Romesh Vaitilingam interviews Markus Brückner for Vox

August 2010

Transcription of a VoxEU audio interview [http://www.voxeu.org/index.php?q=node/6530] 

Romesh Vaitilingam:  Welcome to Vox Talks, a series of audio interviews with leading economists from around the world. My name is Romesh Vaitilingam, and today's interview is with Markus Brückner, from the University of Adelaide. Markus and I met at the European Economic Association's annual meetings in Glasgow in August 2010, where we spoke about his research on the relationship between aid and growth in the least developed countries of the world.

Markus Brückner:  This is clearly a very policy relevant debate because, of course, that's where the taxpayers' money is and international organizations like the World Bank and the IMF clearly have to know what happens with the average dollar that goes into foreign aid. Unfortunately, empirically, it's very difficult to obtain an estimate of the causal effect that foreign aid has on economic growth and the aid recipient countries, because changes in income per capita in the aid recipient countries are likely to affect the demand and supply of foreign aid. For example, if donor countries act as good Samaritans, then they will increase aid flows during times when the economy of the aid recipient countries is in bad shape relative to times when the economy there is booming.

So, this would mean that economic growth in the aid recipient country itself has an effect on foreign aid. So, this means, then, that the partial correlation between aid and growth doesn't really tell us anything about causal effect. It doesn't tell us anything about the causal effect of foreign aid on economic growth, nor does it tell us anything about the effect that growth has on foreign aid.

So the aid allocation and aid effectiveness literature, in fact, they're well aware of this problem, at least since the 1970s. This issue has been discussed. The issue of simultaneity problem has been discussed in the literature. However, it seems to be the case that even though the literature is aware that economic growth itself may have an effect on aid, there has not really been any within-country time-series evidence on the causal effect that growth has on aid.

And having an understanding how growth affects aid is important, because not only from aid allocation criteria, because it gives us a prior on how large the reverse effect—so to speak, the reverse causal effect—of growth on aid is. And this is important to know when we look at what we call econometrics and ordinary least squares estimate of the effect that aid has on growth. This is to estimate if there is an upward bias to reverse causality or is it a downward bias?

So, in my paper, I take a so called two-step approach. In the first step I estimate the response of foreign aid to economic growth in the aid recipient country, using instrumental variable techniques. And I do this for the least developed countries, because in the least developed countries, these countries are highly dependent on the agricultural sector and also on the commodity exporting sector. So, we can use rainfall variations, and variations in international commodity prices, as instrumental variables for economic growth. So, we can generate plausibly exogenous variations of economic growth, and then see, using these instrumental variable techniques, how foreign aid responded to changes in GDP per capita growth of the aid recipient countries.

And then in the second step, after I quantified with the instrumental variable estimates of the response of aid to economic growth, I adjust the instrumental variable estimation in the growth regressions, where economic growth is regressed on foreign aid to reverse the effect that growth has on aid. And these are my two main findings.

My first main finding is that, based on within-country time-series variation for the least developed countries during the 1996-2000 period, that foreign aid significantly decreased when economic growth in these countries went up. And not only is the sign of this relationship negative, it's also quantitatively large. My instrumental variable estimates yielded, on average, a one percent increase in GDP per capita growth, reduced foreign aid flow to that country by about four percent.

So, this means that aid is highly elastic to GDP per capita growth of the aid recipient country. So, this is interesting from an aid allocation perspective, but it is also highly relevant for the question of, "Is aid effective?" Because it means that when we just regress economic growth on foreign aid, the estimate that we get on aid is, in econometric terms, downward biased, or more economically stated, it understates the effect that aid has on growth.

And so, right now, the status in the aid allocation literature is that aid has no significant average effect on growth. My first main finding, that aid is highly elastic to growth, means that this reverse causality bias that applies that the least squares estimate and underestimates the effect of aid on growth, may potentially be one of the explanations.

And my second main finding is that, indeed, when I adjust for the negative reverse effect of growth on aid, that I find a positive, a significant and economically quite meaningful effect of aid on economic growth. I find that, on average, about a one percent increase in foreign aid to the LDCs, to the least developed countries, during the 1996-2000 period, raised GDP per capita growth by about 0.1 percent.

So, this is, if you think of it like a solid model, with the basic model in economic growth, the standard model, this is pretty much in line with the effect that we find that investment would have on economic growth. A one percent increase in investment, it depends on the capital-output ratio. But more or less, this would be the effect that we would expect, on average, taking into account the capital-output ratio. And indeed, I further explore the effect of aid on investment, and I find, indeed, that when aid goes to these countries, investment picked up.

So it was not the case that all of this aid was consumed, no, some of it did go into investment, and the fact that I find that, on average, it's positive, it's significant and it's quantitatively, indeed, in line with what we would expect from a standard growth model, where part of this aid goes into investment.

Romesh:  Can you try and explain a little more how you think the mechanisms are working, by which growth reduces aid by which growth reduces aid. I mean, when you think about... How does it happen? Does a country, or do donor countries see growth happening and say, "Right. We've had an impact there. Let's reign back and put the money elsewhere." How does it work the other way, because people have said aid just crowds out other activities and goes straight into consumption or goes into the pockets of corrupt dictators?

