World Development Vol. 40, No. 1, pp. 11–20, 2012 Ó 2011 Elsevier Ltd. All rights reserved 0305-750X/$ - see front matter www.elsevier.com/locate/worlddev

doi:10.1016/j.worlddev.2011.05.012

Social Transfers and Growth: What Do We Know? What Do We Need to Find Out? ARMANDO BARRIENTOS * University of Manchester, UK Summary. — The paper examines the extent to which emerging social transfer programs in developing countries can improve the productive capacity of households in poverty and contribute to micro-level growth. It introduces a basic framework linking transfers to growth mediating processes and productive capacity outcomes. This is then used to organize and assess relevant findings from program evaluation studies. The paper concludes that social transfer programs can have positive effects on the productive capacity of poor and poorest groups; but further research is needed to identify and measure the size and significance of these effects. Ó 2011 Elsevier Ltd. All rights reserved. Key words — social transfers, social assistance, pro-poor growth, poverty, poverty reduction

1. INTRODUCTION

Rural Employment Guarantee Scheme (NREGS) in India reached 48 million households in 2009/10; a Safety Net program in Indonesia is expected to reach 6.5 million households in extreme poverty; Bolsa Familia reaches a further 12 million households; Ethiopia’s Productive Safety Net Programme (PSNP) reaches 1.7 million households; in South Africa, social assistance grants reach one half of the households, and Progresa/Oportunidades in Mexico and Bolsa Famı´lia in Brazil reach one quarter of all the households. In low-income countries the growth has been slower, but noticeable. Social transfer programs have the potential to make a significant contribution to global poverty reduction. Transfer programs in developing countries show diversity in program design. Pure income transfers, like South Africa’s Old Age Grant, supplement the purchasing power of households in poverty. In other countries, income transfers combined with access to basic services or improvements in infrastructure have been introduced. Mexico’s Oportunidades, for example, links transfers to school attendance and primary health care utilization with the objective of reducing intergenerational poverty persistence. Social transfers conditioned on labor supply link the transfer to local infrastructure development (as in Ethiopia’s PSNP or India’s NREGS). Integrated poverty reduction programmes, like Chile’s Chile Solidario or Bangladesh’s Challenging the Frontiers of Poverty Reduction—Targeting the Ultra Poor, combine transfers with a range of other interventions aimed at ensuring social and economic inclusion. In the main, these programs focus on extreme and persistent poverty. Examining the potential effects of social transfers on growth in a developing country context is challenging. The following three key questions underline these challenges and introduce the core issues. Which is the dimension of growth relevant to social transfers? The focus here is on growth among poorest households. Social transfer programs, even large-scale programs in middleincome countries, absorb a very small share of resources. Mexico’s Oportunidades and Brazil’s Bolsa Famı´lia absorb just below 0.5 of GDP. Transfers constitute a fraction of household consumption, typically around 20–30%, and are

A feature of the last decade has been the emergence of largescale social transfer programs in developing countries focused on households in extreme and persistent poverty. A large and fast growing literature assesses the poverty reduction effectiveness of these programs. 1 The broad findings from this literature, covering social transfer programs of different types and in different settings, suggest that, in combination with policies promoting growth and strengthening basic services, they have the potential to make a significant difference to global poverty and vulnerability. Less attention has been paid to examining the potential impact of social transfers on growth and development. 2 Permanent exit from poverty requires conditions for sustained income growth among households in poverty. This is well understood by policy-makers in developing countries, as demonstrated by their concerns to ensure social transfer programs are consistent with growth and development objectives. The main objective of this paper is to throw light on how social transfers may affect growth at the micro-level, by setting out a basic framework with which to collect, organize, and assess the findings from the program evaluation literature. In this paper, social transfers describe programs providing direct assistance in cash or kind to individuals or households, with the primary objective of reducing poverty and vulnerability. The focus is on programs providing regular transfers, as opposed to programs providing emergency and humanitarian assistance or programs providing short-term support. In the main, social transfer programs are noncontributory, financed from tax revenues (where international aid is involved, tax revenues are collected in a different jurisdiction), and are focused on the poorest groups. Increasingly, social transfer programs combine direct transfers with other interventions, access to basic services or credit for example. The emergence of large-scale social transfer programs in developing countries has been swift. Rough estimates suggest that currently between 0.75 and 1 billion people are reached by social transfers. 3 Most large-scale social transfer programs have emerged in middle-income countries, but there are also examples from low-income countries. The Minimum Living Standards Scheme in China reaches 24 million households in urban areas, and a rural equivalent currently under implementation reached 46 million households in 2010; the National

* I am grateful to the Editor and three anonymous referees for helpful comments and suggestions. Final revision accepted: May 11, 2011. 11

12

WORLD DEVELOPMENT

focused on households facing adverse social and economic conditions. They are, therefore, unlikely to produce macroeconomic effects on a scale sufficient to influence aggregate GDP growth. 4 The dimension of growth relevant to social transfers is growth among the households in poverty. How can social transfers aimed at raising consumption support growth processes? Some programs, perhaps even a majority, explicitly aim to influence households’ productive capacity, through facilitating human, physical or financial asset accumulation. Even among pure income transfer programs which focus on raising consumption, studies have observed that receipt of transfers is often associated with a range of household investment in human capital and other productive assets, often on a small scale. In developing countries, social transfers focused on those in poverty are developmental. They have effects which go beyond the supplementation of current consumption in so far as they strengthen the productive capacity of households in poverty with a potential to support income growth among beneficiary households and their communities. Under what circumstances could relatively low level transfers support growth processes? This is perhaps the hardest question to answer, but one that goes to the core of the issues covered in the paper. The poverty traps literature focused at the micro level suggests circumstances in which small increments (decrements) in the income or assets of households could lead to growth (decline) in household productive capacity and welfare. In particular, nonlinearities in the dynamics of production and consumption, as well as complementarities in asset accumulation, would be consistent with small transfers potentially having important effects upon households. The empirical evidence on the existence of poverty traps is patchy, but the approach throws important light upon this, third question. The paper draws on the growing impact evaluation literature on social transfer programs in developing countries. 5 In the main, this literature is focused on establishing whether program objectives are met. Information on improvements in productive capacity is often a by-product of program evaluation, and as a result important gaps remain. Taken as a whole, this literature provides reasonably strong findings on the extent to which social transfers lead to improvements in the productive capacity of beneficiary households, which in turn suggests potential effects on growth; but information on the size and significance of the growth effects is weak or nonexistent. The structure of the paper is as follows. Section 2 introduces a framework for examining the potential effects of social transfers on growth, and discusses two sources of heterogeneity in these effects. Section 3 reviews the findings from studies on social transfers relating to growth-mediating processes. Section 4 reviews and organizes findings on the outcomes of social

