Great new research on African political economy

I am just heading home now from the African Studies Association meetings in Philadelphia, and I have to say, I was impressed by several really interesting presentations that make me quite optimistic about what we can learn about initiatives to enhance democracy and governance in Africa; and about both the practice and deeper understanding of ethnic politics. Political scientists working on Africa are doing a lot of innovative and interesting research on substantively important topics.

Yesterday, at a panel on information and government accountability, Jeremy Weinstein presented some of his work (joint with Macartan Humphreys) in which they described their massive experiment in Uganda creating and distributing parliamentarian “scorecards” to provide citizens information about the quality of work being done by their elected representatives. Another paper, presented by Guy Grossman (also joint with Humphreys and with Gabriella Sacramone-Lutz) investigated the impact of mobile phone technology on “interest articulation” or the inclination of voters to contact their representatives, again in Uganda. Kelly Zhang presented her Kenyan-based research investigating the impact of providing information about the quality of government spending on citizen attitudes and behaviors. And Lily Tsai presented a paper (joint with myself and Dan Posner) on the effects of some aspects of the Uwezo initiative in Kenya, which provided parents information about their children’s literacy and numeracy levels, and information about how to be more active citizens. (I apologize for including my own paper in a post entitled “Great new research…”)

Lily Tsai presents at ASA (lousy photo by me)
Lily Tsai presents at ASA (lousy photo by me)

The papers provided a systematic look at some of the possibilities and limitations of “open government” for improving accountability, action, and service provision. I will not summarize all of the nuanced findings here, but it’s clear that openness and transparency do not lead to immediate sea-changes in citizen-government relations. This is unfortunate news because a lot of money is being spent with potentially overly-optimistic results in mind. But it’s better to identify what’s not working and to try to explain why, than to continue operating under the assumption that any initiative to make citizens more informed with lead to better quality government. In a deeper way, this work forces us to reflect on the role of an informed citizenry in democratic government.

As Lily, Dan, and I try to point out in our paper, we need to try to really clarify the many nuanced conditions under which it’s even plausible that these types of initiatives would have the desired impact, and hopefully all of this research will help “democracy entrepreneurs” to do better, more impactful work. Despite the many null findings, I think many of the scholars working in this area still believe that information campaigns and technologies of some form will have the desired effects.

Today, I discussed four great papers on ethnic politics in Africa: Willa Friedman’s investigated the determinants of participation in the Rwandan genocide, using new villeage-level data on the numbers of people accused in the Gacaca courts of perpetrating crimes. She finds, among other things, that more people were accused — and thus likely more participated — in villages where there was a high level of Hutu education and Hutu unemployment. A reasonable interpretation: personal frustration contributed to individuals’ decision to participate in the holocaust.

The other three papers focused on ethnic voting. Liz Carlson described some experimental research and analysis of Afrobarometer survey data that shows the extent to which many Africans will under-report their bias towards voting for co-ethnics in situations where other people are present. This type of mis-representation says something about the negative connotation of ethnic politics among Africans, and forces us to question the accuracy of uncorrected surveys. Claire Adida presented her related work from Benin, in which she experimentally induced citizens to express (non) support for their ethnically ambiguous president (of both Yayi and Nago descent) following a “prime” that indicated his association with one or the other ethnicity. Finally, Nahomi Ichino and Noah Nathan presented their paper – forthcoming in the American Political Science Review – which showed that in Ghana, in areas where the president’s ethnic group constitutes an increasingly large share of the population, individuals from other ethnic groups are likely to vote for him. They argue, plausibly, that this is because those individuals, despite being from a different ethnic group, will actually benefit from the President’s largesse in ways that would not happen if a candidate of their own ethnic group were elected.

Sometimes I leave academic conferences wondering why I do what I do. This one actually left me pretty energized and quite impressed by the ambitious work of colleagues in the field.

Kenya-bound for more Uwezo research

I head off tomorrow for a few weeks of meetings and research in Kenya. I’ll join my collaborators, Lily Tsai and Dan Posner, and our project manager, Jessica Grody to continue our work trying to understand the impact of the Uwezo initiative. We’ll get started in Nairobi, and from there we’ll do some traveling around the country.

In our research so far, we have examined whether the direct provision of information to parents – specifically, their own children’s literacy and numeracy test results as well as information about what parents can do to be more active citizens – actually leads them to higher levels of local involvement in education. Like many other scholars, we are trying to understand if and when the provision of information leads to greater government accountability in terms of the delivery of key services.

For good reason, a great deal of impact evaluation research these days is being carried out using experimental methods. And if the definition of an experiment requires that the analyst randomly assign treatment to subjects, then technically speaking, ours was not an experiment. But fortunately, Uwezo had decided for various reasons to carry out this portion of their initiative to a random sample of villages around the country. Thus, one could not make the claim that there was something systematically distinctive about the villages that received the intervention. In order to estimate the effects of these informational “bundles” we required a control group. So prior to beginning our study, we identified a set of villages that were highly similar to the ones that had been randomly selected by Uwezo. Then we conducted surveys, focus groups, and other research in both sets of villages. We will complete a draft of a scholarly paper on this study fairly soon, and we will be presenting our results at the late August meetings of the American Political Science Association.

As for our estimates of the effect of those “information bundles”…. Well, we didn’t find any substantial impact. But in fairness, we didn’t hold particularly high expectations for an effect. It was important to see if this aspect of the initiative could drive desired outcomes on its own. It can’t. And now, we will investigate whether some of the much larger information dissemination campaigns make a difference.

On this trip we won’t be carrying out any systematic research – just learning about what Uwezo has been doing and what they plan to do going forward. Hopefully, we will finalize some interesting proposals for research in the coming years.