Book Reviews

Small Change

The big theories of economic development turned out to be wrong. Finding out what works on the ground offers a path to curbing global poverty.

By Charles Kenny

Tagged DevelopmentEconomicsPoverty

Poor Economics: A Radical Rethinking of the Way to Fight Global Poverty By Abhijit Banerjee and Esther Duflo • PublicAffairs • 2011 • 303 pages • $26.99

We’ve cycled through many theories of economic development over the past 60 years, with some championing investment in infrastructure and industry, others stressing spending on education and social welfare, and still others touting the expansion of trade and the market. But none of these models has fit particularly well with the facts of global growth—in particular the problem that some countries appear to be persistently poor.

That’s perhaps why the dominant theory of development today is that slow-changing institutions determine failure or success. Based on studies of world history, growth economists conclude that whether a country today is rich or poor, democratic or autocratic, developed or underdeveloped can be traced to factors such as the mortality rate of colonial settlers, which among them had oceangoing ships in 1500, or even the shape of the continental landmass that was home to their ancestors. The mechanism by which history determines success is through the laws, rules, and customs that emerged as a result of who was the colonizer and who the colonized, and how local conditions shaped the form of colonial rule. And these laws, rules, and customs are very slow to evolve.

If history is destiny, and development is due to slow-changing institutions, then there would seem to be little hope for countries that were on the losing continents back when Pangaea split up. Because of this theory, doubts about the effectiveness of foreign assistance have never been stronger—if it is all about history, the role for aid is unclear. And even if there is progress, it will be slow and halting. Dreams of a world free of poverty will remain just that.

And yet, over the last 20 years, the proportion of the world living on a dollar a day or less has halved. The percentage of children fully vaccinated against childhood disease worldwide climbed from 5 percent to 80 percent between 1974 and 2000. Partially as a result, child mortality in developing countries has dropped by a third since 1990. The children who survive are far more likely to be in school. Primary enrollment has climbed from 47 percent to 87 percent since 1950. This progress has been truly global—Africa, for instance, has seen literacy rates more than double since 1970.

How to reconcile all of this rapid, ubiquitous, historically unprecedented global progress with an empirically supported theory that suggests such progress should barely happen, if at all? Many of the clues are to be found in Abhijit Banerjee and Esther Duflo’s wonderful new book, Poor Economics. Banerjee and Duflo, co-directors of the Poverty Action Lab at MIT, show that a whole range of small and cheap interventions can make a big difference to the quality of life, and even the quality of institutions. These interventions are behind the unlikely progress that we’ve been seeing.

Take three of the examples they offer regarding education. Providing iodine capsules to pregnant mothers is an intervention that helps brain development in fetuses. It costs around 51 cents per dose—and leads to kids who stay in school about five months longer because they are cognitively better able to learn. De-worming pills that cost $1.36 for a course of treatment have such an impact on school attendance that they raise lifetime earnings of treated kids by as much as $3,269. Or just publishing in local newspapers the amount of money that is meant to get to Ugandan schools from the central government increased the proportion of funds that actually arrives at the school from 13 percent to 80 percent. It doesn’t have to take a government full of Franklins, or a treasury full of Benjamins, to see dramatic improvement in a country’s education (or health or infrastructure) outcomes.

Banerjee and Duflo are most widely known for introducing the randomized controlled trial (RCT) to the field of development economics. This technique, originally used for medical trials, tests the efficacy of development projects by taking test subjects and dividing them by lottery into groups that get the treatment or intervention and those that don’t. When the division between those who get the treatment and those who don’t is random, it is fair to assume that any difference in outcomes between treated and untreated test subjects is due to the intervention itself. So, if many more kids get vaccinated in randomly selected villages where parents received a reward than in randomly selected control-group villages where they didn’t, one can conclude that the reward was responsible for the higher rates of vaccination.

Banerjee and Duflo’s evaluations have had them spending much time in developing countries, devising experiments and working with government officials and civil society organizations across a range of sectors. They’ve been widely lauded for their work—Duflo won the John Bates Clark Medal in 2010, given to the best economist in the United States under age 40 and often seen as a stepping-stone to a Nobel. Accessible and largely jargon-free, Poor Economics is in large part a summary of what Banerjee and Duflo have learned from the development experiments that they and their colleagues have run.