transfer programs on the productive capacity of households in poverty. A final section draws some policy implications and concludes. 2. A BASIC FRAMEWORK This section introduces a basic framework for tracing the growth effects of transfers by focusing on the linkages existing between social transfers, growth mediating processes, and productive capacity outcomes at the household level. Several approaches to constructing a framework are available. One approach examines the design properties of transfer programs and models the behavioral responses of beneficiary households, in a partial equilibrium setting. 6 A different approach is to model the welfare effects of transfers in a general equilibrium setting. 7 Both approaches have been applied successfully to the study of the welfare effects transfers. For the purposes of this paper, the first approach provides a limited perspective on the broader effects of transfers, and their financing, on communities and the economy, while the second approach throws limited light upon intermediate processes. In the paper, the focus is on developing a basic framework which could usefully help identify key intermediate growth processes, link these to outcomes, and suggest the likely direction of growth effects. The framework is a device to organize the findings from available studies. The second part of the section examines two potential sources of heterogeneity in the size of the effects, across programs. (a) Framework Figure 1 below summarizes a basic framework tracing the main linkages existing between social transfers and improvements in the productive capacity. They are discussed briefly below. The framework distinguishes the effects on beneficiaries and on nonbeneficiaries. Nonbeneficiaries are primarily affected through taxation. 8 In the literature, taxation impacts negatively on work and saving incentives, though the size of the effects is uncertain. 9 This issue is not taken further here, as the focus is on households in poverty. As regards beneficiary households, the framework suggests that transfers to households in poverty influence growth-mediating processes by lifting restrictions on their productive capacity. The impact on beneficiaries is measured in terms of program outcomes related to their productive capacity: human capital, physical and financial asset accumulation, labor supply, and local economy effects. These outcomes are expected to have positive effects on the income growth of households in poverty, except for the impact of transfers on labor supply, which can be

BENEFICIARIES Growth-mediating Processes: Credit and liquidity constraints Consumption and asset security Household resource allocation

Outcomes: Asset protection and accumulation inc. human capital Labour supply Local economy effects

(+) (+) (-/+) (+)

Expected direction of growth effects

Social Transfers

Taxation

Saving

(-)

Labour supply

(-)

NON-BENEFICIARIES

Figure 1. A basic framework.

SOCIAL TRANSFERS AND GROWTH

positive or negative. 10 Information on productive capacity outcomes is available for a good number of social transfer programs and will be presented and assessed in Section 4. Improvements in the productive capacity of households in poverty might not lead directly to market income growth. The absence of employment opportunities could ensure that improved productive capacity remains un(der)used and declining economic opportunity, lower wages, or lower commodity prices, might lead to contracting market income. The focus of the discussion below will be primarily on the effects of transfers on the productive capacity of households in poverty. A useful framework for tracing the growth effects of social transfers would need to identify the growth-mediating processes through which transfers lead to these outcomes at the household level. Three such processes identified in the literature are particularly relevant here (Banerjee & Duflo, 2004). Firstly, social transfers could lift credit constraints. Credit markets often exclude the poor and poorest, but regular and reliable transfers can help overcome barriers in access to credit. This can work in two ways: through enhancing the saving capacity of poor households; or through facilitating improved access to credit. Secondly, social transfers could provide greater certainty and security in consumption and thus investment outcomes. Poor households have fewer buffers to protect their consumption and their assets against hazards. Insurance services seldom reach the poor and poorest. Uncertainty and insecurity make investment especially risky and, therefore, undermine investment in productive capacity. Transfers can provide increased security and in the process make investment possible. Thirdly, social transfers could facilitate improved household resource allocation. Household resources allocation can be less than optimal if they experience poverty. Social transfers can help overcome cost restrictions, such as service user-fees, or migration costs. The design of social transfer programs can generate additional effects by influencing decision-making processes within the households. Social transfers channeled through the mother or carer can ensure greater investment in children’s education and health. These growth mediating processes suggest ways in which social transfers could generate outcomes which facilitate growth among households in poverty. (b) Two potential sources of heterogeneity in effects across programs In view of the diversity of program design and implementation, and because of the importance of context, similar A. Linear effects of transfers Future income

13

programs could lead to very different effects, and different programs could generate similar effects. It is important to take account of two sources of heterogeneity in observed effects. A first important source of possible heterogeneity in the effects of transfers on growth arises from the presence of nonlinearities in the dynamics of consumption and production at the household level. The poverty trap literature demonstrates analytically that nonlinearities in the function linking current and future income can generate large and persistent effects from relatively small changes in household resources. Figure 2 below will help illustrate the points made here. Suppose that the rise in consumption translates itself directly into an increase in productivity, via improvements in nutrition for example. In Panel A, it is assumed the effects of nutrition on work capacity are linear. Therefore, transfers equivalent to a–b and c–d can at best lead to proportionate increases in future income a0 –b0 and c0 –d0 , respectively. In this case, income growth would depend solely on the amount of the additional income transferred. Now take Panel B, where the function linking current to future income is assumed to be nonlinear. Transfers equivalent to a–b and c–d have different effects on future income. The transfer a–b has a less than proportionate effect on future income; while the transfer c–d generates a more than proportionate increase in future income, as in d0 –c0 . The literature on poverty traps suggests several potential factors responsible for nonlinearities in the dynamics of consumption and production: thresholds in nutrition and work capacity (Dasgupta, 1997; Dasgupta & Ray, 1986); thresholds in asset collateral in accessing credit (Banerjee, 2001, 2005); asset thresholds in production (Barrett & McPeak, 2005; Barrett & Swallow, 2006; Carter & Barrett, 2006); liquidity constraints in human capital investment (Baland & Robinson, 2000). The insights from poverty traps perspectives apply to household resource allocation too (Duflo, 2006; Udry, 1996). Context is important. 11 A second source of heterogeneity in effects across programs is associated with nonincome effects of transfers, arising from program design and implementation. In particular, household investment in productive capacity could be linked to design features of social transfer programs in developing countries, such as preferential access and utilization of basic services, conditions attached to transfers, and payment modalities. Social transfers conditional on schooling or health utilisation, for example, are expected to encourage investment in human capital over and above the additional income effects from the transfer. Figure 3 below illustrates the effects of schooling conditions on household consumption and investment (Das, Do, & Ozler, 2005; Skoufias, 2005).

B. Non-linear effects of transfers Future income

d’

d’

c’ c’

b’ a’ a

b

c

d Current income

0

a

b

c

d

Current income

Non-linear effects can arise in the presence of credit constraints, ‘lumpiness’ in asset accumulation or production, or poverty-related constraints in household resource allocation

Figure 2. Nonlinear effects of transfers.

14

WORLD DEVELOPMENT

(a) Alleviating credit constraints

All other goods

c i a h

g e

o

f

z

b

d

Schooling

Figure 3. Conditions and nonincome effects of transfers.

