DIRECT INTERVENTIONS IN MODERN TIMES

At the opposite extreme to perfect targeting one can imagine a basic income scheme, which provides a fixed cash transfer to every person, whether poor or not.¹¹ This has been advocated by (among others) Paine (1797), Rhys-Williams (1943), Meade (1972), Raventos (2007), and Bardhan (2011). The idea has spanned both rich and poor countries, and the political spectrum from left to right. There are no substitution effects of the transfers because there is no action that anyone can take to change their transfer receipts, but there will be income effects (including higher demand for leisure, though how much so is unclear). There is no stigma associated with participation, given that there is no purposive targeting to poor people. A complete assessment of the implications for efficiency (and equity) must take account of the methods of financing the scheme. The administrative cost would probably be low though certainly not zero given that some form of personal registration system would probably be needed to avoid “double dipping” and to ensure that larger households receive proportionately more money.

The bulk of the direct interventions found in practice fall somewhere between the above extremes of “perfect targeting” and a basic income with no targeting. In countries where income means testing is a feasible option (mostly rich countries), the benefit level can be progressively phased out as income rises above some level, below which some guaranteed support is provided. The rate of benefit withdrawal depends on the strength of the expected labor supply response. With the better data and analytic tools available today, it can be hoped that future policy debates will be better informed about actual behavioral responses. However, from what we know already about labor supply responses, it is evident that poor people gain significantly from transfers in a country such as the United States (Saez, 2006).

The recent emphasis on targeting in many countries (both rich and poor) has typically been defined as avoiding “leakage” of benefits to the nonpoor, implicitly downplaying concerns about coverage of the poor (as pointed out by Cornia and Stewart, 1995).

120

22.9.2 State-ContingentTransfers Financed by Taxation

Recall that the essential idea of England’s Old Poor Laws was state-contingent transfers financed by taxation. There was little effort at explicit targeting of relief (before the 1834 reforms, which we return to later), although there was some degree of self-targeting given that relatively well-off families would be reticent to turn to the parish for assistance after some economic shock.

The idea of untargeted state-contingent transfers (as in the Old Poor Laws) reemerged in twentieth century Britain in the form of the Beveridge Report (Beveridge, 1942), which outlined detailed proposals for social insurance, whereby all those of working age would be obliged to pay a national insurance contribution to finance state-contingent transfers to the unemployed, the sick, the elderly, or widowed. However, unlike the Old Poor Laws, this was to be a national scheme rather than implemented locally. Two other ele­ments completed the social protection policy. First, family allowances were proposed, to cover the costs of dependent children (after the first). Second, an income top-up was pro­posed for those who fell below absolute standards taking account of all income sources.^[512] Although the aim of these proposals was squarely to eliminate poverty, Beveridge was opposed to means-testing—universal provision at a flat-rate was seen to avoid the costs of targeting and to encourage social cohesion.^[513] The past, deliberately stigmatizing, approach typified by the workhouses was to be abandoned. Beveridge’s plan formed the basis for the policies of the new Labour government elected in 1945; the Conservative resistance to the (popular) Beveridge plan helped ensure a Labour victory (Thane, 2000, p. 369).

America’s Social Security system had also grown out of prior relief efforts (notably those established during the depression) and came to provide fairly comprehensive state-contingent transfers, financed by taxation, soon after World War II. As with the Poor Laws, there was much debate about these policies. (America’s Social Security system was, and still is, decried as “socialism” in some quarters.) Similar to the 1834 reforms to the Poor Laws, calls for targeting have become common since 1980 in an attempt to reduce the fiscal cost of social insurance.

Uniform but state-contingent transfers are not common in developing countries today.

As we will see, although better targeting may help, finely targeted policies have costs that are often hidden but must be considered in any proper evaluation of the policy options.

22.9.3 Workfare

The workhouses that emerged in Europe around 1600 can be interpreted as a means of getting around the information and incentive problems of targeting. Design features encouraged those truly in need of help to turn to the workhouse and encouraged them to leave it when help was no longer needed, given that there were better options in the rest of the economy. This solves the information problem of targeting. However, it does so by imposing costs on participants, notably the forgone earnings and the welfare costs of stigma and subjugation (as Oliver Twist experienced). A truly utilitarian-welfarist assess­ment relative to untargeted transfers would clearly be ambiguous without further evi­dence. Arguably England’s workhouses of the nineteenth century went too far in imposing costs, which came to be widely seen as objectionable, on participants to ensure self-targeting. But the idea of self-targeting had lasting influence.

