Cash Transfers, Children and the Crisis: Protecting Current and Future Investments

In a mix of responses to the food, fuel and financial crises of 2008–9, some developing countries have introduced new safety‐net programmes, while others have modified and/or expanded existing ones. Many have introduced conditional cash transfers (CCTs) in recent years, and these have been used as an important starting point for a response. This article aims to describe these various experiences with CCTs, to distil lessons about their effectiveness as crisis‐response programmes for households with children, to identify design features that can facilitate their ability to respond to transient poverty shocks, and to assess how they can complement other safety‐net programmes.

2. Barrientos and DeJong (2006) find that different types of cash-transfer programmes, including incometargeted and conditional programmes, are effective at reducing child poverty. 3. It also builds on two recent studies of social safety nets and CCTs worldwide, Grosh et al. (2009) and Fiszbein et al. (2009). 4. For example, the Republic of Yemen extended its cash-for-work programme to an additional 22,000-26,000 households to offset higher food prices. Similarly, Ethiopia expanded its public works programme to 1.4 million households in response to the food and fuel crises (World Bank, 2010). Argentina provides another example: the emergency public works programme Jefes y Jefas reached 2 million beneficiaries in 2003 after that country's financial crises. For more on school feeding programmes see Bundy et al. (2009). 5. There is evidence that transfers have intra-household effects on children. For example, a study of old-age pensions in South Africa found that girls whose grandmothers received pensions have improvements in height and weight (Duflo, 2003).

Cash transfers and crises
Safety-net programmes encompass a wide range of cash and in-kind transfer programmes and services. 6 This article focuses on national-level cash-transfer programmes targeted on households with children. This subset of programmes includes both conditional cash-transfer programmes (CCTs), which provide money to poor families contingent on their making investments in human capital, such as keeping children in school or taking them to health check-ups, and unconditional cash-transfer programmes (UCTs), which provide direct payments without conditions. Both CCTs and UCTs can be targeted on poor and vulnerable households based on criteria such as income, geographic location and other household characteristics.
As crisis-response measures, cash transfers have notable advantages. They can be an important vehicle for injecting purchasing power into the economy and transferring resources rapidly to poor households. They can also be important instruments for mitigating transient poverty arising from shocks. On the other hand, drawbacks exist. If administrative capacity is low, it may not be possible or straightforward to scale-up programmes rapidly. Personnel and infrastructure, including offices, information and payment systems, may simply not exist nor be possible to install easily to cope with a rapid increase in the number of beneficiaries. In this regard, programme design -in addition to administrative capacity -may also affect how quickly programmes can be adapted. For example, those with relatively simple targeting mechanisms will be easier to scale-up than others. A programme that is self-targeted may be less personnelintensive and easier to scale-up than a benefit which relies on a targeting methodology requiring home visits by programme staff to assess eligibility.
CCTs pose additional considerations in a crisis context. They have the same benefit of providing poor households with income to mitigate shocks, but because they support investments in human capital, they are primarily instruments to address longterm, structural poverty rather than sudden income shocks, particularly if those shocks are expected to be short-term ones. A number of the properties of CCT programmes are inconsistent with the type of flexibility required to manage risk during a crisis. First, CCTs are not designed as crisis-response measures. The administrative targeting methods they use are such that it is difficult to add new beneficiaries in the short term and to remove them from the roster when a crisis has passed. Second, households that are already receiving transfers from a CCT programme may not be those which are worst hit by an aggregate economic shock. Third, CCTs require households to make 'lumpy' investments in child education and health -investments that only make sense with a longer-term horizon. And finally, it is not clear that households will always disinvest in child human capital during systemic shocks. A study of the impact of food price rises in Bangladesh during the 2008 crisis found that mothers protected the nutritional status of their children, even when total household food intake was cut (Sulaiman et al., 2009).
In addition to concerns about the appropriateness of CCTs as crisis-response measures, there are also debates about the effectiveness of conditions. For example, a recent study of the impact of a CCT in Malawi on girls' enrolment in school found no impact of the condition itself (Baird et al., 2010). 7 For these reasons, CCTs may not be ideal instruments for dealing with transient poverty. Transfer programmes that do not involve long-term commitments (such as those implicit in CCT conditions), that are self-targeted (and thus do not involve complex administrative decisions for programme entry or exit), and that involve the participation of beneficiaries in activities that can help address the source of the shock (for example, job-related activities) appear to be better suited to crisis contexts, in some respects, than CCTs. Workfare programmes may be preferable in some crisis contexts, although these have drawbacks as well, including limited coverage of vulnerable groups which are unable to work. Workfare programmes also require administrative capacity to target beneficiaries, and to identify, allocate and monitor the public works projects. 8 Nevertheless, as the experience of the 2008-9 crisis illustrates, having a cash-transfer programme in place -conditional or unconditional -clearly is better when there is a crisis than not having any large-scale social assistance programme at all.