In the figure, ab represents the quantities of schooling and all other goods available to poor households, given their present income. An unconditional transfer equivalent to eg makes it possible for poor households to invest more in schooling and raise their consumption of all other goods, as their budget line is now cd. Suppose that in the absence of the transfer a poor household is at h, then with the transfer household consumption will be in the range ig (at i the household consumes more of other goods but does not invest in more schooling, and at g the household invests in more schooling but does not raise consumption of all the other goods; points in the range ig, therefore, represent feasible combinations). Program designers concerned that poor households consume at least oz schooling could make the transfer conditional on achieving this minimum level of schooling. The new budget line for the poor households is indicated by aegd. A poor household at point h before the transfer will now need to invest more in schooling, at least to point g, to secure entitlement to the transfer. A poor household at point e before the transfer will now be able to move to a point in the range gf. Given conditions represented in the figure, a schooling condition will ensure investment in schooling over and above an unconditional transfer. Taking account of the possible influence of income receipt on the bargaining power of household members, policy designers who believe mothers have a stronger preference for higher consumption/investment among their children could make them recipients of the transfer, in the expectation that this would strengthen their bargaining power within the household and ensure cash transfers do result in improvement in children’s consumption. For example, a poor household consuming at point e before the transfer will be more likely to consume at, or close to, f if the transfer is paid to the mother. Nonincome effects arising from the design and implementation of transfer programs can, therefore, further facilitate investment, and constitute an additional source of heterogeneity in the effects of transfers on productive capacity. This section provided a basic framework for examining the potential effects of social transfers on productive capacity outcomes. The second half of the section then considered potential sources of heterogeneity in the effects on growth across different programs, due to context and diversity in program design and implementation. The next section will review the relevant findings from studies assessing social transfer programs. 3. SOCIAL TRANSFERS AND GROWTH-MEDIATING PROCESSES This section organizes available findings from studies on social transfer program throwing light on growth-mediating processes.

The theoretical and empirical literature confirms that households in poverty face binding credit constraints. They have difficulties in providing collateral to secure loans from financial institutions, especially as the urgency of their consumption needs makes them more likely to default (Banerjee, 2005). The development of micro-credit institutions in developing countries is a policy response to the well documented barriers to accessing credit. However, the literature on microfinance and micro-credit institutions also documents the difficulties involved in reaching the poorest households. Social transfers could help lift credit constraints for poor and poorest households in at least two ways. Firstly, social transfers, providing they are regular and reliable, can encourage small-scale saving and investment, opening another route to lowering credit constraints. Secondly, social transfers could prove effective, in combination with other interventions, in enabling access to credit. There are indications across a variety of social transfer programs, in middle- and low-income countries, that beneficiaries are able to save and invest a fraction of their income following the receipt of transfers, and also that access to credit can be facilitated by the transfer. The evidence is largely qualitative, but has been carefully measured in some programs. In Bolivia, a social pension Bono Dignidad is paid once a year to persons aged 60 and over. At US$246, it is a significant injection of liquidity for rural farmers who have land but lack sufficient cash or credit to purchase seeds and other agricultural inputs. A study has estimated that among pension beneficiaries in rural areas, overall consumption rises by twice the amount of the benefit, suggesting that improved household production was facilitated by the transfer (Martinez, 2007). The effect is observed only among rural households with land, and is stronger for goods which are typically produced by these households, such as diary, meat, and vegetables. Social transfers can also help lift credit constraints for poorest households as a component of a package of interventions including micro-credit. Direct evidence of improved access to credit is available for BRAC’s Challenging the Frontiers of Poverty Reduction—Targeting the Ultra Poor program in Bangladesh. This program provides a mix of transfers in kind and cash to households in extreme poverty in preparation for more standard micro-credit programs after 18 months. A sample of selected beneficiary households was compared with a sample of eligible but nonselected households, the latter slightly better off, 12 in 2002 at the start of the program and in 2005 (Ahmed, Rabbani, Sulaiman, & Das, 2009; Rabbani, Prakash, & Sulaiman, 2006). Beneficiary households showed significant improvements over time in the incidence and size of loans they held, in part explained by their access to the micro-credit component of the program. The studies also show a shift in the motivation for credit among selected households. In 2002, credit is primarily a means of smoothing out consumption, but in 2005 the dominant motivation is investment in productive assets. Studies have also shown nonincome effects of social transfers relating to access to credit. In Brazil, Preˆvidencia Rural, a social pension, was introduced in 1991 to cover informal workers and their households. The regularity of the pension benefit enables pension beneficiaries to access loans from banks, by showing the magnetic card which is used by them to collect their pensions (Schwarzer, 2000). A detailed study of the program observed a high incidence of investment in productive capital among beneficiaries (Delgado & Cardoso, 2000).

SOCIAL TRANSFERS AND GROWTH

While providing important insights, the evidence reviewed is not systematic across all social transfers. In some transfer programs, especially pure income transfers such as social pensions or human development transfer programs, access to credit is simply a by-product of the income transfer and not an explicit objective. The capacity of social transfers to help lift credit constraints is likely to vary across programs, target groups and environments. These effects are stronger where credit constraints are directly targeted, as in Bangladesh. Broadly, transfers have greater effects among rural households with deficits in complementary ‘productive’ assets (e.g., inputs, labor). These studies also indicate that transfers can support growth in household production and, therefore, have multiplier effects. It is harder to observe these effects in urban settings. (b) Improving consumption and asset security Households in poverty adopt a range of strategies to protect their consumption and assets from the impact of hazards. Insecure and precarious livelihoods are bound to limit investment by poor households. The gaps in effective policy instruments in this area are well recognized. The theoretical and empirical literature finds that insurance markets seldom reach those in poverty, with the implication that they remain insufficiently protected (Dercon, 2005; Jalan & Ravallion, 1999). Insufficient insurance protection has damaging effects on the ability of the poor and their households to exit poverty and benefit from economic opportunity. Insecurity leads to inefficient use of resources by those in poverty, for example, by forcing rural poor households to opt for low-risk/low-return crops and production methods, which reduces growth opportunities (Morduch, 1995). Insecurity also forces poor households to hold liquid but less productive assets (Dercon, 2003). It could also lead to distortions in inter-temporal resource allocation, forcing a focus on current consumption in preference to investment. This is typically the case when households withdraw children from school or ‘economize’ on health care in response to crises. In the absence of security, responding to short-term shocks can lead to poverty persistence. Regular and reliable social transfers can improve household consumption and asset security, firstly through supplementing household income, and, therefore, ameliorating the impact of consumption shocks; and secondly through integrating insurance features protecting consumption, assets, and investment (Lichand, 2010). In practice, few social transfers have explicit insurance components. Some variants of public works are an exception. The National Rural Employment Guarantee Scheme in India, for example, provides entitlement to up to 100 work days for unemployed rural households, on demand. The design of the employment guarantee is specifically aimed at smoothing consumption among rural households (Kannan, 2006). The Unemployed Heads of Households program in Argentina was crucial in supporting households affected by the 2001 crisis in Argentina (Galasso & Ravallion, 2005). By contrast, social pensions or human development transfers programs have no explicit insurance component, aside from the buffer provided by the supplementary income transferred. This is mainly because transfers are fixed in level, and up-rated at regular intervals. Human development conditional social transfer programs in Latin America have at best indirect insurance components, for example, through the enforcement of schooling conditions which are sustained through crises (de Janvry, Finan, Sadoulet, & Vakis, 2006). This is an important

