I’m happy to announce that I’ll be spending the 2013–14 academic year at American University’s School of International Service. It will be nice to be able to stay in DC and keep working with such great colleagues and I’m really excited to be teaching courses on African political economy, African politics, and international development. I’ve posed the syllabi for the fall courses on my teaching page.
This semester I’ll be presenting a poster on foreign aid politics in Kenya at APSA on Friday, August 30th at 2pm. I’ll also be presenting work on the impact of colonial institutions on education in Ghana at ASA on November 23rd. If anyone wants to meet with me at either conference, be sure to send me an email or a message on twitter.
Below is a simple map of district-level aid to Malawi since 2000. The information on aid came from AidData and the map file came from the UNSALB. The code was written in R. There is nothing overly complicated about the map, but it was one of the first times that I sat down with R and figured out how to make a map.
I learned a few lessons along the way:
- While the geocoded data from AidData are amazing, I really wish that they broke down disbursements by year. Unless I’m missing something, then all that you get in the Malawi dataset are date of commitment, planned date of completion, and amount disbursed to date. I don’t want to complain, but data by year would be really useful to researchers.
- It is hard to find trustworthy shapefiles of maps with subnational boundaries. I tried using a few dodgy ones (look at the odd boundaries in the south) before stumbling across the UN Second Level Administrative Boundaries data set project. I will start there next time that I need maps.
- If you are mapping in R, maptools is your friend. If you are merging datasets, then I found it far easier to use join (part of the plyr package) than merge. I still am very much learning R, so interpret my recommendations accordingly.
On marginal revolution I came across a new Stata package by Damian Clarke. It makes it very simple to show both maps and time series graphs of World Bank data. Even better, because it uses wbopendata it can work with any of the World Bank’s open databases. This gives you access to thousands of variables.
After installing (type “ssc install worldstat”), I graphed net ODA as a percent of GNI to Africa (type “worldstat Africa, stat(DT.ODA.ODAT.GN.ZS)”). The cleaned up version of the graph is presented here. Damian deserves a lot of praise. He’s made it very easy to make beautiful, useful representations of data. I can see myself using this frequently for slides.
Foreign aid is very unpredictable. This unpredictability makes recipient government planning more challenging than it ought to be and lessens the value of aid. One of my research interests is the effect of aid volatility or large swings in aid on different political outcomes in recipient countries. One challenge of this work is conveying the magnitude of aid volatility.
The figure above is my current best attempt at showing just how volatile aid can be. Each year shows two bars, each representing an aid change in a country in sub-Saharan Africa in that year. The blue bars show the largest aid increases and the red bars show the most extreme aid decreases. I’m not particularly good at presenting data, so if you have criticisms please pass them along: @ryanbriggs.
The data used are ODA disbursements and I subtracted out technical assistance and debt relief. The sample of recipients contains all countries in sub-Saharan Africa between 1961 and 2008. I measured aid volatility as percentage changes from the aid level in the previous year. This way of measuring aid changes will produce large values if the previous year had an unusually small amount of aid. For example, Zimbabwe went from 0.46 million in aid in 1979 to 227.42 million aid in 1980, for an increase of roughly 49,000 percent. To avoid these situations, I dropped any observations where a country received less than 50 million dollars of aid. This means that all of the percentage changes in the figure above are happening on a base of at least 50 million (2008) USD of aid.
The figure reveals that each year some countries experience very large changes in aid. Social scientists often focus on averages—and average aid volatility is large enough that it is a problem—but the tails of this distribution are really scary. Each year of the figure shows two instances of real countries getting knocked around by aid changes, so I’d say that tails of the distribution matter. Some countries went from over 50 million in aid one year to less than 0 the next. Many other countries saw their aid double or triple in one year, and one saw a six-fold increase. How can a government plan future expenditures if a large fraction of their revenue can either quadruple or be cut in half. This is what is at stake in the calls for donor coordination to reduce aid volatility.
Update: For those who asked, I don’t have the data to measure aid changes as a percentage of government revenue or expenditure. To the best of my knowledge, these data do not exist across time for most of the countries in SSA. If I normalize aid changes as a percentage of GDP, I often see swings larger than +/- 5% of GDP.
A final note: it is hard to believe that these kinds of aid changes don’t affect recipient country politics, right?
Randomized control trials (RCTs) are the gold standard for some kinds of development interventions, but they are naturally limited in how broadly they can generalize to new contexts. This point was recently raised in an excellent debate and discussion between Abhijit Banerjee and Angus Deaton.
I mentioned this before and drew a distinction between doing RCTs with people and doing RCTs with larger social units within countries. Countries are far more heterogeneous than people, and so it is easier to generalize from some sample of people to the rest of humanity than it is to generalize from some sample of villages to all villages.
The reason I’m bringing this up again is that I didn’t realize how far medical science has moved towards increasing internal validity at the cost of external. From the Aid on the Edge of Chaos blog:
Ironically, “the canonical tenets of ‘scientific excellence’” are threatening to undermine the whole enterprise. One rather shocking – for me, at least – example relate to the latest developments in research on mice, where a lot of resources and funding have been poured into the cloning of genetically identical animals in order to enable fully controlled, replicable experiments and rigorous hypothesis-testing. However, the findings of this research have turned out to be useless when applied in the real world of diversity and change.
I recommend reading the whole article. It is sadly comforting to know that every science has these problems.