The effects of crises
A starting point for assessing the choice of safety-net instrument for a particular context is to consider the types of effects that economic shocks can have on human capital. While it is too early for a full assessment of the effects of the 2008-9 downturn on human-development outcomes, it is illustrative to consider the past experience of shocks. Evidence suggests that, in practice, shocks are likely to have very different effects on different dimensions of human capital, and the impact will depend critically on whether shocks are idiosyncratic or systemic (Ferreira and Schady, 2008). In many developing countries, child health and nutritional status deteriorate during crises. 9 On the other hand, the evidence on the effects on school attendance is mixed. For example, Jensen (2000) finds that negative weather shocks have large negative effects on school enrolment in Côte d'Ivoire; and Thomas et al. (2004) find that the Indonesian financial crisis of 1998 had a negative effect on school enrolment, although its size is very small. On the other hand, schooling may increase during downturns if the decrease in the opportunity cost of going to school is large enough to offset the negative income effect for credit-constrained households. 10 The social impact of the crisis has the potential to persist after the crisis itself. The experience of previous downturns illustrates that poverty can endure beyond macroeconomic recovery, and responds with a lag. For example, the national poverty level in Mexico only returned to the pre-crisis level approximately ten years after the 7. It should be noted, however, that the target population for this evaluation was older girls and the transfer amount was large. So the income effect may have overpowered the effect of the condition. 8. There is also plenty of empirical evidence on how workfare programmes operate (for example, see Drèze and Sen, 1991;Ravallion et al., 1993;Ravallion, 1994, 1995;Jalan and Ravallion, 2003). 9. See, for example, Cutler et al. 2002;Paxson and Schady 2005;Baird et al., 2007. 10. This appears to have been the case in Nicaragua; it was also observed in Peru during the deep recession of the late 1980s (Schady 2004), in Mexico in the 1990s (Mckenzie 2003), and in the United States during the Great Depression of the 1930s (Goldin 1999).
1994-5 peso crisis. Similarly, losses in non-income dimensions of welfare, including labour markets, educational attainment and health status, may have lasting effects. Diverse effects of crises may be due to heterogeneous responses by households. Households may apply alternative strategies to cope with the effects of shocks, potentially including increasing their working hours, drawing down savings and other assets, or turning to informal social networks. If these mechanisms are insufficient, households may turn to purchasing cheaper but less nutritious food, selling productive assets such as land and livestock, withdrawing children from school or delaying their enrolments and encouraging them to enter the labour market, and deferring the use of essential health services. While these risk-management mechanisms may aid households over the short term, they may also imply long-term losses in productivity and child welfare, in particular among the poorest households, increasing the likelihood of intergenerational transmission of poverty, and underscoring the importance of investments in child health, nutrition and schooling for long-term development.
Even when chronically poor households are not the hardest hit by shocks, their economic conditions can be expected to worsen from already low levels, causing severe consequences for their well-being. They are likely to become poorer as a result of the crisis and may be most at risk of suffering irreversible losses because they have fewer strategies to mitigate the crisis, such as seeking credit, increasing work or private transfers. In the absence of an effective social protection system, they can be compelled to use inefficient private responses, such as reducing investments in human and physical capital that affect their long-term earning potential. The possible impacts of a crisis on children are of particular concern. In response to economic crises, households generally reduce their consumption levels and change consumption patterns, and in some cases this can lead to an increase in child malnutrition. All of these adjustments are difficult to reverse in the short term, causing negative impacts on welfare to persist even after the economy recovers and generating long-term losses in poverty reduction and human development.
In this context, cash transfers may help cushion the impact of systemic and idiosyncratic shocks, including possible effects that these may have on children's accumulation of human capital. The findings from prior crises suggest that there is an important role for cash transfers to provide income support to both households affected by transient shocks and the chronically poor, and to maintain household investments in human capital. The caveats noted earlier suggest that cash transfers -both conditional and unconditional -should be elements of a comprehensive social protection strategy, and can complement other measures such as workfare and unemployment benefits.

