B000QJLQXU EBOK

by William Easterly

  More anecdotally, people in Buenavista have noticed the difference. Emma García says that she can feed her children meat twice a week now to supple ment the tortillas, thanks to the money she receives from PROGRESA. Schoolteacher Santiago Días notices that attendance is up in Buenavista’s two-room schoolhouse. Moreover, Días says, “because they are better fed, the children can concentrate for longer periods. And knowing that their mothers’ benefits depend on their being at school, the children seem more eager to learn.3

  Because the program was such a clearly documented success, it was continued despite the voters’ rejection of the longtime ruling party in Mexico’s democratic revolution in 2000. By that time, PROGRESA was reaching 10 percent of the families in Mexico and had a budget of eight hundred million dollars. The new government expanded it to cover the urban poor. Similar programs began in neighboring countries with support from the World Bank.4

  The lesson for aid reformers is that a combination of free choice and scientific evaluation can build support for an aid program where things that work can be expanded rapidly. The cash-for-education-and-nutrition segment in itself could be expanded, with suitable local adjustments, to more countries and on a much larger scale than it is now.

  Helping Children Learn in Kenya and India

  Education promoters have offered a bewildering variety of interventions to educate children: abolishing school fees, providing uniforms, building classrooms, hiring more teachers, training teachers, providing free meals at school, making children healthier with free medicine, providing textbooks and flipcharts, offering remedial education, awarding prizes to teachers whose students perform well, and handing out vouchers for private schools. In the typical utopian program, everything is done at once and it is impossible to learn what works and what doesn’t.

  Fortunately, some Searchers for solutions have taken a different approach. They study each intervention in isolation, identifying its effects relative to a control group that did not get the intervention. Esther Duflo of MIT and Michael Kremer of Harvard are pioneers in this approach to development. One researcher studying a project in Western Kenya found that attendance at preschool programs was 25 percent higher when a free breakfast was provided, and test scores were higher even though the meals cut into instruction time.

  Did textbooks raise students’ performance? More research in the same district in Kenya found that those students who were in the upper 40 percent of the class did have an improvement in test scores after getting textbooks. How to reach the other 60 percent of students? One bright idea was to use flip charts, which were thought to be easier to follow for less literate students. Unfortunately, the same careful research found that flip charts had no effect on test scores. The research giveth, and the research taketh away.

  In India, another program to help poorly performing students hired remedial education assistants from the communities to tutor the students. The students in Mumbai and Vadodara got two hours a day of tutoring from the assistants. What were the results? This program did increase test scores, with the most dramatic results among those students who had been doing worst prior to the program. The researchers estimated that the program of hiring remedial education workers from the community achieved ten times the results of hiring more conventional teachers for the same cost.

  Another promising idea was to improve teacher incentives. An experiment in Western Kenya had parent groups give prizes to teachers whose students performed the best on standardized tests. Test scores of students covered by the program initially rose, but subsequent monitoring of these students found that they quickly regressed back to the same level as those students not covered by the program. Teachers were apparently “teaching to the test,” with no lasting impact on student performance.5

  The lesson of all this research is that some equally plausible interventions work and others don’t. Aid agencies must be constantly experimenting and searching for interventions that work, verifying what works with scientific evaluation. For learning to take place, there must be information. The aid agencies must carefully track the impact of their projects on poor people using the best scientific tools available, and using outside evaluators to avoid the self-interest of project managers.

  What Works?

  MIT professor Abhijit Banerjee gives other helpful examples besides those already mentioned that have been verified as cost-effective uses of foreign aid: deworming drugs; dietary supplements such as those for iron, vitamin A, and iodine; education in using condoms and treating other sexually transmitted diseases to slow the spread of AIDS; indoor spraying to control malaria; fertilizer subsidies; vaccination; and urban water provision.6 None of these are keys to development according to some utopian scheme; they are just modest interventions that make people’s lives better.

  Other evidence on what works has come from cross-country comparisons of country practices and certain outcomes. And not all piecemeal interventions are for social services and projects. James Barth of Auburn, Gerard Caprio of Williams College, and Ross Levine of Brown University found a strong association between banking regulations that force banks to disclose accurate, timely, comparable information on their finances and the level of banking development in a country. They controlled for possible reverse causality and found that the result still held. Contrary to the widespread view that tough official bank supervision is needed to protect savers from rogue banks, they found a negative relationship between powers of bank supervisors and development of a healthy banking industry (also controlling for possible reverse causality). If you want to promote healthy banking, which is an ingredient in enabling the poor to help themselves through credit, then these statistical associations point you in the direction of regulating information disclosure rather than having powerful official bank supervisors.7

  Experiments and statistical analysis are a long way short of a panacea for effective aid. There is also development problem solving that is more like learning to sail than testing a theory of fluid dynamics. Aid agencies must let their staff gain experience in a particular local setting on a particular problem, and then let the experienced staff decide on the ground what is working and what is not. Instead, we see aid bureaucracies shifting staff around before they can gain enough experience, with the result that those agencies are full of generalists without local or specialized knowledge.

