Education and health outcomes: the potential contribution of conditionality

Studies of poverty dynamics have amply documented the cumulative effects of disadvantage and the processes of intergenerational transmission of poverty. The analysis of childhood transition to adulthood underlines how the experience of early disadvantage weakens individual adaptation and in what ways the detrimental effect is then carried on into the future (Schoon, Reference Schoon2006). The long-term effects of malnutrition are explored by Alderman et al. (Reference Alderman2006), who show that improved pre-schooler nutritional status is associated with increased height as a young adult, a higher number of grades of schooling completed and an earlier age at which the child starts school. Others have stressed the cyclical nature of the processes of risk. Garmezy (Reference Garmezy1993), for example, traces the effects of family poverty on children via poor maternal health.

Such studies suggest that improvements in outcomes in education and health are crucial to enabling the poor to cope with adversity and to promote resilience. In terms of policy design, these findings reinforce the emphasis placed on early years interventions to improve education and health outcomes. These would target retardation in cognitive skills acquisition and improve the life chances of children and young people by addressing the risk factors undermining individual development (Schoon, Reference Schoon2006).

The introduction of a conditionality component that targets education and health in cash transfers is largely justified by most policymakers with precisely these objectives in mind: to improve human capital outcomes and promote processes of resilience. By design, CCTs in the Latin American countries reviewed here target households with young children or pregnant and nursing women with this objective. Perhaps the most widely cited definition of CCTs reflects this rationale: ‘The cash transfer is aimed at providing short-term assistance to families often in extreme poverty . . . while the conditionalities aim to promote longer term human capital investments, especially among the young’ (Rawlings, Reference Rawlings2004).

Yet debates regarding whether ‘to condition or not to condition’ have pointed to mixed understandings about how conditionality operates and in what ways it may influence policy impact on human capital. Conditionality has been justified to modify poor people's behaviour, to encourage decisions that will lead to increased ‘investments in human capital’. It has been criticised for potentially incurring high administrative costs, drawing resources that would be better spent directly on the poor. Others still, have argued that conditionality can help secure public support for a targeted intervention, leading to increased resources for the poor. The arguments in favour and against conditionality can be mapped to three types of considerations related to: behaviour, administration and political economy.

Behaviour

A widespread understanding of the purpose of conditionality is its potential impact on beneficiary behaviour. When (if ever) it is legitimate to condition eligibility for welfare benefits on the performance of prescribed activities will not be discussed in detail here.Footnote ² Rather, this section provides an overview of the principal justifications advanced in favour of conditionality's potential impact on behaviour with respect to perceived beneficial effects on outcomes of individuals or particular social groups.

Conditionality has been justified to address the distortion of choices that arise when myopia and information asymmetry lead individuals to make decisions that do not take into account the benefits of mid to long-term human capital accumulation. Lack of knowledge concerning the benefits of health-care check-ups, pre-natal treatment and other types of preventative health care for example might bring individuals to underutilise health services.

Secondly, conditionalities may be used to address intra-household bargaining processes and preferences to strengthen the bargaining position of individuals who have weak or no bargaining power within the household. Das et al. (Reference Das, Do and Ozler2004) illustrate this case in relation to child labour: where parents favour positive short-term outcomes such as increased income due to child labour over potential long-term returns from increased school attendance. In such cases, the justification of conditionality rests on its ability to address the differences between child and parent preferences.

Moreover, proponents of conditionality argue that it can promote individual behaviour that has a beneficial effect on the rest of society. When making decisions about their children's education, girls’ schooling for example, parents may not take into account the benefits that society derives from educating girls, and, as a result, decide to underinvest in girls’ schooling relative to levels that are beneficial from a societal perspective. Similarly, governments may be aware of the benefits of immunisation or screening for chronic diseases from a societal perspective, while individuals may be unaware or unconvinced of these benefits.

The above examples illustrate ways in which a change in beneficiary behaviour might be desirable and contribute to improved human capital outcomes. By requiring beneficiaries to send their children to school and undertake health visits, conditionalities in CCTs aim to influence behaviour in this direction. Yet two considerations arise when assessing the effects of conditionality on behaviour.

