Arrested development

Books Two young economists attempt to resolve the mysteries of international aid with "natural experiments" but, Bradford Plumer writes, economics doesn't have all the answers.

A man carries a bag of fertilizer as he leaves a grain depot in Sauri, eastern Kenya, 22 March 2007. Sauri is the first "Millenium Village" and part of a United Nations development program, stressing the importance of agriculture, health and education, and aiming to prove that it is possible to eradicate extreme poverty.  AFP PHOTO/Roberto SCHMIDT  TO GO WITH STORY BY Beatrice DEBUT
Powered by automated translation

Two clever young economists attempt to resolve the mysteries of international aid with "natural experiments" but, Bradford Plumer writes, economics doesn't have all the answers.
Economic Gangsters: Corruption, Violence and the Poverty of Nations Raymond Fisman and Edward Miguel Princeton University Press Dh88
The small rural village of Sauri, on the south-west edge of Kenya, would seem unexceptional if it had not been transformed four years ago into a kind of human laboratory. In 2004, experts from the UN and Columbia University's Earth Institute, led by megastar economist Jeffrey Sachs, offered up hefty sums to turn the impoverished town, with its faltering farms and malarial plagues, into a gleaming new "Millennium Village". According to Sachs, poor communities like Sauri have been caught in a poverty trap composed of many interlocking parts, which cannot be undone by sporadic injections of aid: you can build a new school, for example, but if the students remain malnourished and have few job prospects, they aren't likely to learn much. The only hope for such places is a massive influx of aid that can attack all of its problems simultaneously.

In Sauri, at least, the early returns have been positive. In 2005, Sachs - with Angelina Jolie and MTV cameramen in tow - trekked into the village to find test scores up, corn production booming and villagers thriving. So, Sachs argued, why stop there? The old development model erred by pumping aid into Africa without achieving measurable results - but with a fat enough checkbook, the UN can eradicate poverty, reduce child mortality, stem the tide of Aids and malaria. All that's needed is the will.

Not everyone shares Sachs' optimism, though, and the sceptics - led by former World Bank economist William Easterly, argue that foreign aid is too often squandered, either lavished on unproven projects or siphoned away by kleptocrats. Just look at Chad, where the World Bank spent the past decade financing a 650-mile oil pipeline, with the understanding that President Idriss Déby would dedicate the revenue to reducing poverty. Yet soon after the last bolt was tightened, Déby went back on his word and started spending money on border wars with rebel groups instead.

Even in the best-case scenario - when local strongmen aren't lining their own pockets with aid money - economists still only have a dim grasp of what actually works. Easterly, then, suggests that aid donors should set aside grand, well-meaning schemes that frequently do more harm than good, and focus their attention on trial-and-error experiments in search of small projects that can be proven to make a difference.

Although the UN and many aid groups prefer the Sachs approach - optimism has a way of attracting crowds - Easterly's argument has found its share of converts. Since 2003, the Poverty Action Lab at MIT has conducted an array of randomised controlled trials to figure out what development projects are actually achieving. After all, even if a country receiving aid shows signs of improvement, how do you know the aid made the difference? For the past few years Kenya has enjoyed an economic boom and healthy rains: perhaps the success of Sauri owes as much to these factors as to UN largesse. By contrast, a closely monitored experiment, in which randomly chosen test subjects are compared with a control group, is a surer way of getting answers. The MIT lab has found, for example, that handing out free textbooks to students does surprisingly little to boost education outcomes, whereas deworming medicine has a sizeable effect.

Trouble is, it's not always so simple in the real world to carry out controlled experiments. Take bureaucratic corruption. In recent years, some economists have suggested that increasing the salaries of officials might reduce bribery, since a police officer will presumably be less likely to risk his career over a bit of palm-greasing if his paycheck is large enough. But in countries where this has been tried, it's tough to know for sure if the higher salaries are reducing corruption, or whether improvements might be due to other factors. And, unfortunately, it's not really possible to stage a test wherein you pay some lucky officials more than others and then check to see if they're extorting fewer bribes than their colleagues. So what happens here? Do we just give up?

Not necessarily. In recent years, economists have become adept at uncovering "natural experiments" - random events that essentially divide people into treatment and control groups. Say we wanted to figure out whether military service gives people an advantage in the civilian workforce later on in life. A comparison between veterans and non-veterans might be misleading, since the former group could well be self-selecting. So, instead, we might examine American men born in the 1950s who were randomly assigned draft numbers and compare those who were called up with those who weren't. Steve Levitt of the University of Chicago put this forensics method on the map (not to mention bedside tables everywhere) with his best-seller Freakonomics. True, some critics worry that modern economists are burning too many brain cells dreaming up clever natural experiments to answer trivial questions, as when Levitt sniffed out an ingenious way to discover that sumo wrestlers sometimes cheat. But the method can help wrestle those big important questions, too.

Raymond Fisman and Edward Miguel, two young rising stars of economics, apply the Freakonomics approach to the problems of development in their new book, Economic Gangsters. (Levitt even gave them a glowing blurb for the cover.) It's a superbly crafted set of essays that raise the bar for clear, accessible pop-economics writing, and offers an excellent overview of recent research into the corruption, violence and poverty that have long bedevilled the developing world.