Box 1: Do CCTs help protect human-capital investments during economic shocks?
Evidence from Nicaragua, Mexico and Indonesia illustrates that, despite constraints, CCTs can contribute to mitigating shocks and complement other safety-net responses. Maluccio (2005) considers patterns of occupational choice, employment, consumption, school enrolment and child nutritional status in the 2000-2 period in Nicaragua. The data cover communities randomly assigned to the CCT programme, the Red de Protección Social (RPS), treatment and control groups, and households in coffee-growing and non-coffee-growing areas, which are found in both groups. The 2000-2 period saw a sharp downturn in the price of coffee. Maluccio shows that in those communities that did not receive RPS transfers during this period, household per capita expenditures fell by 18%. Nevertheless, school enrolment of children aged 7-12 increased -particularly in coffee-growing areas. Among boys, for example, school enrolment increased by 15 percentage points, which suggests that the opportunity cost of going to school fell sharply. Turning next to a comparison of changes in enrolment in RPS treatment and control communities, Maluccio shows that increases in enrolment were larger in RPS communities than in the control communities, and even larger in RPS communities that were also in coffee-growing areas. But it would not be accurate to conclude that the RPS 'protected' school enrolment during a downturn because enrolment increased during the period in control communities, especially in coffee-growing areas. Maluccio next analyses changes in nutritional status over the period. He shows that height for age deteriorated in control communities between 2000 and 2002, but not in RPS communities. However, the positive impact of the RPS on child nutritional status was larger in noncoffee-growing areas than in areas where coffee is grown -a finding that suggests, if anything, that the RPS was better able to improve child nutritional status in areas in which household incomes were stable than in those affected by the economic downturn.
In Mexico, de Janvry et al. (2006) combine the randomised assignment of the flagship CCT programme, Oportunidades, with data on systemic shocks (drought, natural disaster) and idiosyncratic shocks (unemployment or illness of the household head, illness of pre-school-age children) to compare household responses to shocks in treatment and control villages. They show that shocks generally reduced school enrolment in the sample, but that those effects were offset partially or fully by Oportunidades (in particular, with systemic shocks such as droughts and other natural disasters). During the Indonesian crisis of 1997-8, the government made children in poor households eligible for a 'scholarship' programme. Although, given the crisis context, little attention was paid to evaluation of the effect of the programme, using regression and matching techniques, Cameron (2002) concludes that the programme reduced drop-out levels in lower-secondary school by about 3 percentage points. Sparrow (2007) runs ordinary least squares regressions that suggest a larger effect on enrolment for children aged 10-12 (about 7.6 percentage points).