The workhouses are an example of a class of direct interventions often called today “workfare schemes”—schemes that impose work requirements on welfare recipients as a means of ensuring incentive compatibility. Though not involving workhouses, this idea was embodied in the Famine Codes introduced in British India around 1880, and the idea has continued to play an important role to this day on the subcontinent (Dreze, 1990a). Such schemes have helped in responding to, and preventing, famines, including those in Sub-Saharan Africa (Dreze, 1990b). Workfare was also a key element of the New Deal introduced by US President Roosevelt in 1933 in response to the Great Depression.

An important subclass of workfare schemes has aimed to guarantee employment to anyone who wants it at a predetermined (typically low) wage rate. Employment Guarantee Schemes (EGSs) have been popular in South Asia, notably (though not only) in India where the Maharashtra EGS, which started in 1973, was long considered a model. In 2005, the central government implemented a national version, the Mahatma Gandhi National Rural EGS. This promises 100 days of work per year per rural house­hold to those willing to do unskilled manual labor at the statutory minimum wage listed for the program. The work requirement is (more or less explicitly) seen as a means of ensuring that the program reaches India’s rural poor.^[514]

These schemes can be interpreted as attempts to enforce a minimum wage rate in sit­uations in which there is no other means of legal enforcement. Minimum wages appeared in the late nineteenth century, with the first minimum wage law introduced by New Zealand in 1894. Critics have long pointed to concerns about negative effects on overall employment of minimum wages rates, although advocates have pointed out that those effects may be small in practice and even positive in monopsonistic labor markets. However, enforcement of minimum wage legislation has been famously weak in developing countries with large informal sectors (including traditional farming). For example, Murgai and Ravallion (2005) show that in 2004-2005, three-quarters of India’s casual labor was paid less than the country’s (state-level) statutory minimum wage rates. In an EGS, anyone who wants work can (in theory) get it provided they are willing to do unskilled manual labor at the statutory minimum wage rate in agriculture.

An important difference between an EGS and minimum wage legislation is that an EGS aims to provide comprehensive insurance for the able-bodied poor, in that anyone who needs work can get it, at least on paper. Eligibility is open to all, so that a farmer who would not need the scheme in normal times can turn to it during a drought (say). This concept was explicit from the outset of the idea of an EGS (as it developed in Maharashtra in the early 1970s). Whether this insurance function is served in practice is another matter; Dutta et al. (2012) find evidence of considerable rationing on India’s national EGS. The rationing tends to be greater in poorer states, which may well reflect their weaker administrative capabilities to implement a complex program such as India’s national EGS.

125

These schemes illustrate that even a well-targeted transfer scheme can be dominated by untargeted transfers when one takes account of all the costs involved, such as income forgone or other costs in complying with the conditionalities imposed. Ravallion and Datt (1995) and Murgai et al. (2013) provide evidence that in both the Maharashtra EGS and India’s new national scheme, an untargeted basic income scheme would have been more cost-effective in directly transferring incomes to poor people.

Workfare schemes have typically been seen as short-term palliatives—a form of social insurance. In principle, a workfare scheme can also directly serve promotional goals. One way is by generating assets that could change the wealth distribution or shift the produc­tion function, which could also allow people to break out of the poverty trap illustrated in Figure 22.1. In practice, asset creation has not been given much weight in these schemes in South Asia, although it seems to have greater weight elsewhere, including in Latin America (such as Argentina’s Trabajar Program).

Another way that workfare programs can better serve the promotional goal is by tying benefits to efforts to enhance human capital through training. Welfare reforms in many rich countries since the early 1990s have aimed to make transfers conditional on invest­ments in human capital and to incentivize searching for and finding private employ­ment.^[515] This form of workfare does not actually provide employment, as in the public works form of workfare. Training and encouragements for private-sector employ­ment using wage subsidies have also been used to encourage the transition from public employment in workfare schemes to private employment.^{[516] [517]}

Next we turn to a policy for which the creation of human wealth is seen as crucial to poverty reduction.