  How can aid agencies acquire the incentive to do scientific evaluation, statistical analysis, and on-the-ground learning? For many efforts, you can survey the intended beneficiaries—the poor—before and after every project, along with a control group that did not benefit from the project. Disclose publicly all efforts and outcomes. Let a community of observers who care—from both donor and recipient countries—hold the agencies responsible for outcomes. Increase the pressure of free speech and democracy on aid agencies by having vociferous advocates for the poor who practice science rather than public relations. Have rich-country politicians realize that a negative evaluation of a particular aid effort is a learning opportunity, not an excuse to cut foreign aid.

  Your Ideas Are Crazy, but Are They Crazy Enough?*

  On New Year’s Eve, December 31, 2004, in Washington, D.C., Dennis Whittle and Mari Kuraishi got married at the stroke of midnight. The wedding was a charming blend: Shaker hymns, a Unitarian minister, Mari’s very proper Japanese mother, Dennis’s good-old-boy Kentucky father, and the joint giving away of the bride by an international assortment of their friends and relatives.

  Dennis and Mari are intellectual as well as romantic partners. They come into this story because of the innovative ideas to which they are devoting their non-wedding hours. They decry the central planning approach of the international aid system and have come up with some practical steps to make it act a little bit more like a free market, making the system match the free-market rhetoric of some of its participants.8

  They suggest a marketplace instead of central planning, a kind of eBay meets foreign aid. They see three types of actors: (1) social
entrepreneurs close to the poor who propose projects to meet their needs; (2) individuals and institutions with technical and practical knowledge, and (3) donors who have funds they want to give away. The current system has huge bureaucracies under (3) doing central planning of (1) and (2). Dennis and Mari envision instead a decentralized market in which each category has many players who seek out players in the other categories and spontaneously form matches (another possible metaphor: aid’s version of an online dating service). Projects would compete for funds, technical specialists would compete to be hired, and donors would compete to get results, and thus attract even more funds. All of the players would build reputations based on their past performance delivering results. Players would interact through personal meetings, telecommunications, and the Internet (much like the business networks described in chapter 3). The trust necessary for successful relationships would be built through repeated interactions. Dennis and Mari have put together an Internet platform and a company that would facilitate these matches, called GlobalGiving.com. They summarize their vision:

  The number of potential participants on the project design, funding, and implementation sides have all dramatically risen over the past decades. Official agencies, foundations, and even private companies that want to combat poverty abroad, assist with HIV/AIDS projects, or invest in schools in developing countries all have no place to “shop,” so to speak, for projects. On the other side of the equation, people with projects cannot find donors…. GlobalGiving.com…treats foreign aid and philanthropy like a marketplace, where donors and recipients come together to exchange information and resources…. [R]ecipients post projects that need funding, and individual and institutional donors can provide anywhere from a few dollars to thousands of dollars to the projects they find most worthy.

  This decentralized approach would avoid the coordination problems. It would bypass the narrow administrative funnel in the recipient government through which official aid must pass. It would avoid the strategic manipulation of aid by donor governments and the corruption of recipient governments.

  How does this work in practice? Take the schoolteachers in Coimbatore, India, who noticed that their girl students often left school after puberty and thought of a possible solution. As BusinessWeek tells the story:

  The teachers posted a small, bedraggled project on GlobalGiving—so tiny, in fact, that it initially embarrassed Whittle…. The project ad read: “New Toilet Block for School. $5,000.” Within a few weeks, four donors from around the U.S., including a writer from New York City and a banker from JP Morgan, put up the money. In less than three months, the school had its own separate toilet block for girls; the donors had thank-you letters and photos from the kids. Turns out the teachers had guessed right: The girls were dropping out in droves because of the embarrassment they felt once they started menstruating and had no private facilities. Now, two years later, 100 of them have stayed in school because of this tiny addition..9

  Remember the networks that substituted for missing institutions to make markets possible, in chapter 3? Think of the power of networks in foreign aid. If local social entrepreneurs, development workers, and rich-country donors were tied together in a network based on repeated interactions, the networks of relationships could find the local social entrepreneurs—like Patrick Awuah, the founder of Ashesi University in Ghana—who had good projects and good reputations. The local social entrepreneurs would be completely in charge, motivated only by the rewards of good reputation if they did deliver results to the poor and the threat of a lost reputation in the network if they didn’t.

  Think of the potential for creativity if thousands of potential donors, project proposers, technical advisers, and advocates for the poor were freed from the shackles of the large centralized bureaucracy and could find solutions that worked on the ground. This is not a panacea for redesigning all of foreign aid; it is just one promising experiment in how aid could reach the poor.