Safety-net responses
While it is too early to assess the outcomes of policy responses to the crisis, an overview of selected examples across regions suggests that countries that had existing cash-transfer programmes in place, or that had plans for reforms or new programmes in the works, were able to respond more quickly than others. Across regions, countries expanded coverage of existing programmes, increased benefit levels, or introduced new programmes. Table 1 summarises some of the programmes that were used in response to the crisis by type of response. In some cases, such as the Philippines and Pakistan, countries had pilot cash-transfer programmes under way and were able to move rapidly to implement new programmes. This section looks across the experience of a selection of countries which made use of cash-transfer programmes during the 2008-9 crises.
At the outset it is important to note that the recent crises had varying effects across countries, depending on their level of exposure to the external shocks, their fiscal position, and their ability to respond with safety-net measures and other countercyclical programmes. The extent of the downturn in countries with cash-transfer responses ranged from declines of 14.9 percentage points in GDP in Latvia between 2007 and 2008, to 3.2 percentage points in the Philippines and 1.9 points in Mexico.
While many countries have begun to recover from the shocks of the crises, the longer-term impact of increased unemployment and poverty is likely to persist. Recent World Bank estimates suggest that, while poverty rates have been falling, by 2015 the global poverty rate is projected to be 15%, in contrast with a projected 14.1% if the crisis had not occurred (World Bank, 2010). The existence of social programmes -including cash transfers -has been an important factor in mitigating the effects of the crisis.

Expanding coverage
The most common approach for countries with existing programmes has been to expand their coverage. In some cases, such as Brazil's Bolsa Familia and Mexico's Oportunidades programme, this involved adding beneficiaries to CCT programmes which already had substantial coverage. In other cases, such as the Philippines and Pakistan, this has involved scaling-up of pilot programmes. In Kenya and Malawi plans were under way to scale-up pilot transfer programmes before the crisis, and they were accelerated. Scaling-up has been achieved in different ways. Some countries increased eligibility by adding new groups to the programme while others expanded coverage geographically.
In other cases, countries revised formulas that target beneficiaries. This happened in Latvia which raised the eligibility threshold for its Guaranteed Minimum Income (GMI) Programme. Another approach was to change the targeting methodology completely. For example, the Philippines shifted from geographic targeting to proxy means-testing of the Pantawid Pamilyang Pilipino Programme (4P), and Pakistan moved from targeting the Benazir Income Support Programme (BISP) by parliamentarians to proxy means-testing. Brazil's strategy involved a combined approach of raising the benefit eligibility threshold, as well as revising the targeting formula to take account of variations in household income.
Countries have accompanied changes to targeting methodologies with efforts to strengthen the administrative infrastructure of programmes; for example, by investing in improvements in programme management as well as strengthening information and payment systems. In the Philippines, the scaling-up of the 4P programme has involved a major investment in creating a unified registry of beneficiaries. These efforts have the benefit of increasing the capacity of programmes to respond during the crisis, as well as contributing to longer-term improvements in efficiency and effectiveness.
Expansion of programmes has continued in the aftermath of the 2008-9 crisis. Guatemala is expanding its newly introduced MIFAPRO programme from covering 89 to 145 municipalities; Mexico is introducing a new benefit for families with children below school age; and El Salvador is introducing a new transfer programme in rural areas, Comunidades Solidarias Rurales, based on a former CCT programme.

Increasing benefits
Another frequent reform during the crisis has been to increase transfer amounts to enable households to tackle the crisis. This has been an important development, as in many cases, even where households were eligible for transfers, the payments were sometimes too low to provide poverty relief. For example, prior to recent changes, the average benefit of the Guaranteed Minimum Income Programme in Latvia was 6. In some cases, such as Latvia, these increases were considered permanent adjustments, while in the case of the Kyrgyz Republic, the programme was planned as a temporary increase for 10 months. Similarly, in 2009 the Chilean government paid two one-time benefits -one in April and the other in August of 40,000 pesos per dependant. Those eligible included participants in the CCT programme Chile Solidario, any family receiving the Single Family Allowance (social assistance) and those formal workers receiving the contributory family allowance but with wages below a certain ceiling. The aim was to cover the bottom 40% of families or 1.7 million individuals, with the expectation that it would cover 3.5 million.
Other Latin American countries, including Mexico, increased the level of payments to CCT beneficiaries. These additional payments were presented as supplemental payments made on an exceptional basis in order to give policy-makers room to scale-down payments to their pre-crisis levels in the future (Walker et al., 2010). Benefit increases are another example of crisis responses potentially leading to longer-term improvements in the programmes' abilities to alleviate poverty. However, the increased budget commitments can also undermine the fiscal sustainability of the programme and raise expectations for beneficiaries that the increases will be permanent.