A central theme of Banerjee and Duflo’s work is an effort to take people as they are and the rules as they might be (to misquote Rousseau). And that means much of their work is about better understanding people—the poor themselves, as well as the government officials and non-government actors who provide them services. What emerges is a rich understanding of the poor not just as agentless, passive recipients, but as people who make complex and demanding choices every day. Poor people are people, too. They want sweets along with spinach, they make mistakes, and they suffer from the usual range of cognitive biases. But because they are poor, and because they live in poor countries, the help they get in making these decisions is limited, and the cost of mistakes is high.

For example, to return to the vaccine study, Banerjee and Duflo point out that in rich countries parents aren’t allowed to send their kid to school unless they can show that junior has completed his vaccine regimen. No such guide to correct behavior is available to the vast majority of the world’s poor people. In villages in Udaipur, India, the full child-vaccine regimen was available for free to kids who were brought to regular monthly clinics. And yet fewer than one-fifth of all kids were fully vaccinated. Given that up to three million people—the great majority of them children—die worldwide each year from vaccine-preventable diseases, low vaccination rates are a significant tragedy. But parents have to make the effort to go to the clinics and wait in line to vaccinate their kids. With the benefits of the vaccine not immediately obvious and knowledge of germ theory lacking, it’s not surprising they often stay at home instead.

The good news, according to the authors, is that a small incentive can easily change their minds. An RCT they conducted found that a parental reward of a two-pound bag of dried beans, worth less than $2, was enough to dramatically increase immunization rates in the villages. And this effect was observed not just among those in the village. When people from other villages saw mothers and fathers taking their kids to get vaccinated, they jumped on the bandwagon. A “new normal” of vaccination began to develop, so that the default course of action was no longer to stay at home, but to get children their shots.

Just as parents need to be nudged to do the right thing, so too do service providers. Bureaucrats who are often corrupt and malingering feature heavily in Poor Economics, from teachers, doctors, and nurses who don’t turn up for work or don’t do anything when there, to police who can’t be bothered to register crimes. Private practitioners come under scrutiny, too. In a survey of medical practice in Udaipur, for instance, Banerjee and Duflo found that diagnostic tests were performed in only 3 percent of patient visits to private doctors. However, patients were given a shot in two-thirds of the visits, meaning that many people were getting unnecessary—sometimes harmful—treatments. The clients wanted their shots, and doctors eager to collect their fees were happy to oblige, whether or not it was the right thing to do.

But again, there is an upside. With the right incentives these same malingerers can deliver quality education and health care. In one RCT in India, the authors found that teachers who appear to make little effort to teach and are often absent from the classroom during the school year can still be highly effective volunteers in summer reading programs. The difference was that during the summer they were asked to teach basic literacy rather than rush through an overly demanding syllabus, all while being watched by NGO staff as they did their work.

That small changes in incentives can make such a big difference in outcomes give Banerjee and Duflo hope that the provision of public services can be improved. Returning to education, they argue that there could be a huge impact on learning if teachers were given the incentive to focus on imparting a basic set of skills to every child rather than trying to complete a curriculum designed for an educational elite (and thereby abandoning most kids as hopeless cases early in the process). Such a comparatively simple change in curriculum design could have a dramatic impact, without the need for a complete overhaul of national governance. The above-mentioned program in Uganda that publicized the amount of money schools should be getting from the central government is another example of a small, innovative solution that can have major consequences. Banerjee and Duflo note that if “rural school headmasters could fight corruption, perhaps it is not necessary to wait for the overthrow of the government or the profound transformation of society before better policies can be implemented.”

The book offers lessons in other areas as well. On population control, Banerjee and Duflo argue that people have more kids when adult offspring are the surest safety net in old age—so social safety nets and pensions may be a key to lower fertility rates. Looking at microcredit, the authors contend that most poor people—like the rest of us—really just want a steady job, and that setting up their own very small enterprise is what they do when there’s no other option. That means relatively few borrowers will become rich through small loans at high interest, however useful such services might be. In turn, that suggests microfinance can play a role, but is unlikely to be a silver bullet for poverty reduction.