15

deficiency of these types of social transfers in both middleincome and low-income countries, one which became apparent in the context of the 2008 global financial crisis. In the context of the crisis, program agencies reacted by expanding the coverage of programs through inclusion of adolescents aged 16–17 and by speeding up the processing of applications; and in other cases by raising the level of the transfer. The crisis will generate some ‘learning’ among program agencies on how best to address covariate shocks. (c) Improving household resource allocation Household dynamics is an important factor in determining the capacity of poor households to access economic opportunity, and especially unequal bargaining power within the household. Udry (1996) finds that differentials in yields in men’s and women’s plots in Burkina Faso are accounted for by differences in input intensity, especially labor and fertilizers. These differences in turn could be explained by women’s concerns to avoid having their plots and yields expropriated by their husbands. Decisions about allocation of time and effort made within the household, often reflecting unequal bargaining influence, can generate inefficiencies. These in turn could become an important impediment to investment and income growth among the poor. There is an emerging literature looking at the impact of transfers on intra-household resource allocation, with implications for income growth among those in poverty (Duflo, 2006). In human development social transfers, it is common that transfers are paid to the mother, in the expectation that children stand to benefit more directly if it is the mother who receives them. An interesting issue is whether this modality of transfer payment has any effect on intra-household resource allocation, especially in the context of bargaining models suggesting that household decision making is strongly influenced by relative shares of income contributed by different household members (Haddad, Hoddinott, & Alderman, 1997). Studies of the impact of Oportunidades transfers on patterns of consumption in Mexico observe a shift in consumption toward children-related goods and services (Attanasio & Lechene, 2002). This finding could be interpreted as supporting the view that transfers are capable of effecting changes in intra-household resource allocation. A related issue is whether transfers encourage changes in household labor allocation, especially through migration. This point will be discussed in more detail below, but studies on Oportunidades identify impacts of transfers on domestic and international migration (Angelucci & De Giorgi, 2006; Stecklov, Winters, Stampini, & Davis, 2005). The findings on the effects of transfers on household resource allocation are far from systematic, but they are especially insightful. This issue has not been sufficiently well researched. It can be argued that changes in household allocation have the greatest potential to make a large and long-term contribution to poverty eradication and prevention. Further research is needed to identify with greater precision and certainty these effects.

4. SOCIAL TRANSFERS AND PRODUCTIVE CAPACITY OUTCOMES This section reviews the findings evidence on the effects of social transfers on changes in the productive capacity of households in poverty.

16

WORLD DEVELOPMENT

(a) Asset protection and accumulation Several studies have observed improvements in asset holdings among beneficiaries of social transfer programs. For Oportunidades, Gertler, Martinez, and Rubio-Codina (2005) estimate that, on average, around 12% of transfers to beneficiaries were invested in productive assets. Sadoulet, de Janvry, and Davis (2001) who compared transfers under Oportunidades and Procampo—a human development conditional transfer program, and a productive transfer to owners of small farms, respectively—find that the latter had income multipliers of around 1.5–2.6. The main effect of Oportunidades was instead on human capital investment. Ethiopia’s Productive Safety Net Programme addresses persistent food insecurity through a public works component, which helps prevent asset depletion and build community assets. Participants in the public works component can also access productivity-enhancing interventions and asset transfers. A study comparing outcomes between this group and a control group 18 months after a baseline survey observed improved food security (0.36 months), increased borrowing for productive purposes (12 percentage points), and increased use of fertilizers (10.7 percentage points) (Gilligan, Hoddinott, & Seyoum Taffesse, 2008). A dimension often overlooked is investment in housing. Studies of activities associated with first receipt of social pension benefits have indicated that housing investment is important (Schwarzer, 2000). Human development-focused social transfer programs deliver large improvements in school enrollment attendance among the poorest beneficiaries, especially where they explicitly target schooling through conditions. This is a common finding for programs not only in Latin America, but also in Bangladesh. The transfers are made conditional on school age children attending school. In Oportunidades, the design of the transfer strengthens the incentives for household investment in girls’ schooling at secondary level, by raising the level of the transfer for them. This program has ensured high school attendance for girls in rural areas. Many Latin American programs also ensure primary health care utilisation through conditions. Oportunidades has been very successful in achieving higher rates of utilization and improved health status among beneficiaries (Skoufias, 2005). Positive findings are not limited to transfer programs specifically targeting human development or children. Studies on the impact of social pension receipts in South Africa and Brazil, for example, find that households with a pension beneficiary have higher enrollment rates among children of school age, and improved health status (Carvalho, 2008a; Case & Menendez, 2007; Duflo, 2003). This also applies in lowincome countries, although the effects are strongly mediated by supply-side conditions in these countries. Raising educational and health service utilization in low-income countries requires that attention is paid to improving infrastructure in line with growth in service demand. Supply-side policies implemented alongside such programs may be necessary to ensure that services do not decline in standards due to increased demand. These findings describe short-term effects, but investment in human capital will raise the future productive capacity of beneficiary children. Todd and Wolpin (2006) extend the mediumterm estimates of impact with the help of a structural model representing household fertility and schooling decisions. They predict that children participating in Oportunidades will complete the schooling cycle with almost one extra year of education. Given the strong correlation existing between years of schooling and labor market earnings, it is expected that trans-