New programmes
The crisis has also encouraged the introduction of new cash-transfer programmes. In some cases, such as the BISP in Pakistan, this involved the scaling-up of a pilot programme. The BISP was introduced in the second half of 2008 on a pilot basis. The government is in the process of expanding the programme to cover 1.6 million families. Senegal also introduced a one-off cash transfer with the goal of reducing food insecurity in children below 5 years of age. The programme paid out US$14 per month for 6 months during the crisis in 2009. Indonesia reinstated its Bantuan Langsung Tunai (BLT) programme during the crisis. It existed in 2005 and was started again in 2008 as a one-off cash-transfer programme to offset increases in energy prices. The transfer was made to approximately 18.5 million households, 30% of the total. Finally, the MIFAPRO programme in Guatemala was also introduced in 2008 as a social protection programme covering 89 municipalities.

Reflections and early lessons
Although it is premature to assess the impact of the recent wave of reforms on welfare outcomes, it is evident that many countries have made substantial use of cash-transfer programmes -especially existing programmes during the 2008-9 crisis -and that countries that had programmes in place were better positioned to respond during the crisis than those that did not have existing programmes. The scale of responses to the recent shocks varied. While some countries have put in place massive new programmes, or made unprecedented increases to existing programmes, others have made more marginal adjustments, either to improve programme coverage, by adding previously excluded beneficiaries, or to enhance the adequacy of benefits by increasing benefit amounts. Regardless of the scale, many of these reforms have the potential to strengthen safety nets over the longer term and leave countries better prepared for the next crisis. This underscores the importance of investing in programmes now that can be scaled-up in the future. However, the policy changes also raise potential risks that need to be considered, including for fiscal sustainability.