  Development Vouchers

  Let’s keep getting crazy about market mechanisms. Suppose we issue development vouchers to target groups of the extreme poor, which the poor could redeem at any NGO or aid agency for any development good they wanted—for example, vaccinations, life-saving drugs, a health worker’s visit, an improved cookstove, textbooks, seeds, fertilizer, or food supplements. The official aid agencies would set aside some of their money for an independent “voucher fund” separate from the agencies. The poor would choose both the goods they wanted and the agency they wanted to deliver the goods and would give their vouchers to that agency. The agency could then turn in the vouchers to the voucher fund for real money to cover the costs of providing the development services.

  Since the poor would be choosing which agency would deliver the goods, the agencies would feel competitive pressure to deliver results. They would feel cost pressures the way private firms do, trying to deliver more for less. (One study estimates that the least-efficient donor delivers one square meter of primary school space in Guinea for $878, while the most efficient can do the job for $130.10 The aid agencies would be forced to act as social entrepreneurs, trying to offer innovative services that would prove attractive to the poor. Any agency that was not delivering what the poor wanted would see its budget go down, as it turned over money to the voucher fund but none of the vouchers came back. The poor would be firmly in charge, giving feedback on what they did and did not want from the aid agencies.

  Individual vouchers would not work with development services that had to be chosen collectively to benefit a group, such as roads, health clinics, boreholes, or schools. We could think of vouchers being given to a village rather than to an individual to cover such goods. The villagers could then vote on how to spend the vouchers, so the feedback would be at the village level rather than the individual level. Again, the aid agencies would find out how well they were doing at satisfying the poor based on how many village vouchers they attracted. The vouchers would at long last provide a “market test” and a “voter test” to the aid agencies.

  Giving vouchers to the poor may be the stupidest idea ever, except for all the ideas that have already failed in foreign aid. Both the GlobalGiving and the voucher schemes should be treated as experiments that have to be tested to see whether they work. They should be tried on a small scale. Again, let me say that I do not have the next utopian scheme to fix aid, because there isn’t any. We would have to scrutinize these schemes for results just like we do conventional aid schemes. Do the players in GlobalGiving have the right incentives to give the poor what they need? Could it work on a larger scale than its present modest incarnation? Are voucher markets going to be efficient for small transactions?

  We have seen that markets and democracy (such as a village voting on vouchers) require many conditions to work well. One worry is that local elites could cheat the poor by buying up all their vouchers or votes and pursuing projects for the benefit of the elites. We really don’t know, but the instinct that something like a market could make aid work better sounds promising.

  One idea that is too quickly dismissed is for foreign aid to just give cash grants targeted to the poorest people. This would be the purest solution to letting the poor choose for themselves what they needed. Although there are many potential pitfalls, it’s surprising that aid agencies have not experimented with this approach in any serious way. This would be a promising area in which to do a randomized controlled trial.

  Rachel Glennerster of the MIT Poverty Action Lab and Michael Kremer of Harvard have another promising idea to give foreign aid in a way that promotes results. They suggest creating a fund that would make an advance-purchase commitment to whoever succeeded in developing a vaccine against malaria. This would remedy the incentives that led to the present pitifully inadequate R&D on malaria. We could think of expanding this idea to other areas of foreign aid. Have worldwide or regional competitions for effective aid projects and programs, with large rewards to be given to the agencies, NGOs, or individuals that an indepe
ndent panel judged as having designed the most effective aid programs, as measured by results.

  Feedback from the Poor

  If the main problem with foreign aid is the lack of feedback from the poor themselves, and accountability to these same poor, then why not attack the problem directly? The World Bank produced a fascinating three-volume publication called The Voices of the Poor, from which I have drawn some of the country examples in this book. Why not give the poor voices on whether aid is reaching them?

  Americans fought a revolution on the principle of “no taxation without representation.” Could Americans and other parts of the West extend that principle to the Rest: “no intervention without representation”? (And I hope I have made clear that nonrepresentative tyrants from the Rest cannot provide such “representation.”) Westerners: don’t do things to or for other people without giving them a way to let you know—and hold you accountable for—what you have actually done to or for them.

  Is aid reaching the poor? Well, let the agents of foreign assistance ask them. Evaluation efforts could include surveys of the poor. Just ask them if they got what they most needed and if they are better off because of an aid intervention, and hold the aid agencies accountable for the results. Hold surveys of the population’s well-being both before and after the aid program, to compare the results on specific outcomes.

  The main mechanism for feedback and accountability for public services in the West is democracy. Could aid agencies find democratic mechanisms for local communities to vote on what services and projects they wanted? Could independent local watchers make sure the goods actually arrived and delivered what the agencies promised? Myriad volunteers such as local college students could simply monitor a sample of potholes, missing textbooks for schoolchildren, or out-of-stock drugs in health clinics. They could make the call to the responsible party to repair the pothole, supply the textbooks, or restock the drugs. Publicize the results and thus put pressure on the aid donors and their local partners. It is strange that aid agencies talk so much these days about “good governance” in the recipient countries without worrying about “good governance” of their own aid projects.