22.9.4 Schooling for Children from Poor Families

Children from poor families tend to get less schooling. This “economic gradient” in schooling persists to this day almost everywhere and has long been seen as a factor that perpetuates poverty across generations—a potential source of a poverty trap. As noted in Section 22.5, the inability of poor families to finance their children’s schooling given credit market failures came to be recognized as a key factor in perpetuating poverty and entailing that a more unequal initial wealth distribution will generate aggregate effi­ciency costs.^{1 8} Thus, policies that can promote the schooling of children from poor fam­ilies can be seen as an important part of social policy that could improve both equity and efficiency, and credibly allow people to escape poverty permanently.

Such policies are a modern idea, advocated at times but little known in practice before the nineteenth century (see Section 22.5). Past and ongoing policy debates over mass education have raised many issues, but a fundamental one is whether compulsory school­ing is even in the interest of poor families, for it was typically their children who were unschooled. Opponents (on both the left and right) of compulsory schooling pointed to the costs (primarily their forgone earnings) to poor families of sending their children to school. While compulsory schooling could break the poverty trap, a short-term trade-off was created by the costs to poor families. Advocates argued, in effect, that the longer-term benefits from breaking out of a poverty trap outweighed these costs.

After much debate, compulsory schooling emerged in virtually all industrialized countries by the early twentieth century, with a significant state role in both public provision and support for private schooling. In England, the Elementary Education Act of 1870 was a breakthrough law that established a secular public sector institutional framework, including democratic school boards. Implementation was uneven geo­graphically, and there was a continuing struggle for control of schools between the democratically elected local bodies and religious organizations (Stephens, 1998). It was not until the 1880 act of the same name that education was compulsory in England for children aged 5-10. A similar act was passed in France about the same time. In the United States, 34 states had compulsory schooling laws by 1900, 30 of which required attendance until at least age 14. Japan in the Meiji period (1868-1912) was not behind the West in promoting mass education, which was vir­tually universal by the end of the period. Mass public education (with tertiary educa­tion left largely to the private sector) was given high priority throughout developing East Asia, with educational attainments far surpassing those of most developing coun­tries and even some developed countries.

The payoffs from mass public education were huge. Equitable, broad-based education has been identified by Goldin and Katz (2008) as a key factor in the US record of rela­tively equitable and rapid economic growth in the period 1940-1980. The ability of the school system to support a relatively rapid increase in education attainments in the United States in this period (though slowing down greatly after 1980) meant that the supply of skilled workers kept up with the extra demand stemming from new technologies—what Tinbergen (1975) dubbed the “race between education and technology”—thus attenu­ating the inequality-increasing effects of technical progress favoring demand for relatively skilled labor. The fact that American educational expansion was so broad-based in this period was key. A more elitist school system would have entailed a more unequal dis­tribution of the gains from growth. And Goldin and Katz argue that rising inequality in the United States since 1980 stems in large part from the fact that the education system has not allowed the supply of the types of skilled labor required for the new technologies of the time to keep up with the demand. And it tends to be children from poor families who are most disadvantaged in this race.

Broad-based education has also been identified as a key factor in East Asia’s relatively equitable growth. Using a regression of GDP per capita growth rates from 1960 to 1985 on primary and secondary education attainments in 1960—with controls for initial GDP per capita, population growth, and the share of investment in GDP—an influential report by the World Bank (1993) identified primary education as the most important single fac­tor, accounting for somewhere between 58% (Japan) and 87% (Thailand) of GDP growth. Of course, such calculations can be sensitive to model specification; the educa­tion variables could well be correlated with other omitted factors. However, it is none­theless striking that primary education is found to account for a greater share of the variance in growth rates than private (nonhuman) investment.

There is also evidence that education attainments have interacted strongly with India’s growth process in determining the impact of that growth on poverty. This was demonstrated by Ravallion and Datt (2002) by comparing rates of poverty reduction among India’s states. While the elasticities of measured poverty to farm yields did not vary significantly across states, those for nonfarm output did. The nonfarm growth pro­cess tended to reduce poverty more significantly in states with initially higher literacy rates, and interstate differences in literacy rates were the dominant factor among those identified by Ravallion and Datt. The importance of mass education has long been acknowledged in principle in India. A “directive principle” of state policy in the 1949 Constitution was free compulsory education to the age of 14.^[518] However, implemen­tation of this policy has lagged considerably, with large interstate differences and often poor quality schooling across the country (Probe Team, 1999). The state that has made the most progress in mass public education is Kerala. Expanding literacy to the whole population was a high priority of the state government from the 1950s (building on a history of prior successes in schooling provided by Christian missionaries dating back to the early nineteenth century). The results of Ravallion and Datt (2002) indicate that Kerala’s success in mass schooling has generated a far more propoor process of nonfarm economic growth than is found in other states.