Firstly, as in any evaluative exercise, the central question is whether the inclusion of the conditionality in a targeted intervention induces additional changes in behaviour or human capital investment decisions to those that would have been achieved in the absence of the conditionality by the income transfer alone. The inclusion of conditionality in a CCT ‘presumes . . . that stimulating demand through income effects is insufficient to induce major changes in human capital investment . . . Instead, a condition that transforms the income effect into a price effect needs to be attached to the cash transfer’ (De Janvry and Sadoulet, Reference De Janvry and Sadoulet2006a).

Empirical evidence indicates that decisions on human capital investments do respond to pure income transfers, as revealed by numerous studies on unconditional cash transfers. Filho's (Reference Carvalho Filho2001) study for example, shows that an expansion of Brazil's non-contributory, unconditional old-age pension programme led to an increase in educational attainment among girls and to a decrease in child labour among boys.Footnote ³ Duflo (Reference Duflo2003) shows that the expansion of an old-age pension programme in South Africa led to an improvement in the health and nutrition of girls.Footnote ⁴ Aguero et al. (Reference Aguero, Carter and Woolard2006), studying the nutrition impacts of South Africa's unconditional Child Support Grant (CSG), find that large dosages of CSG treatment early in life significantly boost child height.Footnote ⁵

A second consideration in relation to behaviour arises from the possibility for conditionality to generate unintended behavioural responses. While evidence of these remains largely anecdotal, conditionalities introduce incentives and potential for distortions in behaviour that could be taken into account when assessing conditionality performance. In their account of the implementation of the Red de Protección Social (RPS) in Nicaragua for example, Maluccio and Flores (Reference Maluccio and Flores2005) describe the initial design and subsequent redesign of the RPS conditionality component on the basis of unintended behavioural consequences triggered by the implementation of the original conditionality. One of the requirements for benefit receipt over time was for beneficiary children at school to be promoted to the next grade. By design, in the original RPS, failure of successful grade promotion was a cause for expulsion from the programme. When policymakers learned that some schools practiced automatic promotion, enforcement of the grade promotion condition was not implemented.

Administration

Conditionality administration implies financial and social costs. Caldes et al. (2006) analyse the administrative costs of different CCT components for three Latin American programmes: Progresa (Mexico), Programa de Asignación Familiar (PRAF, Honduras) and RPS (Nicaragua). They find that the costs devoted to targeting and conditioning form a substantial part of the ongoing operations of these programmes.Footnote ⁶ The authors argue that the cost effectiveness of conditioning transfers will depend on the share of transfers reaching the poor and on the improvements in human capital that can be attributed to conditionality. As this paper's final section on CCT impacts indicates, while there is some evidence that the ‘payoff’ in terms of targeting performance has been high across CCT programmes, little is known about the effects of conditionality on human capital outcomes and on whether the administrative costs sustained for conditionality implementation are offset by positive outcomes.

Conditionalities act as additional screening or targeting devices. Especially in countries where administrative capacity is weak, logistic difficulties associated with implementing means testing and other targeting mechanisms can be at least partly offset by the additional requirements exercised through the conditionality. Work requirements in Argentina's Jefes y Jefas workfare programme for example, helped to assure self-selection into the programme by those that are unemployed (Galasso and Ravallion, Reference Galasso and Ravallion2003). In the same fashion, conditionalities requiring school-aged children to attend school and health clinics help ensure the cash transfer targeting of households with young children.

As with any targeting mechanism, a priority concern is that conditionalities can incur errors of exclusion and screen out some of the poor and most vulnerable. By overburdening the poor with behavioural requirements that they are not able to comply with, poor households might drop out of the programme. Poor households may find the conditions too difficult to meet, and if these households are among the poorest participants, imposing conditions may detract from the targeting of the CCT.

In a study on Mexico's Progresa, Alvarez et al. (2006) find that at an aggregate level, the cost of the program's conditionality does not seem to be overly burdensome for the extreme poor as they do not appear to systematically dropout of the programme. At the same time, they find two specific instances in which the conditionalities are contributing to the disproportionate abandonment of the programme by particular groups: the case of indigenous populations and the case of the extreme poor in communities where inequality is likely to be highest.