Markus Brückner:  Right. So clearly, there may be many channels through which aid affects growth. One channel, the classical channel, is just investment. If you believe that there is a financing problem in these countries, that these countries can not just go to the international market and borrow, and that there are many high return projects sitting around, especially in the rural markets--in very many poor African countries, for example, the rural farmers, they just can't obtain credit. They may have a lucrative project, but they just can't get the credit.

So if the aid dollar goes to these farmers, and clearly they can realize projects that have a high return…There may, of course, be other channels, more macro channels. Real exchange rate appreciation. There may be a rent-seeking channel. All these channels, of course, are captured by the estimate of average effect. The important message of that estimate is that taking into account all these channels of potential rent-seeking activity, of potentially just having an effect on labor supply, on the real exchange rate of appreciation. That on average, taking all these potential different channels into account, aid had a positive and significant effect on income per capita levels in these countries.

Romesh:  But the reverse one, the growth on to aid, tell me about how you see that playing out. That seems surprising to me, in a way. You see growth happen year on year, and then you see the aid budget decline.

Markus:  Right. This could have several explanations. It could have an explanation from a donor supply point of view, so then when these countries grow, that means that their income per capita levels increase. Donors see that these countries are doing relatively better. I mean there are many different countries that donors can potentially give aid to. So in countries that are doing relatively well, tf donors have some sort of equality motive, then this could be partially an explanation why some countries that are having a high GDP per capita growth do not receive so much aid.

From the demand side, that could also be an explanation because aid is usually associated with conditionality. This is something that you do not like. It's kind of like saying when you're doing well, why should I take aid? In addition, take this conditionality. While if you're doing bad, you just have no outside option. This is potentially the last resort.

Romesh:  Can your research say anything about the different forms that aid might take, and why some forms of aid might be more effective than others?

Markus:  Yes, those cases. Certainly, there may be food aid that is just given to countries in crisis. This is certainly the kind of aid that we would expect just to be consumed by the people living there. So this wouldn't necessarily have an effect on GDP per capita unless you think that, of course, you need a certain income of nutrients, of food, in order to work. You can even think, even if this aid dollar is just given to these people to eat, well these people need to produce something, so even if it just goes into consumption, these people still work.

So, of course, if work effort is a function of nutrition, then clearly even if there's just a consumption effect, because these people now have enough food to actually work, then even then you can think that it has a GDP per capita effect. This is the case of food aid, where usually people think, "No, it doesn't have so much of a GDP per capita effect."

If it goes into investment, if these are tractors that are going to the country, if it goes into building roads, into building schools, well then clearly you, via the capital build up, you have an effect on GDP per capita.

Romesh:  What about the initial conditions, if you like? You talked about focusing on the least developed countries. Is there an issue around the fact that you might have more impact on a very poor country, or actually you might have more impact on a slightly richer country?

Markus:  Right. So the least developed countries ‑‑ there are 47 of them in my sample ‑‑ clearly they're highly dependent on aid. About 4%, 3 to 4% is a share of aid in GDP. So clearly aid is a very significant variable in these countries. For the least developed countries, clearly they're a very relevant group. For the other countries, they're not in my sample, so it would be very difficult to extrapolate for them. I focus on the least developed countries because there we have a natural experiment to estimate, due to their dependence on the agricultural sector and commodity export. There we have a natural experiment to use the rainfall and the international commodity prices to estimate very unsymmetrical variable estimation, the effect that growth has on aid.

This is the reason why I focus on the least developed countries, because there we this natural experiment, but also because they're a very focal point of foreign aid flows. For the other countries, extrapolation is always an issue about that. My statement is about the least developed countries. They're very important, and for the other countries I think my research can not address that. It's not aimed to address that.

Romesh:  What about the business side? What about the impact of the cyclical things going on in these economies and the impact of aid at different times?

Markus:  Right. The effect of short-run versus long-run effects. I address that, indeed, in my research. One can address that with distributed lag regression, so when you look at the effect in year T and then the effect in year T minus one, T minus two, T minus three. So after three years, after four years. So to address your point, I find that the largest effect is actually on impact, and then the effect declines over time. This means that I find that there's a significant effect on the level of GDP per capita, so a dollar given to that country in the long run, it increases GDP per capita.

But in the long run, what we think of the long-run growth rates, so some sort of Solow TFP residual on that growth rate there's no effect. So there's a level effect of aid, in the sense that I give you a dollar and you use that dollar. You build something. You build a road you can then consume more, in terms of the level of consumption. But in terms of the growth rate, there's no long run growth rate effect.

Romesh:  Final question, Markus. Let's talk about the policy implications. You've sort of got a good news story for aid.

Markus:  I think policy makers, they always have it more difficult than academic researchers in the sense that when they make the decision of whether to give aid to one country or the other, it's always very high context specific. My message is clear and simple: On average, aid works. Clearly for the policy maker, for the IMF, or for the World Bank, or for the United Nations, each project needs to be decided context specific. They know this. My research simply says this to the taxpayer who pays the money, on average, this aid flow significantly helps in raising GDP per capita in the poorest countries in this world. That's an important message, I think.

Romesh:  Markus Brückner, thank you very much.