fers will contribute to improved labor market outcomes for beneficiaries. It is, of course, too early for the actual effects to be observed. (b) Labor supply effects Standard economic models predict that income transfers will have adverse labor supply effects among beneficiary households, on participation and hours of work (Saez, 2006). A concern with minimizing labor supply disincentives has dominated discussions on the optimal design of transfers in developed countries, and has informed policy discussions around welfare reform (Moffitt, 2002). In developing countries, studies have examined this issue in some detail, and in a variety of contexts. 13 Some transfer programs have direct labor supply objectives, for example, programs with a work condition or programs aiming to reduce child labor or improve school attendance. Among programs focused on human capital investment, a reduction in the labor supply of children of school age is an implicit objective of programs requiring school attendance. The success of these programs in reducing child labor has been mixed. Given that children’s time could be split into schooling, working and other activities, education subsidies providing incentives for schooling may not, by themselves, result in a proportionate reduction in the incidence of child labor (Ravallion & Wodon, 2000). Impact assessments of Bangladesh’s Food for Education program suggest that the reduction in child labor time was less than proportionate to the rise in schooling, and in Mexico’s Oportunidades the evidence of a rise in enrollment and attendance from beneficiary children is stronger and more compelling than evidence that education subsidies reduced child labor (Ahmed & del Ninno, 2002; Skoufias & di Maro, 2008; Skoufias & Parker, 2001; Skoufias, Unar, & Gonza´lez-Cossı´o, 2008). The success of Brazil’s Programme for the Eradication of Child Labor (PETI) in reducing child labor was in large measure due to the requirement that participating children spend time in afterschool activities (Yap, Sedlacek, & Orazem, 2002). 14 Transfers targeted on older people are likely to reduce their labor supply even in the absence of inactivity tests as a requirement of eligibility. Social pensions in developing countries very rarely include inactivity test as an eligibility requirement, a feature which distinguishes them from noncontributory pensions in developed countries (Barrientos, 2006). Labor force participation rates for those eligible to receive a pension are very low in South Africa, and decline rapidly when individuals reach the age of pension entitlement (Lam et al., 2004). This is to be expected, as the combination of the generosity of the pension benefit and the means test provide strong incentives for withdrawal from the labor market. Given a high incidence of co-residence of pensioners and their extended households in developing countries, and evidence of widespread pension income sharing, it is of some interest whether labor supply effects of social pensions can also be observed among nonpensioners. This has been investigated in some detail in South Africa. Bertrand et al. (2003) suggest that the effects of pension income on hours of work and employment of 15–50 year-olds co-residing with pension beneficiaries in South Africa are significant. Using 1993 cross-section data from three-generation African households, they estimate that 15–50 year-old co-residents with a pension eligible person undertake on average 6.4 fewer hours of work and a 4.3% lower probability of employment. They also find that the labor supply effect is strongly significant among 15–50 year-old males, but not significant among females.

SOCIAL TRANSFERS AND GROWTH

Critically, however, this study samples co-residents only and misses out nonresident household members. Posel et al. (2004) note that a high proportion of household members of working age migrate to urban centers in search of work—as many as 30% of rural households in South Africa have a migrant. Replicating the work of Bertrand et al., but now including migrant household members, Posel et al. find that the negative association between pension receipt and labor supply in that study becomes positive, and conclude that 15–50 yearold individual members of a household with a pension eligible person have a 3.2% higher probability of employment. Disaggregating by gender, they find no significant effect associated with pension income received by male pensioners on labor supply of adult household members, but a strong and positive effect when the pension recipient is female. They suggest that pension income received by women is particularly important for rural households, “not only because it helps prime-age women overcome income constraints to migration, but also because it makes it possible for grandmothers to support grandchildren” (p. 24). A recent study has confirmed this finding with longitudinal data helping to track household changes around pension receipt (Ardington et al., 2009). Overall, the effects of social transfers on labor supply could be summarized as follows. Firstly, an increase in unearned income leads to a reduction in the labor supply of children and older people, especially where education subsidies or pensions target these groups specifically. The effects are stronger where social assistance programs include schooling conditions for children or inactivity tests on older people. Secondly, the reduction in labor supply from these groups is often compensated for by changes in the labor supply of other household members. In the case of Oportunidades, a small reduction in child labor appears to have been compensated by an increase in the labor supply of adults, but otherwise adult labor remained unaffected by the income transfers (Skoufias & Parker, 2001). To the extent that social transfers lift credit and care constraints, they could have the effect of encouraging additional labor supply from other household members, as appears to have been the case in Chile (Carneiro, Galasso, & Ginja, 2008). In these circumstances, the impact of social transfers on productivity and growth could well be positive. 15 There is scant evidence that social transfers have large adverse labor supply effects on beneficiaries and their households. It will be important to reassess this finding in the future, especially as transfer programs expand their coverage, improve the level and generosity of transfer levels, and implement stricter recertification procedures. Attention will also need to be paid to structuring pathways to employment among beneficiaries of transfer programs. 16 (c) Local economy effects In areas with high poverty incidence, household income growth can be constrained by community-level factors, such as the absence of adequate infrastructure or the scarcity of local liquidity and trade. Transfer programs conditional on labor supply could in principle have an impact on both these factors, by transferring income to households and improving liquidity, while at the same time upgrading available infrastructure. In practice, the assessment of such programs has on the whole been very mixed. Many public works programs transfer only a fraction of their budget to beneficiary households, due to the cost of inputs, equipment and technical advice; and in some cases the value of the newly created infrastructure is marginal. Studies assessing the impact of public works programs find that their impact at the community le-

17

vel depends to an important extent on program design, especially the level and periodicity of the transfer (McCord, 2007). Transfer programs can also improve liquidity and trade at the local economy level by stimulating effective demand. Only a handful of studies have addressed this question. In the context of direct income transfer programs, existing studies suggest social transfers could generate multiplier effects, especially in poor rural areas, but evidence on multiplier effects from social transfers must be considered with great care. 17 Studies on Oportunidades have observed an increase in consumption and productive assets among nonbeneficiary households in treatment areas, compared to nonbeneficiaries in control areas (Angelucci & de Giorgi, 2009). This can be interpreted as reflecting local economy effects, and applies more strongly to nonbeneficiary households with low asset levels at the start of the program (Barrientos & Sabate´s-Wheeler, 2010). The knowledge base for the presence of local economic effects is at present very ‘thin’, as few studies on this issue are available. Impact evaluation studies normally focus on beneficiaries, and disregard wider effects. This is an important area for further research. One aspect that could well turn out to be significant in the context of social transfers in poorer areas concerns potential ‘demonstration effects’ and community development effects. Demonstration effects could be responsible for observed increases in health care utilisation by nonbeneficiary households in Oportunidades (Handa, Huerta, Perez, & Straffon, 2000). There is only limited information on the impact of transfers on community institutions and action. These can be important in lifting the constraints on local economy growth in areas with high poverty incidence. 5. CONCLUSIONS The paper studies the potential growth effects of social transfer programs in developing countries. It examined how transfers could strengthen the productive capacity of households in poverty by setting out a basic framework with which to collect, organize, and assess the relevant findings from existing evaluation studies. The objective was to discuss what we know and what we need to know about these effects. This is challenging in a developing country context. Despite their rapid growth, social transfers in developing countries redistribute a very small share of GDP to households in extreme or persistent poverty, often facing adverse economic circumstances. The paper focused on the impact of transfer programs on micro-level growth among poor and poorest groups, as it is unlikely that they could influence aggregate GDP growth. This is the dimension of growth relevant to this study. The basic framework mapped out the potential linkages existing between social transfers, growth-mediating processes, and outcomes relating to the productive capacity of households in poverty. The framework focused attention on the potential role of social transfers in lifting credit constraints, providing greater security, and enabling a re-allocation of household resources, leading to improved household investment and productive capacity. Given favorable economic conditions, improved productive capacity among poor households can be expected to lead to growth outcomes. The paper emphasized the role of context and of diversity in program design and implementation in generating heterogeneity in effects across programs. The paper then examined the findings from the impact evaluation literature on social transfers in order to identify empirical counterparts to the growth-mediating processes