Bans on child labor have often been proposed and legislated. Hazan and Berdugo (2002) model an interesting version of a poverty trap in which, at the early stage of devel­opment, child labor is abundant while fertility is high, and mean output is low. With economic growth stemming from technical progress, the returns to schooling rise, mak­ing child labor less attractive and also lowering fertility. In this model, the economy even­tually converges to a new equilibrium in which child labor has vanished. Hazan and Berdugo show that an effective ban on child labor will speed up the transition to this new equilibrium.

However, in economies with large informal sectors, the enforcement of such bans is difficult. Legislation to set a minimum working age was introduced in some countries from the late nineteenth century, although it is unclear how much this helped reduce the incidence of child labor; Moehling’s (1999) analysis suggests this legislation had little effect. Basu (1999) argues that compulsory schooling is a better way of implementing a ban on child labor than an actual ban, and compulsory schooling can also break the poverty trap.

129

22.9.5 Policy Incentives for Schooling

Although out-of-pocket expenses and the forgone earnings of children figured in the nineteenth-century debates about the idea of compulsory education, there was not much discussion of the obvious policy response: a bursary, or scholarship grant, for poor fam­ilies. Smith (1776) andMill (1859, Chapter5) had advocated tuition subsidies for children from poor families. Marshall (1890, p. 594) took a less sympathetic attitude and proposed instead penalizing poor parents (a public policy of “paternal discipline”) who neglected to send their children to school or to care for their health. Educational institutions have for a long time subsidized tuition fees and other costs for selected students, often based on some sort of means test. England’s 1870 Elementary Education Act recommended tuition subsidies for children from poor families (Gillie, 1996). However, implementation of public policies providing any form of schooling incentive for poor parents had to wait until the middle of the twentieth century, after which it started to become common prac­tice to build in incentives for children from poor families to stay in school. Britain’s 1942 Beveridge Report recommended a universal child allowance paid up to the age of 16 if the child stayed in school.^[519] Australia had a school bursary program from the 1960s that essentially paid parents from poor families to keep their children in school beyond the age the children would normally leave school as long as the children passed a special exam. It is common today for various forms of education subsidies (scholarships, tuition subsidies, subsidized loans) to be means-tested.

In the development literature in the 1990s, targeted bursaries came to be known as CCTs.^[520] The idea was the same: a monetary incentive for poor parents to keep their children in school. Transfers are made under the condition that the children ofthe recip­ient family demonstrate adequate school attendance (and health care in some versions of the policy). Plainly, the promotion benefits of these programs rest on ensuring that the transfers go to poor families, presuming that the children of the nonpoor will already be in school. Thus, targeting has been instrumentally important to both the protection and promotion benefits. The promotion benefits also depend on designing the conditions so that the required level of schooling would not be attained in the absence of the pro­gram. Early influential examples of these programs in developing countries were Mexico’s PROGRESA program (now called Oportunidades) and Bolsa Escola in Brazil. Another early example was the Food-for-Education Program in Bangladesh for which the transfers (targeted to poor families) were made in kind but were also conditional on school attendance. Bolivia’s CCT, Bono Juancito Pinto, introduced in 2006, is an example of a universal (untargeted) transfer program, for which every child enrolled in public school is eligible, irrespective of family income. More than 30 developing coun­tries now have CCT programs, and the number is growing (World Bank, 2014). And other countries have similar policies that are not called CCTs; for example, in an attempt to ensure that poverty did not constrain schooling, since 2002 China has had a “two exemptions, one subsidy” policy for students from poor rural families; the exemptions are for tuition fees and textbooks, and the subsidy is for living costs.

