Helping South Asia Cope with Natural Disasters: The Role of Social Protection

This discussion paper, an input to the South Asia region's Social Protection and Hazard Risk Management Strategies, describes the cash transfer instruments supported by the Bank in South Asia, evaluates their design and implementation, and suggests improvements to increase their effectiveness. Based on available evidence, the paper finds that cash transfers appear to have performed well in providing relief to affected households, suggesting that they should remain an integral part of Bank-financed support for natural disasters. The paper also suggests that the Bank can ensure timely and high-quality support through a best-practice design toolkit, a right-on-time technical assistance facility, and by integrating social protection in emergency preparedness by building the capacity of national social assistance (cash transfers) agencies to respond to natural disasters. Although the focus is on cash transfers, the note also discusses other types of social protection mechanisms used in emergencies in South Asia and worldwide, e.g., workfare or social care for the vulnerable, and which might also appropriate for including in Bank emergency operations. The note covers South Asia, but lessons from this region may also be relevant for governments of other developing countries and donors.

Discussion Papers are published to communicate the results of The World Bank's work to the development community with the least possible delay. The typescript manuscript of this paper therefore has not been prepared in accordance with the procedures appropriate to formally edited texts. Some sources cited in the paper may be informal documents that are not readily available.
The findings, interpretations, and conclusions expressed herein do not necessarily reflect the views of the International Bank for Reconstruction and Development / The World Bank and its affiliated organizations, or those of the Executive Directors of The World Bank or the governments they represent.
The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries.
1 Social protection comprises safety nets and social security programs. Cash transfers are one component of the safety net programs. Income support or social assistance are also terms used for cash transfers, and these terms are used interchangeably in the text. 2 The paper discusses cash transfers in case of emergencies, cash transfers have been successfully used to protect laid off workers during major economic restructuring, economic crises, and to address the needs of post-conflict populations. INTRODUCTION South Asia is prone to natural disasters. In earlier times, these disasters often caused massive famines that prompted government action. A stone inscription found in Bangladesh dating from the third century BC contains an instruction from a higher government official to subordinate officers to supply paddy to distressed areas. The Great Bengal Famine of 1943 was precipitated by crop failures beginning in 1938 and events accompanying the Second World War and had an estimated death toll of 3.5 million. Thus, disasters have been a recurring feature in South Asia.
More recently, between 1990 and 2007, some 682 natural disasters were registered in the region, killing an estimated almost 400,000 people, affecting many more, and causing severe economic damages (See Annex 1). India, Bangladesh, and Afghanistan were the most affected countries in South Asia in terms of number of events, while Bangladesh, India, and Pakistan suffered most fatalities. The most common types of natural disasters were floods, wind storms, and epidemics, while the natural disasters that caused the most deaths were wind storms, earthquakes, wave/surge (including tsunami), and floods. Most disasters are small and localized; at the most 2 percent of (registered) disasters in this period caused more than 1,000 deaths. The disaster types with the average deaths per event were wave/surge, earthquake, and windstorm. Thus, the casualty toll from natural disasters in South Asia is dominated by rapid onset events unlike in the past, where droughts and other slow onset events wrought havoc. 3 South Asian countries have coped relatively well with recent major disasters. Severe disasters such as the tsunami, or the Pakistan earthquake did not cause massive famine or post-disaster mortality. The population affected by these severe disasters was provided food and shelter in the immediate relief phase; social protection (cash support) in the medium-term transition from relief to reconstruction; and housing and infrastructure rehabilitation in the reconstruction phase. Field evidence suggests that early relief plays a crucial role in helping affected communities cope and recover and is a prominent indication to those affected of the government's commitment to rebuild.
The Bank's assistance has played an important role in helping South Asian countries cope with disasters. South Asia has a major disaster-related lending portfolio of more than $10 billion since 1980. Two of the top ten countries on the Bank's global top-10 list of disaster financing are South Asian: India and Bangladesh. Between 1984 and 2005, a total of 71 disaster projects were implemented in these two countries. The Bank's support in South Asia reflects its broader emphasis on helping countries cope with emergencies worldwide. Since 1984, the Bank has financed natural disaster activities in 528 projects (including 89 ERLs) across all its regions totaling $26 billion and constituting some 10 percent of total global Bank lending over the last two decades.