18

WORLD DEVELOPMENT

identified in the framework, and to assess reported productive capacity outcomes. The studies reviewed in the paper cover a variety of social transfer programs and contexts. The findings in these studies are clearer and stronger when they come from middle-income countries and from human developmentfocused programs combining income transfers and access to basic services. Overall, the findings suggest that social transfers are capable of having measurable effects on the productive capacity of households in poverty, and of influencing microlevel growth. More research is needed before a comprehensive and systematic assessment of the growth effects of social transfers can be offered. It is much harder to identify the size and significance of nonincome effects, and it is also harder to get a sense of potential growth effects in low-income countries. The discussion in the paper focused on outcomes relating to the productive capacity of households, but further research is needed to identify the exact circumstances and conditions in which specific types of transfers generate sustainable income growth among the poor and the poorest groups. And more research is needed to quantify these effects with some precision. The available literature on the impact of social transfers focuses on short term effects, but further research is needed to examine medium and longer term effects which are crucial to sustainable exit from poverty. The rapidly expanding evidence base on large-scale transfer programs in developing countries, and the emergence of new programs, provides an opportunity to address these knowledge gaps. How to maximize the growth effects of social transfer programs? The discussion in the paper offers several pointers. In order to maximize their effects on investment, social transfers need to be regular and reliable. Regularity involves incorporating within the program clear and transparent rules on eligibility, and on the time period for the entitlements. Duration is important too. The time period during which households are supported needs to be long enough to influence households’ consumption–investment decisions, and not too long to generate dependency. Beneficiary households must have clear and credible information on the size, time, and schedule of entitlements. The level of the transfer is important too. Broadly, and keeping all other design features constant, a

higher level of transfers should result in a higher level of household investment, but attention needs to be paid to thresholds. Where transfers are set at a very low level relative to household consumption, household investment following the transfer will be at best marginal. Significant levels of household investment are more likely to kick in after basic consumption levels are reached in a reliable and sustainable way. It is important to ensure that eligibility conditions do not incorporate incentives for asset depletion. Transfer programs which make eligibility dependent on relatively liquid asset holdings, such as livestock, could generate incentives for divestment of assets. The same applies to eligibility conditions focused on saving. Eligibility conditions dependent on timeinvariant variables less easily manipulated by potential beneficiaries have fewer disincentive effects, and greater selection effectiveness. Transfers should be designed in ways that facilitate household allocation of productive resources. Changes in household composition and resource allocation in response to social transfers are a large knowledge gap. On this same point, it is important that eligibility conditions do not include inactivity tests. This is important, for example, in the context of social pensions or family allowances. Social transfer programs should take care not to limit the use of productive assets available to the household, either intentionally or unintentionally. The design of social transfers should pay attention to appropriate complementary asset accumulation interventions. This should include human capital and other productive assets. Few social transfer schemes include asset protection components, but it is worth thinking carefully how these components could be incorporated. There are few studies addressing the extent to which the financing of social transfers has adverse incentive effects on nonbeneficiaries. This is not a significant issue at present, because most transfer programmes absorb a very small share of resources in middle-income countries, while being largely externally financed in low-income countries. However, in the medium and longer run, when programs are increasingly financed from domestic resources, distortions associated with higher levels of taxation will need to be studied carefully.

NOTES 1. Overviews are available (Grosh, Del Ninno, Tesliuc, & Ouerghi, 2008; Hanlon, Barrientos, & Hulme, 2010), as well as studies of specific types of programmes (Fiszbein & Schady, 2009; Holzman, Robalino, & Takayama, 2009). For programme information and references see the Social Assistance in Developing Countries database (Barrientos, Nin˜oZarazu´a, & Maitrot, 2010), available from http://papers.ssrn.com/sol3/ papers.cfm?abstract_id=1672090. 2. The linkages between social protection and growth are reasonably well understood in developed countries with comprehensive welfare states (Atkinson, 1999).

3. This estimate is based on the information contained in the database referred to in Footnote 1. 4. Santiago Levy notes that “Progresa-Oportunidades will not directly increase growth. . .The Program can contribute to growth as it gradually fosters a healthier and more educated labor force and it allows poor households to make more productive investments that have longer horizons and higher expected returns. But that will not have a first order

effect on the country’s growth rate. This is because the first two deciles of the income distribution receive less than 2.5% of aggregate income in Mexico. . .If all poor households’ income (net of Progresa-Oportunidades transfers) increased by 5% a year, aggregate income would increase, at most, by an additional 0.12% a year over the growth rate without the program” (Levy, 2006, pp. 19–20). 5. The studies referred to in the paper were selected on the basis that their findings relied on appropriate data and methods. The studies are based on household level survey data, often collected for the purposes of evaluating transfer programs, employ econometric techniques which are appropriate to the data, and describe the structure of the data and estimation results in sufficient detail. The majority of the studies discussed have been published in peer reviewed journals or discussion paper series.

6. See for example Besley and Coate (1992) for transfers with a work requirement, or Skoufias (2005) for human development transfer programs. 7. See Coady (2004) for a general equilibrium approach.

SOCIAL TRANSFERS AND GROWTH 8. This is a simplification; the sources of financing for social protection in developing countries vary across low and middle income countries. 9. See Kaplow (2006) for a survey.

19

Cortez Reis & Camargo, 2007; Edmonds, 2006; Ferro & Nicollela, 2007; Foguel & Paes de Barros, 2008; Freije, Bando, & Arce, 2006; Hoddinott & Skoufias, 2004; Lam, Leibbrandt, & Ranchhod, 2004; Posel, Fairburn, & Lund, 2004; Skoufias & di Maro, 2008; Skoufias & Parker, 2001) 14. The transfer component of the PETI program has been integrated into Bolsa Famı´lia.

10. See Moffitt (2002) for a survey. 11. For an example see the discussion of the differential impact of Bono Dignidad on urban and rural pensioner households below.

15. Some studies find marginal shifts in employment to non-farm activities (Skoufias & di Maro, 2008; Skoufias et al., 2008).

12. Households were ranked through a participatory wealth assessment. Households in the bottom two categories were identified as ultra-poor, around 25% of all households. One fifth of the ultra poor were selected to participate in the program based on eligibility criteria.

16. Active labor market policies are beginning to receive attention in the context of discussions over exit strategies in transfer programs. For a discussion of these issues in developed countries see OECD (2003).

13. There is a growing literature on this issue (Alzu´a, Cruces, & Ripani, 2010; Ardington, Case, & Hosegood, 2009; Bertrand, Mullainathan, & Miller, 2003; Bourguignon, Ferreira, & Leite, 2003; Carvalho, 2008a,b;

17. Multiplier effects from social transfers are usually identified in the context of the assets and consumption of beneficiaries, rather than their communities as a whole (Sadoulet et al., 2001).