These programs are clearly designed with a view to breaking the poverty trap stem­ming from the aforementioned economic gradient in human development. If the sole concern was with current income gains to participating households, then a policy maker would not impose schooling requirements, which entail a cost to poor families by incen­tivizing them to withdraw children or teenagers from the labor force, thus reducing the (net) income gain to poor people. The idea of these programs is to strike a balance between protection and promotion, based on the presumption that poor families cannot strike the socially optimal balance on their own. The program’s incentive effect on labor supply (previously seen as an adverse outcome of transfers) is now judged to be a benefit—to the extent that a well-targeted transfer allows poor families to keep the kids in school, rather than sending them to work. Concerns about distribution within house­holds underlie the motivation for such programs; the program’s conditions entail that relatively more of the gains accrue to children. Some advocates of CCTs have also claimed that they would reduce child labor, although the economic data are unclear about whether such a policy will work for this purpose; Ravallion and Wodon (2000a) show that, under standard assumptions, a tuition subsidy will increase children’s amount of schooling but has theoretically ambiguous effects on the supply of child labor; empirically, the authors find that a tuition subsidy has little effect on child labor in Bangladesh.

There is evidence from impact evaluations that these schemes bring nonnegligible benefits to poor households, in terms of both current incomes and future incomes, through higher investments in child schooling and health care.^[521] The conditions change behavior. In the United Kingdom, means-tested grants paid to secondary students have been found to very effectively reduce the number of school drop outs from poor families (Dearden et al., 2009). The various evaluations of Mexico’s PROGRESA/Oportuni- dades program have been positive; see the survey in Fiszbein and Schady (2010). Baird et al. (2011) found sizeable gains from the schooling conditions in a Malawi CCT. In a study for Burkina Faso, Akresh et al. (2013) found that the conditionality mattered more in encouraging the school enrollment of children who were initially less likely to go to school, including girls—children who are less likely to receive investments from their parents. Cameron (2002) found that a CCT program in Indonesia, Jaring Pengamanan Sosial, had the greatest impact at the lower secondary school level where children are most susceptible to dropping out. The design features have also been critically assessed. A series of papers on PROGRESA revealed that a budget-neutral switch of the enroll­ment subsidy from primary to secondary school would have delivered a net gain in school attainments by increasing the proportion of children who continue onto secondary school.^[522] Although PROGRESA had an impact on schooling, it could have had a larger impact. However, it should be recalled that this type of program has two objectives: pro­motion by increasing schooling (reducing future poverty) and protection by reducing current poverty through the targeted transfers. To the extent that refocusing the subsidies on secondary schooling would reduce the impact on current income poverty (by increas­ing the forgone income from children’s employment), the case for this change in the pro­gram’s design would need further analysis.

132

Impact evaluations have also pointed to high returns for early childhood interventions in some settings. The experimental Perry Preschool Program in the United States in the 1960s provided schooling and home visits for poor children aged 3—4. The benefits included higher adult earnings and reduced crime, and the benefit-cost ratio (even with­out putting greater weight on the propoor distribution of the gains) was estimated to be more than eight to one (Heckman, 2006). Head Start (also begun in the 1960s as part of the United States’s War on Poverty) was a similar national preschool program, which targeted a package of education, health, and nutrition services to poor families; the pro­gram continues at the time of this writing and, as of 2005, some 22 million preschool children had participated in Head Start programs. Head Start has also been found to gen­erate sizeable long-term gains in schooling, earnings, and reduced crime (Garces et al., 2002). The aggregate benefits from Head Start also appear likely to exceed the cost, even without distributional weights (Ludwig and Phillips, 2007). There is also evidence of sig­nificant long-term gains in adult health indicators from an intensive preschool program launched in the United States during the 1970s, the Carolina Abecedarian Project (Campbell et al., 2014). There is a great deal of interest in how effective early childhood interventions might be devised for developing countries.

All these interventions require complementary efforts on the supply side, through effective (public or private) service delivery. This has been an important concern in many developing countries; World Bank (2004) reviews the evidence and discusses how better incentives for service delivery might be developed.

133

22.9.6 Microfinance Schemes

As we have seen, credit market failures have been identified as a cause of poverty and a reason why poverty can be costly to overall economic performance. In addition to long­standing moral arguments, transfers to poor people can be interpreted as a means of relieving the constraints stemming from such market failures. But there is another option, namely, policies that aim to make financial institutions for saving and borrowing work better for poor people. Microfinance programs aspiring to support small-scale credit and savings transactions by poor people have attracted a great deal of interest since the idea emerged in the late 1970s, and there are now many examples of such programs in the developing world.

The classic argument is about promotion, namely, that relaxing borrowing constraints faced by poor people allows them to invest and so eventually escape poverty by their own means. Credit and savings are also potentially important instruments for protection, by allowing poor households to more effectively smooth their consumption in the face of income fluctuations.

