I found no evidence for the credit-taking hypothesis, but clear evidence for the rent-seeking hypothesis. For aid before elections, constituencies with some funding had lower levels of electoral competitiveness than those with no funding, but this difference was not statistically significant. For aid after the elections, constituencies with some funding did have a significantly higher level of electoral competitiveness than those with no funding.
There is obviously much left to explore on this question of electoral competitiveness. For example, we do not know if higher electoral competitiveness is observed because the opposition is more fragmented, or the incumbent performed worse than the opposition. And of course, in future work researchers will need to account for a host of other constituency-level characteristics, such as wealth. It might be that poorer constituencies receive more aid funding, but this same poverty also makes them more susceptible to patron-client relations, and that more village chiefs are running for office.
However, these simple tests help to illustrate the kinds of research questions that can now be pursued with the growing availability of sub-nationally geo-referenced data. The technology needed to manage and manipulate such data has deterred far too many social scientists from engaging in this type of analysis.
The AidData 3.0 platform will help correct this problem. It is a powerful and user-friendly tool that the research community can leverage for many years to come – and help improve over time.
The platform itself was built with generous support from USAID’s Higher Education Solution Network and benefited from lessons learned through collaboration with the World Bank-AidData “Mapping for Results” partnership. However, as many readers of this blog will know, AidData traces its origins to a 2003 grant from the National Science Foundation Political Science Program. If not for the initial funding from the NSF, the platform, mapping tools, and data-sets produced by the AidData team would not be available to researchers, policymakers, journalists or citizens.