REFERENCES Ahmed, A. U., & del Ninno, C. (2002). Food for Education Programme in Bangladesh: An evaluation of the Impact on Educational Attainment and Food Security. Discussion Paper 138. Washington DC: International Food Policy Research Institute. Ahmed, A. U., Rabbani, M., Sulaiman, M., & Das, N. C. (2009). The impact of asset transfer on livelihoods of the ultra poor in bangladesh. Research monograph 39. Dhaka: BRAC. Alzu´a, M. L., Cruces, G., & Ripani, L. (2010). Welfare programs and labour supply in developing countries. Experimental evidence for Latin America. Documento de Trabajo 95. La Plata: CEDLAS-UNLP. Angelucci, M., & De Giorgi, G. (2006). Aid and Migration: The impact of PROGRESA on the timing and size of labour migration. Discussion Paper 1955. Bonn: IZA. Angelucci, M., & de Giorgi, G. (2009). Indirect effects of an aid program. How do cash transfers affect inelegibles’ consumption. American Economic Review, 99(1), 486–508. Ardington, C., Case, A., & Hosegood, V. (2009). Labour supply responses to large social transfers: Longitudinal evidence from South Africa. American Economic Journal: Applied Economics, 1(1), 22–48. Atkinson, A. B. (1999). The economic consequences of rolling back the welfare state. Cambridge: MIT Press. Attanasio, O., & Lechene, V. (2002). Tests of income pooling in household decisions. Review of Economic Dynamics, 5, 720–748. Baland, J.-M., & Robinson, J. A. (2000). Is child labor inefficient? Journal of Political Economy, 108(4), 663–679. Banerjee, A. V. (2001). Contracting constraints, credit markets and economic development. Department of Economics Working Paper 0217. Cambridge MA: MIT. Banerjee, A. V. (2005). The two poverties. In S. Dercon (Ed.), Insurance against poverty (pp. 59–75). Oxford: Oxford University Press. Banerjee, A. V., & Duflo, E. (2004). Growth theory through the lens of development economics. Working Paper 05-01. Cambridge, M.A.: MIT. Barrett, C. R., & McPeak, J. G. (2005). Poverty traps and safety nets. In A. de Janvry, & R. Kanbur (Eds.), Poverty, inequality and development: Essays in honour of Erik Thorbecke. Norwell MA: Kluwer Academic Publishers. Barrett, C. R., & Swallow, B. M. (2006). Fractal poverty traps. World Development, 34(1), 1–15. Barrientos, A. (2006). Pensions for development and poverty reduction. In G. L. Clark, A. H. Munnell, & M. Orszag (Eds.), Oxford handbook of pensions and retirement income (pp. 781–798). Oxford: Oxford University Press. Barrientos, A., Nin˜o-Zarazu´a, M., & Maitrot, M. (2010). Social assistance in developing countries database version 5. Report Manchester: Brooks World Poverty Institute. Barrientos, A., & Sabate´s-Wheeler, R. (2010). Do transfers generate local economy effects? Applied Economics. doi:10.1080/00036840903493259.

Bertrand, M., Mullainathan, S., & Miller, D. (2003). Public policy and extended families: Evidence from pensions in South Africa. World Bank Economic Review, 17(1), 27–50. Besley, T., & Coate, S. (1992). Workfare versus welfare: Incentive arguments for work requirements in poverty-alleviation programmes. American Economic Review, 82(1), 249–261. Bourguignon, F., Ferreira, F. H. G., & Leite, P. G. (2003). Conditional cash transfers, schooling and child labour: Microsimulating Bolsa Escola. World Bank Economic Review, 17(2), 229–254. Carneiro, P., Galasso, E., & Ginja, R. (2008). The impact of providing psycho-social support to indigent families and increasing their access to social services: Evaluating Chile Solidario. London: University College London, mimeo. Carter, M. R., & Barrett, C. R. (2006). The economics of poverty traps and persistent poverty: An asset-based approach. Journal of Development Studies, 42(2), 178–199. Carvalho, I. E. d. (2008a). Household Income as a determinant of child labour and school enrollment in Brazil: Evidence from a social security reform. IMF Working Paper WP/08/241. Washington DC: IMF. Carvalho, I. E. d. (2008b). Old-age benefits and the labour supply of rural elderly in Brazil. Journal of Development Economics, 86, 129–146. Case, A., & Menendez, A. (2007). Does money empower the elderly? Evidence from the Agincourt demographic surveillance site, South Africa. Scandinavian Journal of Public Health, 35(Suppl. 69), 157–164. Coady, D. P. (2004). Designing and evaluating safety nets: Theory, evidence, and policy conclusions. FCND Discussion Paper 172. Washington DC: International Food Policy Research Institute. Cortez Reis, M., & Camargo, J. M. (2007). Rendimientos domicilia´res com aposentadorias e pensoˆes e as decisoˆes dos jovens quanto a` educaca˜o e a participacßa˜o na forca de trabalho. Texto para Discussao 1262. Rio de Janeiro: IPEA. Das, J., Do, Q.-T., & Ozler, B. (2005). Reassessing conditional cash transfer programs. World Bank Research Observer, 20(1), 1–28. Dasgupta, P. (1997). Nutritional status, the capacity to work, and poverty traps. Journal of Econometrics, 77(1), 5–37. Dasgupta, P., & Ray, D. (1986). Inequality as a determinant of malnutrition and unemployment. Economic Journal, 96(384), 1011–1034. Delgado, G. C., & Cardoso, J. C. (Eds.) (2000). A Universalizacßa˜o de Direitos Sociais no Brazil: A Preˆvidencia Rural nos anos 90. Brasilia: IPEA. de Janvry, A., Finan, F., Sadoulet, E., & Vakis, R. (2006). Can conditional transfer programmes work as safety nets in keeping children at school and from working when exposed to shocks? Journal of Development Economics, 79, 349–373. Dercon, S. (2003). Risk and poverty: A selective review (or: can social protection reduce poverty?). Oxford: Department of Economics, Oxford University, mimeo. Dercon, S. (Ed.) (2005). Insurance against Poverty. Oxford: Oxford University Press.