Much of the early (and ongoing) enthusiasm for microfinance was really little more than advocacy, with weak conceptual and empirical foundations. In recent times, there has been a rise in popular concern in the media (in South Asia especially) about over­borrowing by poor people once they are given access to microfinance. Much of this con­cern appears to stem from anecdotes, and the debate has also become politicized. Positive average impacts do not, of course, mean that there are no losers among the recipients. This is probably true of all antipoverty policies, but it is especially so in the case of credit-based interventions. Risk is not eliminated, shocks do occur, and mistakes are made, such as due to faulty expectations. There will be both winners and losers in these types of interventions.

The earliest and still most famous example of this class of policies is Bangladesh’s group-based lending scheme, Grameen Bank (GB). GB has made a conscious effort to reach the poor both through their eligibility criteria and their branch location decisions, which (in contrast to traditional banks) have favored areas where there are unexploited opportunities for poor people to switch to nonfarm activities (Ravallion and Wodon, 2000b). Research on GB has indicated that the scheme has helped in both protection and promotion; in the former case, by facilitating consumption smoothing and, in the latter, by helping to build the physical and human assets of poor people.^[523] This result was found by Pitt and Khandker (1998), who exploited the design features of GB, notably that it is targeted to the landless, to identify effects. Given that access to GB raises the returns of being landless, the returns of having land will be higher in villages that do not have access to GB credit. Thus, comparing the returns of having land between

134

villages that are eligible for GB and those that are not (with controls for other observable differences) reveal the impact of access to GB credit. Put another way, Pitt and Khandker measure the effect by the mean gain among households that are landless from living in a village that is eligible for GB, less the corresponding gain among those that have land. They found positive impacts on measures relevant to both protection and promotion. This was confirmed in a subsequent study by Khandker and Samad (2014) using survey data on 3000 households spanning 20 years. The success of GB has led to a proliferation of microfinance schemes in Bangladesh, with over 500 providers at the time of this writing.

Even careful observational studies such as that by Pitt and Khandker require identi­fying assumptions that can be questioned, and there has been some debate about the robustness of their results.¹³⁵ This is a type of policy intervention for which it will inev­itably be hard to convince everyone of the validity of the identifying assumptions given the likelihood of unobservable factors jointly influencing acceptance and effects. Exper­imental evaluations relying on randomized assignment (typically at the community level) have offered the hope of more robust results, and there have been some interesting exam­ples. A study by Banerjee et al. (2009) of the impacts of opening new microfinance branches in slums of Hyderabad in India found that overall borrowing, business start­ups, and spending on consumer durables (but not nondurables) increased in the areas that were randomly assigned the new branches relative to the control areas. However, the study did not find evidence of positive impacts on health, education, or women’s self­efficacy. Heterogeneity was the focus of a recent experimental evaluation of access to micro-credit by working-age women in Mexico (under the Compartamos Banco scheme) by Angelucci et al. (2013). The authors found positive average effects in a num­ber of dimensions. There was heterogeneity in the impacts, but they found little evidence of significant losses, including among poor borrowers. More research on the benefits and costs of microfinance schemes can be expected.

It is clear that we have seen a shift in thinking about this class of policies over the last 200 years; in the days when poor people were routinely blamed for their poverty, giving them a loan would not have made sense. Of course, identifying credit market failures as one cause of poverty does not imply that credit for the poor will solve the problem. But well-designed programs do have a role, as a complement to other policies for protection and promotion.

22.9.7 Poor-Area Development Programs

Almost all countries have their well-recognized “poor areas,” in which the incidence of absolute poverty is unusually high by national standards. We would hope, and under cer­tain conditions expect, that the growth process will help these poor areas catch up.

See Morduch (1999) and Roodman and Morduch (2009) as well as the latest detailed rejoinder in Pitt and Khandker (2012).

But this process often appears to be slow, and geographic divergence has sometimes been evident. This has led to antipoverty policies focused on lagging poor areas. “Poverty maps” are widely used in geographic targeting, and the method proposed by Elbers et al. (2003) has facilitated many applications.

