Adam and O'Connell (1999), "Aid, Taxation, and Development in Sub-Saharan Africa", Economics and Politics, Vol. 11, No. 3, 225-253, (ungated)
This is the earliest paper I know of to use a form of the theory. Recipient governments maximize the utility of the winning coalition by choosing tax rates and transfers to the winning coalition. Even if transfers are 0, there are still taxes to pay for needed expenditures. The Laffer curve limits taxes to the revenue maximizing rate. A sufficiently representative government does not make any transfers. Donors have fixed resources, maximize everyone's utility, and give only if it is effective. They show that
- Small increases in unconditional aid to a representative government lower taxes, encouraging further investment. This is better than aid given directly to the poor as consumption.
- All unconditional aid to governments that are already making transfers go entirely into transfers and giving aid directly to the poor for consumption would be better.
- For any government that doesn't care about every member of society, there is some level of unconditional aid that will induce the government to begin making private transfers rather than lower taxes.
- With conditional aid, donors can lower tax rates and recipients are willing because they can increase transfers.
This paper differs considerably from the last while sharing the fundamental theory: There are no taxes and consumer behavior is irrelevant; donors don't care about the welfare of these non-existent consumers, "donor leaders give aid ... in return for policy concessions ... that advance the interests of political elites" in their nation; in fact, donors and recipient governments are identical in that they only care about their respective winning coalitions and the only meaningful difference between donors and recipients is average coalition size and availability of finance. All politicians seek to maximize their probability of staying in office, which depends on making their winning coalition happy.
Crucially (from a story perspective), they assume that policy concessions are public goods for donors and public bads for recipients. That is, the World Bank and IMF and the rest are tools of political hegemony and the Washington Consensus was specifically designed to be bad for developing countries.
They come with a number of unusual conclusions from the model and estimations, but their proxies are questionable enough that time and future iterations of this exercise will be needed to tell us more. More fundamentally, aid increases with GDP/capita up to about $1000 and then falls again as it gets more expensive to buy policies. This leads them to conclude that most OECD countries do not seem to be giving aid for humanitarian reasons, and they are show that the US behaves very similarly to the other OECD countries - the US just has more resources to spend buying policies from wealthier governments (Israel and Egypt) and the results hold even when those countries are taken out. "Corrupt" uses of aid by small, autocratic governments are not a bug, but a feature of the system that are what enables richer countries to buy policy change. Interesting throughout.