20

WORLD DEVELOPMENT

Duflo, E. (2003). Grandmothers and granddaughters: Old age pensions and intrahousehold allocation in South Africa. World Bank Economic Review, 17(1), 1–25. Duflo, E. (2006). Poor but rational? In A. V. Banerjee, R. Be`nabou, & D. Mookherjee (Eds.), Understanding poverty (pp. 367–378). Oxford: Oxford University Press. Edmonds, E. (2006). Child labor and schooling responses to anticipated income in South Africa. Journal of Development Economics, 81(2), 386–414. Ferro, A., & Nicollela, A. (2007). The impact of conditional cash transfer programmes on household work decisions in Brazil. Rio de Janeiro: PUC, mimeo. Fiszbein, A., & Schady, N. (2009). Conditional cash transfers. Reducing present and future poverty. Washington DC: The World Bank. Foguel, M. N., & Paes de Barros, R. (2008). The effects of conditional cash transfer programmes on adult labour supply: An empirical analysis using a times series cross section sample of Brazilian municipalities. Rio de Janeiro: IPEA, mimeo. Freije, S., Bando, R., & Arce, F. (2006). Conditional transfers, labour supply, and poverty: Microsimulating Oportunidades. Economı´a(Fall), 73, 124. Galasso, E., & Ravallion, M. (2005). Social Protection in a Crisis: Argentina’s Plan Jefes y Jefas. World Bank Economic Review, 18(3), 367–399. Haddad, L., Hoddinott, J., & Alderman, H. (Eds.) (1997). Intrahousehold resource allocation in developing countries. London: John Hopkins University Press. Gertler, P., Martinez, S., & Rubio-Codina, M. (2005). Investing cash transfers to raise long term living standards. Washington DC: The World Bank, mimeo. Gilligan, D. O., Hoddinott, J., & Seyoum Taffesse, A. (2008). The impact of Ethiopia’s Productive Safety Net Programme and its linkages. Journal of Development Studies, 45(10), 1684–1706. Grosh, M., Del Ninno, C., Tesliuc, E., & Ouerghi, A. (2008). For protection and promotion. The design and implementation of effective safety nets. Washington DC: The World Bank. Handa, S., Huerta, M. C., Perez, R., & Straffon, B. (2000). Poverty, Inequality and spillover in Mexico’s education, health and nutrition program. Washington DC: IFPRI, mimeo. Hanlon, J., Barrientos, A., & Hulme, D. (2010). Just Give Money to the Poor: The development revolution from the global South. New York: Kumarian Press. Hoddinott, J., & Skoufias, E. (2004). The impact of PROGRESA on food consumption. Economic Development and Cultural Change, 53(1), 37–61. Holzman, R., Robalino, D., & Takayama, N. (2009). Closing the coverage gap. The role of social pensions and other retirement income transfers. Washington DC: The World Bank. Jalan, J., & Ravallion, M. (1999). Are the poor less well insured? Evidence on vulnerability to income risk in rural China. Journal of Development Economics, 58, 61–81. Kannan, K. P. (2006). Employment and social security for the working poor. Two major initiatives in India. New Delhi. Kaplow, L. (2006). Optimal income transfers. NBER Working Paper 12284. Cambridge MA: National Bureau of Economic Research. Lam, D., Leibbrandt, M., & Ranchhod, V. (2004). Labour Force Withdrawal of the Elderly in South Africa. Working Paper 118. Cape Town: SALDRU/CCSR. Levy, S. (2006). Progress Against Poverty. Sustaining Mexico’s ProgresaOportunidades Program. Washington DC: Brookings Institution Press. Lichand, G. (2010). Decomposing the Effects of CCTs on Entrepeneurship. Policy Research Working Paper 5457. Washington DC: The World Bank.

Martinez, S. (2007). Invertir el Bonosol para aliviar la pobreza: Retornos econo´micos en los hogares beneficiarios. In G. Aponte, L. C. Jemio, R. Laserna, S. Martinez, F. Molina, E. Schulze, & E. Skinner (Eds.), La Inversio´n Prudente. Impacto del Bonosol sobre la familia, la equidad social y el crecimiento econo´mico (pp. 109–128). La Paz: Fundacio´n Milenio. McCord, A. (2007). The social protection function of short-term public works programmes in the context of chronic poverty. In A. Barrientos, & D. Hulme (Eds.), Social Protection for the Poor and Poorest: Concepts, Policies and Politics. London: Palgrave. Moffitt, R. A. (2002). Welfare programs and labour supply. In A. J. Auerbach, & M. Feldstein (Eds.), Handbook of Public Economics, Vol. 4 (pp. 2394–2430). London: Elsevier Science B.V.. Morduch, J. (1995). Income Smoothing and Consumption Smoothing. Journal of Economic Perspectives, 9(3), 103–114. OECD (2003). Benefits and employment, friend or foe? Interactions between passive and active social programmes. Employment Outlook, 2003, 171–235. Posel, D., Fairburn, J. A., & Lund, F. (2004). Labour migration and households: A reconsideration of the effects of the social pension on labour supply in South Africa. Economic Modelling, 23, 836–853. Rabbani, M., Prakash, V. A., & Sulaiman, M. (2006). Impact Assessment of CFPR/TUP: A descriptive analysis based on 2002-2005 panel data. CFPR/TUF Working Paper 12. Dhaka: BRAC. Ravallion, M., & Wodon, Q. (2000). Does child labour displace schooling? Evidence on behavioural responses to enrollment subsidy. Economic Journal, 110(462), C158–C175. Sadoulet, E., de Janvry, A., & Davis, B. (2001). Cash transfer programs with income multipliers: PROCAMPO in Mexico. World Development, 29(6), 1043–1056. Saez, E. (2006). Redistribution towards Low Incomes in Richer Countries. In A. V. Banerjee, T. Be´nabou, & D. Mookherjee (Eds.), Understanding Poverty (pp. 187–202). Oxford: Oxford University Press. Schwarzer, H. (2000). Impactos socioeconoˆmicos do sistema de aposentadorias rurais no Brazil - Evideˆncias empı´ricas de un estudio de caso no estado de Para´. Discussion Paper 729. Rio de Janeiro: IPEA. Skoufias, E. (2005). Progresa and Its Impacts on the Welfare of Rural Households in Mexico. Washington: International Food Policy Research Institute. Skoufias, E., & di Maro, V. (2008). Conditional cash transfers, adult work incentives and current poverty. Journal of Development Studies, 44(7), 935–960. Skoufias, E., Unar, M., & Gonza´lez-Cossı´o, T. (2008). The impacts of cash and in-kind transfers on consumption and labour supply: Experimental evidence from rural Mexico. Policy Research Working Paper WPS 4778. Washington DC: The World Bank. Skoufias, E., & Parker, S. W. (2001). Conditional cash transfers and their impact on child work and schooling: Evidence from the PROGRESA program in Mexico. Economı´a, 2(1), 45–96. Stecklov, G., Winters, P., Stampini, M., & Davis, B. (2005). Do conditional cash transfers influence migration? A study using experimental data from the Mexican Progresa program. Demography, 42(4), 769–790. Todd, P. E., & Wolpin, K. I. (2006). Assessing the impact of a school subsidy program in Mexico: using a social experiment to validate a dynamic behavioural model of child schooling and fertility. American Economic Review, 96(5), 1384–1417. Udry, C. (1996). Gender, agricultural production, and the theory of the household. Journal of Political Economy, 101(5), 1010–1045. Yap, Y.-T., Sedlacek, G., & Orazem, P. F. (2002). Limiting child labor through behavior-based income transfers: An experimental evaluation of the PETI program in rural Brazil. Washington DC: Inter-American Development Bank, mimeo.