Most recently, the Bank has redoubled its efforts to improve its capacity for responding quickly and in a poverty-sensitive way to emergencies, such as natural disasters. In response to recommendations of an IEG evaluation report, the Bank has issued a new Policy on Rapid Response to Crises and Emergencies. This policy sets out a framework for faster and more effective response to crises and emergencies better aligned with borrower needs and the Bank's experience and current engagement. 4 Noteworthy provisions include (i) streamlined processing and approval procedures; (ii) clarification that the Bank can support development objectives such as preservation of human, social, and institutional capital; and (iii) contingent emergency loans. Contingent emergency loans are loans to countries with high disaster risk determined to have appropriate institutional and implementation capacity to respond to emergencies. These loans, once triggered after the onset of a disaster, would disburse against a positive list of goods required under the borrower's emergency recovery plan and could offer very fast financial support. A multi-donor global partnership for hazard management, called the Global Facility for Disaster Reduction and Recovery to support this initiative has also been launched This policy change provides an opportunity to streamline the Bank's postdisaster income support. Finally, as recommended by the 2000/01 World Development report, the Bank's emergency response is increasingly incorporating poverty-sensitive disaster response planning, e.g., using social protection instruments, to help households cope with disasters. Social protection support is complementing mainstream components of the Bank's disaster support operations including, e.g., reconstruction and rehabilitation.
The South Asia region has also launched a Regional Hazard Risk Management Program focused on emergency preparedness, risk mitigation, and institutional capacity building. The strategy also includes the provision of social protection support to vulnerable groups affected by disasters. The protection of individual welfare in emergencies is also a tenet of the region's Social Protection Strategy. For this reason, in all three countries which experienced recent natural disasters--Maldives, Pakistan and Sri Lanka--the Bank's emergency response included social protection (cash transfer) components. In addition, in Pakistan, the Bank's emergency response also included support for those disabled in the disaster (as well as other disabled individuals) in the earthquake-affected areas.
The purpose of this note is to review South Asia's recent experience in providing social protection (mainly cash transfers) support to the poor in emergency response operations, with a view to improving its effectiveness. Available evidence suggests that cash transfer programs are in high demand by client governments and affected communities, appear to perform well in helping households cope with disasters, and suggests some guidelines for improving their design and implementation. It concludes that the Bank could provide timely and high-quality support through a best-practice design toolkit on social protection, a right-on-time technical assistance facility, and the integration of social protection in emergency preparedness by building capacity in social assistance/cash transfers agencies for helping households cope in emergencies. While the note focuses on evaluating cash transfers, it also reviews other social protection instruments commonly used for coping with disasters (e.g., workfare and social welfare and care services). South Asia is the focus of this note, but the findings may also be of interest and relevance to other developing countries and donors.
The organization of the note is as follows. The following section overviews the main social protection instruments used in natural disasters (workfare, cash transfers and social care services). A description of the cash transfer programs supported by the Bank, their similarities and differences, available evidence on their impact, and preliminary lessons learned from their implementation is provided in the third section. The paper concludes with implications for the Bank and Governments in making social protection an integral part of ex-ante planning for disaster response. 5 5 The note does not discuss the related, but slightly different, topic of ex-ante insurance against disaster risk These topics are taken up by Alain de Janvry, Elisabeth Sadoulet, Pantelis Solomon, and Renos Vakis (2006) in "Uninsured risk and asset protection: Can conditional cash transfer programs serve as safety nets?", World Bank Social Protection Discussion Paper 604.

ESTIMATES OF RELATIVE LIVING STANDARDS AND POVERTY INCIDENCE
The purpose of social protection instruments, when used in disasters, is to protect basic consumption and, if needed, to provide temporary employment. Without income support, households, particularly the poor, would deplete their physical and human capital and fall deeper into poverty, particularly as disasters can often exhaust community support mechanisms. The main social protection instruments used in time of disasters are (i) social assistance/cash transfers to meet basic needs (ii) workfare, (temporary low wage employment); and (iii) social care services (e.g., community-based rehabilitation for disabled, family placement for orphans, legal support to widows).