Laggingpoorareas have prompted poor-area development projects—one ofthe oldest forms of development assistance, though under various headings (including “Integrated Rural Development Projects” and “Community Driven Development”). Extra resources are channeled to the targeted poor areas for infrastructure and services and developing (farm and nonfarm) enterprises. Emphasis is often given to local citizen participation in decision making, although a survey of the available evaluative research by Mansuri and Rao (2012) found somewhat mixed success given the scope for exploitation by local elites.

It is widely agreed that poor areas are typically characterized by low capital-to-labor ratios, but there is less agreement about the right policy responses, such as efforts to aug­ment local capital—investing in lagging poor areas—versus policies to encourage out migration. Geographic externalities clearly play an important role, but this role is still poorly understood because of a lack of convincing empirical research.

In the case of China, where poor area development has been the main form of direct intervention against poverty since the mid-1980s, there is evidence of pervasive geo­graphic externalities, whereby households living in poor areas have lower growth pros­pects than seemingly identical households living in well-off areas ( Jalan and Ravallion, 2002; Ravallion, 2005). This suggests that there is scope for poor-area development as a means of ensuring longer-term promotion from poverty, as well as protection. However, here, too, the evidence for the success of the policies currently in practice is mixed.¹³⁶

The main concerns about the incentive effects of poor-area programs have related to the responses of local governments to external aid and to migration. An example of the former is found in Chen et al. (2009), who demonstrate that local government spending allocations changed in response to efforts by higher levels of government to target poor villages in rural China, dampening the targeting outcomes. Regarding migration, there appears to be a widely accepted assumption that there is limited intrarural mobility in developing countries, sometimes reflecting institutional and policy impediments (such as local administrative powers for land reallocation, as in China). It is not clear how con­fident we can be in making that assumption.

There is still much we do not know about the impacts of poor-area development efforts, especially over the long term, and the tradeoffs faced by policy options. Although local infrastructure development is clearly crucial to fighting poverty, it has not attracted the degree of attention in evaluative research that has been generated by social policies. Here, an important factor is the extent to which “development impact” is challenged by

For example, contrast the findings of Jalan and Ravallion (1998) with Chen et al. (2009) on poor-area programs in China.

donors and citizens. “Impact” is too often taken for granted with infrastructure develop­ment. By contrast, the “softer” social policies have had to work hard to justify themselves, and evaluative research has served an important role. If the presumption of impact is rou­tinely challenged by donors, aid organizations, and citizens, then we will see stronger incentives for learning about impact and fewer knowledge gaps.

22.9.8 Information Campaigns

There has been recent interest in the scope for using information-based interventions because lack of information is a decisive factor inhibiting poor people from successfully participating in actions to get the services to which they are entitled. There are some signs of support for this premise from past research. Stromberg (2004) reports evidence that US antipoverty programs have worked better in places with greater access to radios. Besley and Burgess (2003) found that the governments of Indian states where newspaper circu­lation is greater are more responsive in their efforts to mitigate negative agricultural shocks. Reinikka and Svensson (2005) found significant effects of information through a newspaper campaign on school outcomes in Uganda.

There have been some evaluations of information interventions. The results so far seem mixed. Focusing on one country and one sector, Pandey et al. (2009) report that a community-based information campaign led to short-term gains in schooling out­comes, while the findings of Banerjee et al. (2010) are less encouraging about the scope for using information interventions to improve the monitoring of education service pro­viders in India. In rich countries facing concerns about the rising incidence of obesity, there have been efforts to post information on the “calorie prices” of food.¹³⁷ A recent review of both experimental and nonexperimental evaluations found mixed evidence of effectiveness (Swartz et al., 2011).

Mixed results of this sort might not be surprising. Three observations can be made. First, public information about a program may well discourage participation; for some people, learning about the program may have the opposite effect; see, for example, Hertel- Fernandez and Wenger (2013), with regard to an information campaign for a US program. Second, incomplete information is only one of the possible reasons why poor people do not access services (Keefer and Khemani, 2005; Cappelen et al., 2010). Third, mixed results might also stem from heterogeneity in the quality of the information intervention itself.Also, for India, Ravallion et al. (2013) report success in changing public awareness of rights and rules under India’s EGS using an entertaining and high-quality fictional movie that can be shown in villages. However, the results also warn that informing poor people of their rights is not sufficient for positive change. Public awareness can be improved, but this must be combined with effective responses on the supply side.

For example, US legislation in 2010 requires restaurant chains with 20 or more outlets to post calorie counts for all food items sold.

22.10.