Furthermore, conditionality risks excluding poor households when CCTs are implemented only in areas where the infrastructure and administrative capacities are adequate for conditionality implementation. In Colombia for example, the Familias en Acción is implemented in poor municipalities that satisfy minimum infrastructure requirements. While this design option aims to ensure that beneficiaries are guaranteed circumstances in which they can formally comply with the conditionalities, it may have the effect of excluding the poor residing in excluded areas. Whether this is the case depends on alternative social policies covering areas and groups excluded from CCTs.

Finally, conditionality may aim to improve service provision. By stimulating the demand for services through modifying beneficiary behaviour, service provision and quality can be stepped up. Furthermore, conditionality implementation as part of the broader CCT initiative in many countries is accompanied by subsidies paid to service providers and the introduction of performance-based mechanisms that enhance the responsibilities of service providers.

Political economy

There is a common perception that conditionality helps legitimise a cash transfer politically. According to such claims, CCTs are more acceptable to policy-makers and to taxpayers than unconditional schemes. De Janvry and Sadoulet (Reference De Janvry and Sadoulet2006b) for example, observe that conditions can be used ‘to increase the political acceptability, and hence the budget size and sustainability, of the programme’. Conditionality would be desirable according to these authors, since ‘Larger budget size and greater sustainability due to higher political acceptability will benefit participants through higher allowances, or potential beneficiaries through a larger selected pool.’Footnote ⁷

Do conditionalities generate public support, making policies targeted to the poor more politically feasible? From a theoretical perspective, the political economy literature on targeting can serve as some reference, given that conditionalities function as an additional targeting device. According to this literature, more narrowly targeted policies risk losing support from the middle classes, leading to smaller budgets for the targeted programmes (Besley and Kanbur, Reference Besley and Kanbur1990). Conversely, political support for a programme may increase with the degree of universalism and leakage to the non poor (Gelbach and Pritchett, Reference Gelbach and Pritchett1997). If we regard conditionality to act as a form of additional targeting, it could be argued that conditionality risks jeopardizing support for a particular poverty targeted policy.

Studies on public attitudes on the other hand suggest that conditionality can have a legitimising effect. Most studies on public attitudes and policy examine how the former shape the political feasibility of broad policy choice. Graham (Reference Graham2002) shows how public attitudes – about redistribution, causes of poverty and opportunities for upward mobility – affect the political viability of social welfare policies in Latin America.Footnote ⁸ The analysis of the association between public attitudes and specific policy parameters in Latin America remains limited for the time being due to the limited available data. Studies linking social attitudes to design parameters however have been undertaken in industrialized countries, for which data are available. Taylor-Gooby's (Reference Taylor-Gooby, Pearce and Paxton2005) study using British Social Attitudes survey data shows that ‘people believe the welfare state should not be simply a funnel to direct resources downwards, but should also be a system of reciprocity, which provides good opportunities and support for those who participate, but does not waste resources on those who fail to do so’. The link then, between welfare payments and some form of valued social participation or reciprocity appears to act as a source of legitimacy for welfare spending.Footnote ⁹

Public attitudes vary from one country to another and by influencing the political feasibility of a welfare programme, shape the available budget. Taylor-Gooby's (Reference Taylor-Gooby, Pearce and Paxton2005) results suggest that in the UK for example, conditionality may indeed act as legitimising force. Discussions about conditionality and its effects would benefit from taking public attitudes and political feasibility concerns into account.

Conditionality design

In most Latin American countries, CCTs are entirely conditional. By design, only families with children and pregnant women are eligible to receive the cash transfer. Yet this is not always the case and in some countries cash transfers combine conditional and unconditional components. Brazil's Bolsa Família for example, includes an unconditional component paid to all the extreme poor, independently of household composition and demographic characteristics.

Conditionality is central to the functioning of CCTs when compliance is a requirement for benefit receipt from the outset. In Mexico's Oportunidades, conditionality compliance is monitored and verified before transfers are paid to the beneficiary. Similarly, in Chile, the Programa Puente's bono de proteccion payments are made once the family has been ‘activated’. Alternatively, in some CCTs, the monitoring of compliance starts in a subsequent phase to beneficiary selection and benefit payment.