Seizing crisis as an opportunity
Countries have taken advantage of windows of opportunity presented by the recent crisis to expand much needed cash assistance to the poor. The crisis accelerated the scaling-up of pilot programmes as well as pending or planned reforms of established programmes. The fact that fewer countries introduced new cash transfers, or other typical crisis-response programmes, such as public works, suggests that it is easier and faster to top-up or expand existing programmes, rather than creating a new programme.
A striking example of a country which scaled-up a pilot programme in a massive way is the Philippines government's rapid scale-up of the 4P CCT programme. In response to the strains from increasing domestic food prices, growing unemployment and declining remittances during the crisis, the government included scaling-up of the 4P programme as part of its stimulus package. The programme increased its coverage to 376,000 households in 2008 and expanded further at the beginning of 2009 to cover 1 million household beneficiaries, or approximately 20% of the poor. The introduction of the 4P programme was a part of a set of reforms in social protection which had begun in 2007. However, the food and fuel crises of 2008 accelerated the government's reform agenda. In addition to the 4P, the government also introduced other temporary cashtransfer programmes including the Tulong Para Kay Lolo and Lola cash transfer for citizens aged over 70 without a regular income, and the Pantawid Kuryente, a one-time cash transfer for low-income energy consumers, that was introduced at the onset of the crisis.
The government made important changes to the 4P programme to expand its coverage rapidly, including shifting from geographic targeting to a proxy means-test. A new targeting system was installed, including a standardised database of poor households, which strengthened the process of beneficiary identification and improved the programme's efficiency. The government also expanded the budget for monitoring and evaluation of the 4P. In addition to direct programme monitoring, there are plans to collect more rigorous household-level data to assess the impact of the programme on households.
Countries with well-established programmes also used the crisis as an opportunity to undertake reforms. In the wake of the 2008 crisis, Brazil implemented planned adjustments to the targeting and coverage of its flagship CCT programme, the Bolsa Familia.
12 These changes were based on the results of a 2008 study which found high 12. This CCT programme was created in October 2003 through the consolidation of four existing cash-transfer programmes. The benefit is conditioned on compliance with health and education requirements, and is available in all 5,563 municipalities in the country, helping around 45 million poor people (approximately 25% of the population). For more information see 'Bolsa Familia Programme: Targeting Strategies', income volatility among families which fit the income profile for the BF programme. Following this, the government decided to review the estimates of the target population based on census data, household data and an index of income volatility. Based on this analysis, it decided to increase the size of the target population for the programme -an adjustment which brought in 1.3 million additional families by October 2009. A new method was also adopted to estimate poverty at the municipal level using poverty maps. This methodology allows the programme to determine more accurately the number of families that would be eligible for the programme across municipalities, and has led to increased participation particularly from urban residents. The Bolsa Familia plans further expansion to reach all poor families in all municipalities of Brazil. This will be facilitated through improvements to implementation, including streamlining programme management and payment systems, especially in rural areas, and increasing the capacities of local municipalities to verify compliance with programme requirements. The other planned area for improvement is the ability to link beneficiaries of the BFP to other interventions such as incomegenerating opportunities operated by federal or municipal governments (see World Bank, 2009).
Similarly, Mexico made changes to its flagship CCT programme, Oportunidades. The Social Development Secretariat (SEDESOL), which oversees the programme, adjusted benefits based on price inflation. In 2008, an additional transfer of approximately US$4 on top of inflationary adjustments was introduced to respond to rising energy prices. An additional US$10 was provided to each household in the programme in 2009 due to rising food prices. The operating rules of the programme were also modified by fixing the minimum number of households in the programme rather than setting a ceiling on the number of households participating. In 2010, a new benefit was added for families with children below school age. Families were given an extra US$7 per child in the household up to a limit of 3 children.
The government planned to expand the programme in 2010 to 5.8 million families from the current level of 5.2 million, to take place primarily in urban areas where the under-coverage is greater. There are also plans to expand another programme that covers areas lacking enough service capacity to handle all families eligible for Oportunidades. This is also a cash-transfer programme, but with fewer and no educational conditions. It is expected to expand to 600,000 families from its current level of 200,000 families.

Adjusting programme design and managing risks
While crisis brings opportunity, the rapid momentum of reforms during a crisis also raises risks that changes made to programmes under the pressures of a crisis can weaken a well-designed instrument for poverty reduction. Such risks include reducing benefit levels because of fiscal cuts, expanding coverage beyond the structurally poor, or undermining the sustainability of programmes -if benefit increases made during the crisis are difficult to scale-back. Some countries introduced targeted benefit increases without signalling whether these changes were temporary or permanent. This raises the risk that it will be politically contentious to repeal changes, by cutting off beneficiaries or reducing payment amounts, once the immediate impact of the crisis has passed.
In addition, rapid scaling-up of cash transfers raises risks of increased error, fraud and corruption in the delivery of cash-transfer programmes. 13 Even if adequate arrangements for monitoring and control of the delivery of the programme are fully specified, and even running effectively in pilot regions, it may be difficult to replicate and implement these systems in a wider geographic area. Cash transfers pose special risks from the perspective of error, fraud and corruption because of the large number of financial transactions they require and the involvement of multiple actors in their delivery -from household beneficiaries to programme managers -frequently at municipal, regional, and national levels (Stolk and Tesliuc, 2010).
In order to mitigate these risks, the Philippines 4P programme took preventive measures, with support from the Australian government (AUSAID), to identify and reduce governance risks as part of the plans for scaling-up the programme. These included activities that supported the integration of risk-mitigating measures within core programme design, such as a transparent proxy means-test-based targeting system, a management information system, monitoring and spot checks, and process risk mapping; demand-side measures which supported activities to enhance the capacity of beneficiaries and the general public to oversee the programme's implementation through public information and grievance redress systems; and supply-side measures which supported the strengthening of the implementing agency's governance and anti-corruption capacity through a focused Integrity Development Review (Arulpragasam et al., forthcoming).

