Despite trillions of donated dollars, the lot of the world’s poor has not greatly improved. Economics has failed: time to try science
FOREIGN aid is a waste of money. At a time when western governments are struggling to cut their spending we should be reducing aid budgets, not ring-fencing them. Such views have been widely aired in recent months, and they cannot be dismissed as mere political posturing or self-interest. Those who believe aid is ineffective argue that it hinders economic growth by encouraging corruption and undermining civil society. Although this view is not widely shared, there is a problem. The $2 trillion in aid which western governments and charities have directed to developing countries in the past 50 years has not done its job.
“The countries at the bottom coexist with the 21st century but their reality is the 14th century: civil war, plague, ignorance,” says development economist at the University of Oxford. The past half-century has seen some improvements: fewer children are dying of curable diseases and the number of people with a daily income of less than $1 (at today’s value) has fallen by 20 per cent to about 900 million. But 8 million under-fives still die needlessly each year, 100 million people are without access to clean drinking water, and the income of the poorest billion people has actually dropped by 5 per cent in the past two decades. Why has so much goodwill met with so little success?
Advertisement
The problem, according to those working in development, is not whether to give aid, but what to spend it on. Should we be investing in education or in peace initiatives? In systems to build trust and cooperation, or in some other area? We just don’t know. Theories about development emerge and programmes get tried, but before they can be properly evaluated the ideas go out of fashion and agencies move on to the next project. “We do the same things over and over and over again, without learning from the experience,” says economist of the Massachusetts Institute of Technology.
Duflo is determined to change that. Earlier this year, in a talk at the TED conference but radical approach to bring more scientific rigour to development work. She and her colleagues at MIT’s are pioneering the use of randomised experiments to find out what actually works in a given situation. “The biggest myth in development aid is that there is a magic bullet,” says Duflo. But the solution to a particular problem in a specific situation may be unique and counter-intuitive. “The best thing to do with aid money is to use it in an experimental way.”
The idea that the scientific method can come to the aid of development is gaining ground. In 2008, Collier, a former director of development research at the World Bank, published The Bottom Billion, in which he used statistical analysis to identify the four factors that correlated most closely with poverty in African nations: civil war, bad governance, mineral riches such as diamonds or oil (which paradoxically make it harder for countries to develop competitive export markets) and being land-locked with bad neighbours. Citizens of countries marked in these ways are caught in socio-economic traps which they cannot get out of on their own. The links between poverty and war, Collier suggests, or between abundant resources and government corruption, may be too strong to be overcome by anything short of external intervention in the form of aid. “Societies can’t readily break out of these traps for themselves,” he says. “They will be stuck in poverty unless we help them far more than we have to date.”
Collier admits that the complexity of the problem doesn’t lend itself to easy solutions. Nevertheless, he believes that the analysis he outlines in The Bottom Billion can help governments and aid agencies wondering where best to direct resources.
Collier’s statistical analysis and concrete suggestions for change have gained some influence among aid organisations and western governments, but some experts remain unconvinced that the causes of poverty, or the steps likely to take people out of it, are really so clear-cut. Development economist of New York University argues that throughout the history of development aid, successes and failures have rarely conformed to any simple picture. “After half a century of research,” he says, “it’s safe to say that economists still don’t have any real understanding of why one country has a high rate of growth and another doesn’t.” The most successful development programmes, he argues, work by searching for the best approach, exploring possible solutions by trial and error, rather than by implementing rigid plans driven by monolithic ideas.
Such thinking resonates with what Duflo and her colleagues are doing. Simple theories have little chance of guiding successful aid programmes when people are often poor for many reasons, and remain so for many reasons. In particular, Duflo points out, human intuition is a bad guide to understanding a problem.
For example, giving villagers free mosquito nets would seem like an obvious way to reduce malaria. However, several projects in Africa that have followed this strategy have had limited success because villagers felt they could put the nets to better use in fishing. And plans that seem good on paper in the offices of aid agencies can be missing a vital piece of the picture. Easterly cites a project sponsored in part by the World Bank to teach agricultural techniques to farmers in Lesotho. After it failed, he recalls, project managers complained that the local people were “defeatist”, but the plan had overlooked the region’s poor weather that made farming almost impossible. “Details matter infinitely,” says Abhijit Banerjee, who along with Duflo and Sendhil Mullainathan founded the MIT Poverty Action Lab in 2003. “In my experience, even when you talk to very competent, well-meaning organisations, that is the step where they fall down.”
Trial and error
That’s where the experimental approach comes in. It uses controlled randomised experiments – akin to the drug trials used in medicine – to test combinations of simple interventions aimed to improve things like education, rates of immunisation or the sustainable use of local resources. The idea is to look at the data to determine what works, rather than relying on intuition or supposedly deep economic theories. “Experiments make it possible to vary one factor at a time, and give us a chance to isolate causal effects,” says Banerjee. “It’s a slow and deliberate process of discovery, but it’s the only way we know that works, and we are seeing more and more of this work around.”
In , for example, Duflo, Banerjee and colleagues carried out randomised experiments in 134 villages to test different ideas for promoting childhood immunisation. They organised monthly immunisation camps in one-third of the villages – making immunisation more convenient for parents – and in another third also gave parents an incentive in the form of 1 kilogram of lentils for each child immunised. The remaining villages acted as a control, with no intervention: parents in these villages could still immunise their children, but had to take the initiative and travel to centres run by the state, with no guarantee that these would be open when they arrived. The results showed that the camps alone more than doubled immunisation rates and the lentil incentive multiplied them by six, at an extra cost of just 40 rupees (less than $1) per child.
Other researchers have had similarly striking results. In 2004, in schools in Kenya, economists Edward Miguel of the University of California, Berkeley, and Michael Kremer of Harvard University tested different ways of reducing absenteeism in primary schools. They found that a , costing just 49 cents for each child for a year, was about 20 times as effective as any other measure they tested, including hiring an extra teacher.
“I think this is very exciting work,” says Easterly. “Humans are suckers for finding patterns where none really exist, like seeing the shapes of lions and giraffes in the clouds. It’s the same with economists. Careful experiments help us overcome this problem.”
But if the approach is so effective, why has it not been wholeheartedly adopted? Duflo has one theory. Politicians and donating organisations don’t like negative results, for understandable reasons. “If you work for a government or big aid organisation, you cannot admit that you got it wrong, ever. If you’re a government you won’t get re-elected; that is a real problem.”
Meanwhile, Banerjee points out that aid organisations are gradually coming to recognise the value of forging long-term relationships with researchers to identify more effective ways of working using an evidence-based approach. Today the Poverty Action Lab links nearly 50 researchers and more than 100 aid organisations around the globe in development programmes that carry out randomised evaluations.
Of course, those who argue that aid money is wasted may not be too keen to see it used in experiments. Others may feel uneasy about the ethics of experimenting with the lives of the world’s poorest people. But randomised evaluations are considered essential in medical research, so why not take the same approach when trying to find the most effective ways to deliver aid? “Take risks and accept failure,” says Duflo. “For governments this is always hard, but donors are allowed to fail. They should act as venture capitalists, trying things, letting things fail, but documenting either success or failure so we can learn from the mistakes.”
“Donors should act as venture capitalists, trying things, letting things fail, but documenting everything so that we can learn from the mistakes”
