Why Foreign Aid Cannot be Regressive?

Branko Milanovic explores the effects of aid at both ends of the transfer.

There are many arguments against foreign aid. Angus Deaton has forcefully argued that foreign aid, by providing resources to the governments  outside  normal taxation, breaks the link of reciprocity which exists between the tax payers and the state. It thus weakens national institutions, and insulates the rulers from the population. Dambisa Mayo believes that aid fosters corruption. Bill Easterly has attacked aid because it goes to the governments and not to the private sector. Many of these recent critiques are reminiscent of the arguments made decades ago by Peter Bauer (“Dissent on development”), and even earlier by Ludwig von Mises.

Now, when the European Union seems to be considering a significant increase in aid for Africa—led this time not by humanitarian concerns but by the well understood self-interest as reduction of migration is hard to imagine without a substantial convergence in incomes between Africa and Europe—it is worth pointing out that one argument against aid cannot hold. This is the argument made sometime in popular press (and at times, in academe too), that aid to the poor countries is just a transfer of resources from the poor people in rich countries to the rich people in poor countries. This is what is in economics called a “regressive transfer”. (“Progressive” transfer is what we desire to achieve: tax a richer person and transfer the money to a poorer.)

The data on global income distribution clearly show that this particular argument against aid (“regressiveness”) has no validity. (I have already written about this in greater detail here.)

To see this consider as an illustrative example incomes along the entire income distributions of the Netherlands and Mali. The graph below shows annual per capita income levels (in 2011 PPP dollars) at different percentiles of Dutch and Malian income distributions. As can be readily seen, even the poorest percentiles in the Netherlands are richer than all Malian percentiles, including the top 1%. In other words, the two distributions do not overlap: the Dutch distribution starts at an income levels where the distribution on Mali had already ended. (The dashed line is drawn at the income level of the poorest Dutch percentile.)

Now, this fact alone would sufficiently indicate that if one taxes a Dutch person and transfers that money to a person in Mali, it is very unlikely that he would make a regressive transfer. But the probability of a regressive transfer is even much smaller than this first impression implies. One has to ask: from whom in the Netherlands would the expected tax euro (that is used for aid) come? In other words, imagine a big bowl where all Dutch income taxes are collected with, on each euro received, written down the income level (or percentile position) of the person who paid it. If the Dutch system were based on a poll-tax, everybody would contribute the same amount, and the expected percentile position of the taxpayer whose euro goes to Mali would be exactly at the Dutch median income. If the Dutch system were such that the tax rate is the same regardless of income level (the so-called “flat tax” system), the expected euro would come from the person at the Dutch mean income level. (Formally, t=αy where t=taxes paid by an individual, α is the tax rate, and y=pre-tax income. Then, the random tax euro is received from the person with income: E(y(t/T)=E(αy2/T)=α(E(y))2/αE(y)=E(y) where T=total taxes.) The person with the mean income in the Netherlands is located at the 63rd  percentile of the Dutch income distribution.

But this is not all. The Dutch system of direct taxation is progressive, meaning that the tax rate increases with income level. Now, richer people will contribute not only more euros absolutely, but also more euros proportionally. Using LIS-provided micro income distribution data, we can calculate the average tax rate by percentile of the Dutch distribution (shown in Figure below)  and also calculate that a random euro that would go for aid in Africa would come from the person who is at the 73rd percentile of the Dutch distribution. (This is the other way of saying that ½ of Dutch taxes are paid by the top 27% of recipients.)

We can now go back to the first graph and look at how likely it is that the beneficiary of the Dutch aid to Mali will be richer than the person at the 73rd percentile of the Dutch income distribution. Since even the Malian top 1% has an income that is vastly inferior (it is about one-fourth) than the income at the Dutch 73rd percentile, that likelihood must be quasi nil. In other words, even if we assume, rather extravagantly, that the only beneficiaries of Dutch aid to Mali are the local top 1%, the transfer would be still progressive. It could be of course argued that if we were to slice the Malian distribution in ever smaller pieces, there may be eventually such a small slice, say the richest five or ten persons in Mali, who would be better-off than the Dutch 73rd percentile. But this is simply saying that—working behind the veil of ignorance of who the beneficiaries of aid are—the likelihood of a regressive transfer is somewhere in the vicinity of one-hundreds of one percent. The onus is therefore on those who argue that foreign aid can be regressive to show that the beneficiaries of aid are people who are probably part of the top Malian millesime or so. This, based on whatever we know on the effects of aid, seems utterly unlikely.

To conclude: aid is not a panacea, it could be even detrimental to the recipient country, but it surely does not (in most cases that we normally have to deal with) represent a transfer of purchasing power from a poor person in a rich country to a rich person in a poor country. This shibboleth needs to be laid to rest.

This post first appeared on Branko Milanovic's blog.

Photo credit: Foter.com  CC0 1.0 Universal (CC0 1.0) Public Domain Dedication