Our Word is Our Weapon

The Globalization Institute have a report out called More Aid, Less Growth (I was going to say “a new report” there, but actually it’s not new, it’s a re-hash of this old article for the Cato Institute’s journal). It claims to show that “for every 1% increase in aid received by a developing country, there is a 3.65% drop in real GDP growth per person”. But in fact it is so seriously flawed that it tells us nothing at all. Here’s how.

Firstly, the report author Tomi Ovaska includes as “aid” humanitarian assistance that is usually by its nature negatively correlated with growth, as it is mostly given to countries who have suffered some serious disaster. Mixing this up with non-emergency aid will automatically reduce the strength of any positive correlation between aid and growth, as conclusively demonstrated here.

That paper (”Counting chickens when they hatch”, by the Center for Global Development), also highlights the second flaw in Ovaska’s analysis. He breaks up his data-set into five periods of four to five years each, and tries to measure the effect of aid in a particular period on growth in that period. But if aid has any positive effect on growth that begins or simply lasts beyond four or five years (for example by improving child educational levels or the quality of political and administrative institutions), the short periods used by Ovaska will not pick it up and may even return a negative correlation. The CGD paper again overcame this problem by analysing only the effects of aid that we would expect to have a short-term impact, “including budget and balance of payments support, investments in infrastructure, and aid for productive sectors such as agriculture and industry.” When they make these two adjustments, they find a strong and significant positive effect of aid on growth. Ovaska makes neither adjustment, so it’s hardly surprising that he finds no such correlation.

The problems don’t stop there though. The equations Ovaska uses to measure the effect of aid on growth control for the level of government consumption, investment, education and life expectancy in a country. But each of these are channels through which aid impacts on growth - by financing government spending, investment, education and health services. So by factoring out the effect of aid on growth through these channels, Ovaska massively reduces any chance of finding a positive effect of aid on growth*.

Taken together, these problems render Ovaska’s analysis pretty much meaningless (a bit like that recent IMF analysis, as Owen Barder pointed out). And yet I predict that the Globalization Institute and various other right-wing think-tanks (and media) will continue to talk it up like it’s the gospel truth. Already it has inspired at least one slavering fool to denounce the ’socialist’ Make Poverty History campaign on the grounds that it will “cause more poverty and more deaths than would otherwise have occurred”. Creeps like Will Stephens seem to have no interest in whether what they say is actually true - the point is to bash lefties wherever and whenever possible. I still like to think that the Globalization Institute has better motives, but in promoting this seriously flawed report and deliberately ignoring the robust and overwhelming evidence to the contrary they are spreading misinformation, simple as that.

Which brings me on to a related point. Globalization Insititute head honcho Alex Singleton here tries to answer a criticism of the report I made on the Samizdata site, but without attribution or link. Now, the GI has never been good at providing enough links to enable people to make up their own minds (and they would certainly never link to this post or any other on this site), but I wonder if it’s also partly because if he had linked to my comment people would have noticed my adjacent remark that “a recent summary of the evidence found overwhelming support for the argument that aid is good for growth”, which referred to this study by Mark McGillivray (discussed in my previous post). Here’s another prediction - the Globlization Institute will never acknowledge the existence of the McGillivray paper- if they do I’ll happily donate �50 to their whip-round (as long as I get the promised invite to the “launch party later this month on the rooftop bar at the fashionable London venue, Soho House”).

Incidentally, Alex goes on to say that

The report’s author, Dr Tomi Ovaska of the University of Regina in Canada, tells us that while many studies in the past thirty years have found aid fostering economic growth, this has often had to do more than anything else with poor quality data, misspecified models, and/or a small data set where a just few countries (China and the Asian Tigers) drive the results. If one looks at the latest studies using more advanced statistical techniques and better data, the unanimity is fast eroding.

This is pretty much exactly backwards. Firstly, Ovaska’s sample is actually smaller than the CGD paper I’ve already mentioned, so that’s another grounds on which, by his own admission, we should consider it less robust. Secondly, it is not the case that research is increasingly sceptical about the merits of aid. In fact, the opposite is happening - McGillivray says that “the clear, unambiguous finding of practically all empirical studies conducted over the last seven or eight years”, that “Aid now appears to work in the sense that per capita economic growth would have been lower in its absence”, is “a remarkable turnaround in the literature on aid effectiveness, which for decades provided rather inconclusive, often contradictory findings”.

*A more trivial point, but perhaps an illuminating one, is that Ovaska gets the basic aid terminology wrong. He talks about using a measure of aid called “Efficient Development Aid”, but in fact the proper term is “Effective Development Aid”.