Fisman and Miguel earned a blast of internet buzz in 2006 when they published a study of the traffic tickets racked up by UN diplomats in New York City. Since diplomats don't have to pay parking fines, the only thing that might keep them honest, Fisman and Miguel reasoned, would be a cultural aversion to corruption. And, voilà, it turns out the countries that rank high on traditional measures of transparency - like those do-goody Scandinavians - have diplomats who are less likely to flout New York City's parking laws. This sort of natural experiment would seem to indicate that there's such thing as a "culture of corruption". But does it really? A handful of objections come to mind, and, to their credit, Fisman and Miguel consider many of them (for one, double-parking ambassadors tend to come from countries that don't much like the United States - so maybe this is less a measure of cultural corruption than of mere anti-Americanism). Still, the study showed that there are clever ways to get a handle on concepts that economists have had a difficult time quantifying.

In another study, the pair turned their gaze toward Indonesia, where observers of the Suharto regime have long suspected - but could not prove - that corruption enriched the business dealings of the president's son and his cronies. (Perhaps the young Suharto just happened to be a talented CEO.) Fisman and Miguel devised another ingenious natural experiment - they looked to see how far the stocks of companies close to the regime tumbled whenever rumours of Suharto's poor health surfaced. Quite a bit, it turns out.

So far, so fascinating. But these studies don't get us any closer to Easterly's call for rigorous data that can help countries actually tackle corruption - not yet, at least. Fisman and Miguel do venture in that direction with their work on smuggling. They discover that in a country like China, which levies higher tariffs on imported chickens than on turkeys, trade data mysteriously show that there are far more turkeys imported into China than are actually exported by other countries - and, conversely, far fewer chickens. Since the exporting nations would have no reason to fudge their chicken and turkey numbers, the likely conclusion is that Chinese importers are mislabelling chickens as turkeys to dodge the higher tariffs. (Most inspectors can't easily spot the difference.) It turns out this sort of behaviour accounts for a hefty fraction of the smuggling business, and countries can crack down on leakage by closing the tariff gap on similar products?say, men's and women's shirts. This won't lift nations out of poverty, but it is an easy way to ameliorate a lingering problem in poor countries.

In another study with possibly far-reaching consequences, Fisman and Miguel (along with two colleagues at New York University) pored over reams of satellite data to find that sudden drops in national income caused by, say, adverse weather are a better predictor of civil conflict than oft-emphasised factors like ethnic strife. This may help explain why Africa is so prone to civil wars - countries like Chad, Kenya, and Sudan suffer from unpredictable and erratic rain patterns, a problem that has only worsened with the warming of the planet. Niger provides a vivid illustration of the dangers: during severe droughts in 1989 and 1990, the Tuaregs in the north began fighting with the central government over resettlement assistance, a conflict that waned during the abundant harvest rains of the late Nineties and flared up again when the dry spell returned. This relationship could use more study, but it's certainly provocative.

By way of remedy, Fisman and Miguel suggest that drought insurance might be a remarkably cost-effective form of foreign aid - most farmers in Africa don't have the savings to see them through dry spells, and many nations lack the developed insurance and banking firms that could smooth the highs and lows of the harvest cycle. Botswana, for one, has a formal drought-relief programme, and the country has remained stable since the 1960s, with many leaders crediting the programme for the country's better-than-average growth during that period. That sounds promising, but it's not clear these successes can be replicated across the continent: Botswana's government, after all, is fairly honest and can be trusted to administer a drought-relief programme - not the case everywhere.

Indeed, all the clever analyses in Economic Gangsters point back to the same problem. No matter how many economists are now doing useful and meticulous work on corruption and poverty, relatively little of this work has translated into clear templates for development. At one point, Fisman and Miguel look to Sierra Leone for a model of how to pacify and rehabilitate a country after years of civil conflict. Most of the lessons they glean seem intuitively obvious: eliminate small arms, build a strong government, offer public-works programmes for unemployed youths. But as the authors themselves note, even if these precepts are widely accepted, it's not always clear how to apply them in other countries - what works in Sierra Leone may prove an awkward fit for a country like Iraq. Indeed, it's notable that the book's short section on Iraq offers no special insight into piecing that shattered country back together. Instead the authors recite a list of well-known mistakes by the US occupiers, which have already been picked over ad nauseam by journalists. We hardly need economists to tell us that Paul Bremer's decision to disband the Iraqi Army, which swollen the insurgency with thousands of well-armed and angry men, was a mistake.

In the end, that might just be another way of saying that not all of the development problems boil down to economics. Yes, it can describe more thoroughly how people behave and reveal some of the incentives that shape their choices?which, in turn, gives hope that those incentives can be tweaked to produce more congenial outcomes. But there are scores of development issues that can't be reduced to simple incentives, like the long-standing debate about whether donors should bypass dysfunctional states and deliver aid directly to poor people, or whether doing so would retard political development in those countries and leave entire populations in thrall to unaccountable aid organisations. It's not an easy question to answer, and experts are still very far from having a grand theory about what works in economic development. As Easterly himself put it in a recent introduction to a series of essays about foreign aid, "Economic success is always very uneven and unpredictable, across almost any possible unit of analysis one might consider." What's valuable about Economic Gangsters is that it essentially confirms this view, and underscores the need for caution. There are many questions economists can shed light on, and a great many more remain shrouded in darkness.

Bradford Plumer is an assistant editor at The New Republic.