Response to non-compliance also varies. When non-compliance translates directly into beneficiary suspension from the programme, conditionality can have a clear punitive effect. To avoid a purely sanctionary purpose of conditionality, CCT responses to non-compliance are designed so that an initial episode of non-compliance triggers a set of reactions in phases, of which definite suspension from the programme is the final one. Once a family is participating in Mexico's Oportunidades for example, non-compliance is initially followed by a suspension of benefit payments; if non-compliance persists, the benefit is suspended indefinitely and eventually definitely suspended. In Colombia, beneficiary families are taken off the Familias en Acción if they fail to comply with conditionalities for three consecutive months or if they fail to comply in four non-consecutive months within 12 months.

In Brazil, where non-compliance to Bolsa Familia conditionalities is understood as a flag of additional vulnerability, policymakers designed a response system that in the first instance ensures a verification of the reasons for non-compliance. Beneficiaries that do not comply with conditionality are in the first instance entitled to additional support and individualised services. Only if non-compliance persists do they initially receive a notification from the Ministry of Social Development. At subsequent stages, benefit payments are suspended, blocked and, in a final step, beneficiaries are removed from the programme. Beneficiaries that exit the Bolsa Familia as a result of conditionality non-compliance may re-apply six months thereafter.

In addition to emphasising beneficiary responsibilities in conditionality compliance, CCTs can pursue directly the objective of enhancing service provider and government responsibilities by including supply side components in their cash transfer design. Honduras’ and Nicaragua's CCTs both include cash transfers paid to service providers, health centres and schools, in the areas in which the conditionalities are set.Footnote ¹⁰ In Brazil, federal subsidies are paid to municipalities to support the additional administrative costs incurred by Bolsa Família implementation. Such subsidy amounts are determined by the achievement of a minimum score based on an administrative performance indicator, thus acting as an incentive for local administrations to improve programme administration and service provision (Bastagli, Reference Bastagli2007).

Conditionality implementation

Debates about conditionalities and their effects typically refer to the definition of conditionality as regulated by legislation and official documents, paying limited attention to conditionality implementation. In practice, conditionality implementation relies on a complex procedure in several stages that include: (a) informing beneficiaries and public officials of their exact definition, (b) the regular monitoring of compliance and (c) the implementation of responses to non-compliance.

Conditionality effects depend on what actors perceive the conditionality to be. Numerous reports on CCT implementation describe circumstances in which CCT beneficiaries are not aware that a cash transfer is conditional and in which both beneficiaries and policy administrators are confused about the exact conditionality definitions.Footnote ¹¹ Conditionality implementation then requires the dissemination among beneficiaries and policy administrators at all government levels, particularly those working directly with beneficiaries, of information on the definition of conditionalities and what compliance exactly entails.

Another phase of conditionality implementation, conditionality monitoring, requires the regular tracking of information on beneficiary compliance. This typically involves a variety of actors including beneficiaries, local service providers and public officials such as schoolteachers and health care workers, local CCT administrators and central-level administrators. The day-to-day compliance by beneficiaries is usually monitored at the local level and information is then passed to central levels of government, requiring several steps of information transmission and regular database management.Footnote ¹²

Once compliance monitoring is underway, if an episode of non-compliance is registered, CCT design foresees the activation of a response mechanism. By design, these can vary considerably, as outlined above. Their implementation in practice is another story. In the experience of most Latin American countries, it is difficult to obtain information on the number of CCT beneficiaries that have been penalized for non-compliance. Where information exists, it confirms the irregularity with which monitoring and responses to conditionality compliance are implemented. In Nicaragua, during the first year of RPS operation, about 10 per cent of beneficiaries were penalized at least once and therefore did not receive a full transfer. The RPS allowed households to receive a partial transfer if they complied with the health requirement and not the education requirement or vice versa (Maluccio, Reference Maluccio2005). In Brazil, where the Bolsa Família was launched in 2003, by 2007, out of a total beneficiary population of 11.2 million households, one million beneficiary households received a notification from the federal government for failure to comply to a conditionality and 11 households were dropped from the programme for non-compliance that same year (Bastagli, Reference Bastagli2007).

In summary, the design and implementation of conditionality, even among a single category of conditional policies, CCTs, vary substantially. Any assessment of conditionality impacts will benefit from the consideration of possible trajectories through which it impacts the well-being of the poor directly by design or implementation. Conditionality for example risks further marginalising the poor, through unintended behavioural responses, possibly linked to misunderstandings of conditionality definition and implementation, and the potential for responses to non-compliance to disproportionately punish particular vulnerable groups.