Protecting human-capital investments with conditionality
The crisis has also raised the question of whether conditional or unconditional cash transfers are preferable. If the focus is on protecting the structurally poor, conditions play the same role in stimulating demand for investments in human capital as they do during non-crisis times. On the other hand, if the focus is on protecting the transient poor from shocks, conditions may have a limited effect, or may create logistical barriers to rapid assistance. From the perspective of protecting children, the urgent priority during a crisis is ensuring the usage and quality of education and health services. Where countries are able to maintain pro-cyclical expenditures in social services, investing in a CCT programme might lead them to protect or prioritise social spending; however, there is no guarantee that this will always be the case, given severe fiscal pressures. As an example, scaling-up of the 4P CCT programme in the Philippines led to increased efforts to invest in the supply of services. This has led to a more co-ordinated response, as the government has had to develop a rapid plan for the expansion of services to underserved areas.
In Kenya, the government accelerated the scaling-up of a new cash-transfer programme during the crisis -the Orphan and Vulnerable Children cash-transfer programme (OVC-CT). However, although it is intended to be a CCT, because of constraints on the supply side, the conditions are not monitored or used as requirements for receiving benefits. Instead, they are intended to encourage households to meet the government's health and educational targets. Strict monitoring of conditions would have weakened the ability of the programme to reach vulnerable children.
The government plans to make the OVC-CT programme a permanent feature of the social safety-net programme in Kenya, and is working to improve the coverage of education and health services to allow for the monitoring of conditions. It aims to cover 100,000 poor households with OVCs by 2012, which would enlarge the total programme to cover approximately 300,000 extremely poor OVCs and 13% of all OVCs in the country. In addition, in 2009 the government was scaling-up another cashtransfer programme, the Vulnerable People's Programme (VPP), based on lessons learned from pilot programmes and implemented by Oxfam and the World Food Programme (WFP) in Nairobi.

Strengthening the overall social protection system
The overriding lesson from current and previous crises is the importance for countries of having a well-functioning system of safety nets in place before a crisis hits, so that mechanisms are in place to offset shocks in bad times, as well as to promote growth and development in good times. If social safety nets are not in place, governments are often pressured to respond with sub-optimal policies (for example, consumer subsidies) which can leave a legacy of lower growth. Investments to scale-up and improve the effectiveness of cash-transfer programmes have the potential to strengthen overall social protection systems to protect the poor and to leave countries better prepared for subsequent downturns. Cash-transfer programmes have a role to play in the context of a 'permanent safety net' that is integral to a country's poverty-reduction strategy Ravallion, 2008).
In some countries, installing cash-transfer programmes was the priority, but in others, where cash transfers were already significant, other policy reforms -addressing needs such as severe unemployment -were the priority. For example, countries in Central and South-east Europe deployed multi-pronged strategies to strengthen income support for the unemployed and poor. Romania extended the duration of unemployment benefits. Similarly, Bulgaria increased the amount of unemployment benefits, strengthened labour-market measures including wage subsidies and training, and introduced measures to improve the effectiveness of its social assistance programme. In Latvia, the sharp increase in unemployment led the government to increase the duration of unemployment benefits to a maximum of 9 months. However, since only 50% of registered unemployed people are covered by unemployment insurance benefits (because of work-tenure requirements), the government also launched a public works programme for those not covered. This programme pays a stipend of about 80% of the net minimum wage and requires participants to engage in labour-intensive work identified by municipalities. In general, a systemic approach is needed which complements cash transfers with other safety-net programmes. Even where the focus is on supporting children, public works programmes and benefits for the unemployed are important for supporting household welfare.
Cash transfers, and particularly CCTs, contain administrative prerequisites which provide countries with more capacity to identify and reach the poor and vulnerable groups (Fiszbein and Schady, 2009). These externalities can contribute to better preparedness during a crisis. Countries with existing programmes are more likely to have targeting mechanisms in place to reach the poor, and information systems including beneficiary registries, management information systems, monitoring and evaluation instruments such as household surveys, as well as trained staff. They are able to use the targeting mechanisms from cash-transfer programmes to direct other social programmes.
For example, by benefiting from the existence of well developed targeting instruments, Chile was able to provide two one-off benefit payments to beneficiaries of Chile Solidario as well as other programmes in 2009. Other countries in Latin America use the targeting instruments for their CCT programme to target other benefits, including health insurance in Colombia and labour-market programmes in Brazil. Further analysis of the experience of specific countries and programmes is needed to draw lessons about how particular design features, such as the choice of targeting mechanisms, or the design of payment and information systems can be most conducive to rapid scaling-up.
In addition to administrative capacities, there may be greater political support for cash-transfer programmes where they already have a demonstrated track record and have been accepted by the population. The 2008-9 crisis has increased pressure on countries to improve the administration of cash-transfer programmes by reviewing and improving their targeting methods, investing in programme infrastructure such as beneficiary registries, and identifying opportunities for improving programme management and efficiency.