With regard to institutions and governance, Poor Economics suggests that there is more to improved outcomes than the veneer of participation. The authors’ findings affirm over and over again that knowledge really can be a powerful tool for change. Information about candidate positions in elections makes voters less likely to cast ballots along ethnic lines. Knowledge that women can be strong leaders—spread more rapidly by the requirement that some village government leadership positions have to be held by women—leads to more votes for women candidates after that requirement is removed.

Poor Economics argues that while policy-makers and bureaucrats often do the wrong things for the wrong reasons (to line their pockets, as it might be), they sometimes do the wrong thing for the right reasons. They are trying to do good, but ideology, ignorance, or inertia leads them to make poor decisions. Knowledge of what works can help in those circumstances, and that is precisely what Banerjee and Duflo have been trying to collect over the past ten years.

Banerjee and Duflo’s methods do have their critics. Harvard’s Lant Pritchett has complained that “many, if not most, of the consequential questions of development economics are simply not susceptible” to RCTs. An RCT will never tell you why Côte d’Ivoire descended into civil war, or why South Korea grew so quickly for so long, or the effects of democracy at the national level, or the impact of privatization on economic outcomes. Princeton economist Angus Deaton chafes at the idea that RCTs are some sort of “gold standard” set apart from other evaluations and the only route to truth. He notes that they usually explain only if an intervention has worked on average, not why or for whom it has worked. Using the results of such experiments to argue for “scaling up” immediately involves analysis that is not “rigorously scientific.”

Discussing evidence of the challenges of “scaling up,” Jim Manzi, a senior fellow at the conservative Manhattan Institute, points out that RCTs show merely that a particular intervention works within a particular test group (say, high school students in three districts of Kerala, or mothers of unvaccinated children in refugee camps in Ethiopia). Change the test group and different outcomes are likely. Manzi’s review of 122 RCTs in criminology suggests that randomized experiments frequently produce contradictory results across different populations for this very reason.

Poor Economics responds to some of these criticisms. In a modest tone, Banerjee and Duflo acknowledge that RCTs may not answer the big questions of development, but assert that getting answers to some of the “smaller” questions can make a real and sustained impact on the quality of life of many people. And a number of interventions they write about have been repeatedly tested in randomized experiments across a number of countries, suggesting the results don’t apply only to particular test communities. For example, studies have found that cash transfers to parents paid on condition that their children go to school or are vaccinated increase enrollments and vaccinations across numerous countries. Meanwhile, RCTs that have produced mixed results in only one location—such as the microfinance study that found a limited impact on income generation—at least put to rest universal claims for the utility of a particular intervention.

In any event, RCTs will remain an important tool in the arsenal of those fighting to make overseas aid agencies and governments in developing countries work better. Their utility is orders of magnitude higher than yet another statistical study in the ongoing and seemingly eternal debate over the evidence of a link between aid and economic growth. That’s in part because RCTs give the obvious and equivocal answer to the question over such a link: “It depends.” What is the aid being spent on? Where is it being spent? How is it being spent? The studies described in Poor Economics are a powerful assault on all-encompassing theories of global development: that “it’s all about institutions,” or all about investment, or all about education. It isn’t. Figuring out what works is not a theoretical exercise, but a practical one.

The good news is that a range of approaches is working. That is the only way to explain the broad-based progress we are witnessing worldwide. Global poverty won’t be eradicated tomorrow, and it won’t be solved by one big initiative. But it is suffering death by a thousand cuts—and Banerjee and Duflo’s work is helping us wield the knives to greater effect.

Read more about DevelopmentEconomicsPoverty

Charles Kenny is a senior fellow at the Center for Global Development, a Schwartz Fellow at the New America Foundation, and author of Getting Better: Why Global Development Is Succeeding and How We Can Improve the World Even More

Click to

View Comments

blog comments powered by Disqus