Does foreign aid buy votes for bad governments? This study from Uganda shows the opposite.

A whopping 40% of Uganda’s government budget comes from foreign aid. This is a regime that is getting more and more autocratic, going the way of an Ethiopia or (I fear) a Zimbabwe. So of all the angst about aid, and critiques of foreign assistance, it’s surprising that I don’t hear this one more often: is foreign aid propping up bad guys? It seems irresponsible not to know the answer.

I would have thought the answer was “of course we are propping up thugs”, but data from one program in Uganda points in the opposite direction: people who got a big government grant for their business worked to get the opposition election. I don’t know exactly why, but it looks to me like a little increase in wealth freed people from patronage machines.

A few years ago I evaluated a fairly successful government employment program in Uganda, where cash grants of about $400 helped young people increase their self-employment and earnings by about 40%. At the time we also collected data on how much people participated in the 2011 elections, what parties they liked and disliked, and other political behavior. But like a lot of academics I have more data than I can analyze. Plus once I found this result I never knew what to make of it. So it took years before I could write up a real paper with my two coauthors, Mathilde Emeriau and Nathan Fiala. It’s now up.

The cash grants went out in 2008. This was a pretty meritocratic program. It didn’t target political supporters, it had little pork, and the government couldn’t take it back. The Ugandan government was hoping that good development policy would build its political support in the north of the country.

But instead of rewarding the government in the 2011 elections, compared to the random control group, the people who actually got the grant increased their opposition party membership, campaigning, and voting. Opposition voting went from 12% to 16%, a one third increase.

We went through a bunch of possible explanations. As with most experiments, it’s hard to figure out why something happened, especially if the result was unexpected. (We’d geared our survey questions to understand the opposite result.) But we did notice one interesting pattern: higher incomes are associated with opposition support, and income changes seem to account for a god part of the treatment effect on voting.

This possibility has been dangled out before. Beatriz Magaloni has some work on Mexico arguing that financially independent voters are less dependent on favors from the ruling party. Nancy Hite has some unpublished work from the Philippines suggesting that microfinance untangles people from politicized loan networks.

I wonder if what we’re seeing in Uganda is a bigger phenomenon: that financial independence frees the poor to express their political preferences publicly, since they’re less reliant on patronage and other political transfers. If so it’s a micro-level version of an old fashioned story about how democratization follows from economic development.

Given how much money countries give away in aid, this seems like an important question to answer. One easy way to add to the evidence is the huge number of randomized trials of anti-poverty programs. Simply adding on post-election questions from the regional barometer studies (Afrobarometer, Latinobarometer, etc) would go a long ways to increasing the evidence.

Downstream studies are also a good idea, where you go back for another round of survey data after an election. That’s how Hite got her Philippines data. And we will go back to Uganda next year for the 9-year follow up, and will get information on the 2016 elections while there. This is low hanging fruit for grad students and junior faculty.