Focusing donor assistance
Finally, social safety nets require counter-cyclical investments. During the crisis middle-income countries which had pursued prudent fiscal policies in times of economic expansion -for example, Chile -had more financing capacity to make these investments and incorporate safety-net spending into national stimulus packages. Chile was able to finance its stimulus package with accumulated reserves from copper revenues in an Economic and Social Stabilisation Fund. This helped finance a countercyclical package of public investment, tax cuts for private investments and expanded social safety-net benefits for the poor (World Bank, 2010;Velasco, 2009).
Low-income countries -where the proportion of the poor is the largest -were more likely to lack resources for safety nets and needed to rely on donor assistance. However, during the 2008-9 crisis, safety-net programmes, and particularly cashtransfer programmes, have been an important focus and channel for donor support across countries. Even in middle-income countries such as Mexico, Argentina and the Philippines, governments accessed international aid for financing benefits, as well as technical assistance for strengthening programmes. During fiscal year 2009, World Bank lending for safety nets reached a total of over $3 billion in 29 countries, seven times the annual average. 14 The largest share of this lending was in Latin America, which had the greatest number of existing programmes that could be scaled-up. Agencies including the UK Department for International Development, the UN Children's Emergency Fund, the World Food Programme and others, have also been actively engaged in work on safety nets across countries. Again, countries that had existing programmes, or plans in place, were able to facilitate donor efforts.

Conclusion
The experience from past crises and the current downturn confirms that cash-transfer programmes can be important instruments for providing a rapid injection of income support to vulnerable households with children. Early indications suggest that countries were able to weather the immediate effects of the 2008-9 crisis more effectively than past crises because of the greater prevalence of safety-net programmes. During the recent crisis, cash-transfer programmes have become more central to some countries' efforts to provide income support to families with children. Reforms undertaken during the crisis have contributed to strengthening the capacity of these programmes to provide income support to the poor over the longer term. The discussion in this article leads to two main conclusions. First, investments in cash-transfer systems can pay off, and countries can make productive investments now in systems that will leave them better prepared for the next crisis. Second, cash transfers are only part of the solution. More needs to be done to ensure an effective policy mix that protects multiple dimensions of child welfare, such as nutritional and educational status, from the effects of shocks. This will involve more rigorous analysis of the effectiveness of cash transfers and other measures. As household data become available, it will be critical to analyse the distributional impact of these policy changes, assess the implications for the long-term sustainability of safety nets, and identify lessons for the future.
first submitted June 2010 final revision accepted December 2010