Cash Support
In normal or non-disaster, conditions, cash transfers or social assistance programs help provide minimum income support to the chronic poor in order to meet a country's distributional objectives. 6 In times of disasters, cash transfers allow affected households to preserve their human and physical assets, and therefore prevent the deepening of chronic poverty that would otherwise result from natural disasters.
In general, transfers to the poor can be given in cash or in-kind, e.g., food. Cash assistance provides households with greater flexibility and choice in meeting their basic needs as compared to in-kind transfers. Cash infusions in the economy can also help stimulate supply, though they can also generate inflation in supply constrained conditions. Cash is also less vulnerable to distributional and storage issues that plague food based safety nets. However, food may be preferred over cash in the circumstances where food markets are not functioning and where households are not able to meet basic food needs. 7 Inferior food goods can also be self-targeted, while cash transfers sometimes need to be targeted more explicitly to particular groups. (See Annex 2 for a more detailed discussion of the trade-offs between food and in-kind transfers). 8 Cash support has been highlighted as a successful and growing area of support in disaster assistance in a recent IEG report: 9 "During the recovery process, getting cash support to victims quickly has positively affected people's sense of safety and security. It has been a 6 World Bank (2007) (2007), "Cash-based responses in emergencies", HPG Report 24, ODI. 8 Bank post-disaster income support in South Asia has been given mostly as cash, and elsewhere has used food only in rare circumstances, in part because the Bank never was a major actor in relief, 9 "Hazards of Nature, Risks to Development". Independent Evaluation Group (IEG), 2006. prominent first sign of the government's support in a time of acute need. Since 1984, the Bank has funded over $850 million in cash support (cash transfer, cash for work, and similar programs) in the context of 11 projects, 5 of which are ongoing. Approximately 94 percent of these funds were lent since the Turkey Emergency Earthquake Reconstruction Loan was appraised in 1999. In projects that have closed and been rated, four out of six were found to be satisfactory. Those rated unsatisfactory accounted for less than one percent of the funds allocated. When promptly provided, cash support enabled people to survive and get local economies moving again, and was reported as highly preferable to in-kind support by beneficiaries. For example, the 1999 Turkey earthquake reconstruction project implemented a cash transfer component that was widely considered successful, and even a model to be emulated, as four subsequent projects have already done." Cash transfers can complement well other types of post-disaster efforts, including housing 10 and infrastructure reconstruction. 11 IEG (2006) tells how this complementarity came to light in Gujarat: "…earthquake victims' need for cash assistance became apparent to the Bank in an indirect way following the 2001 earthquake. Families were using the first installment of house construction funding to purchase food and other necessities to survive, rather than using it toward the construction. As a result, when it became time to issue the second installment, many of the families did not have the first phase of the house to show in order to receive the second installment." 12 These attributes have also made cash transfers attractive in other emergency situations. For example, cash transfers have helped households cope with post-conflict situations (UAS program in Sri Lanka IDPs, conditional cash transfer program for IDPs in Colombia) and have compensated households for income losses due to economic restructuring (e.g., price increases in Romania and layoffs in Eastern Europe).

Workfare (cash-for-work)
Workfare programs provide targeted employment for the poor in many countries around the world. Workfare has multiple objectives: provision of temporary employment to poor able bodied working age individuals during emergencies, providing cash support to the poor, and creating or maintaining assets. The work includes reconstruction or rehabilitation efforts, completion of small works such as clearing debris and reestablishing roads and water schemes. Workfare is needed immediately after the disaster when regular incomes have collapsed and unemployment is high. It has also been used in post-conflict situations to disarm and provide gainful employment to excombatants (e.g., Bosnia, Senegal). Workfare can be used to complement cash grants or it can be phased in after the termination of cash grants to provide more incentive constrained cash support. The best mix of social protection programs: cash grants and workfare, or both depend on the country context.
Workfare has a long history going back to at least 19 th century Britain and India, where it was used to avert famines; 13 since the 1950s, food-for-work has also been used as an avenue for distributing bilateral food aid. 14 In South Asia, workfare is used routinely in rural areas of Afghanistan, Bangladesh, and India both during disasters and in normal times. India and Bangladesh in particular have a long experience of using workfare to respond to disasters. Their governments (with donor support in Bangladesh) have permanent cash-for-work and food-for-work programs that they roll out in response to national and local disasters:  In India, the Government has recently introduced the National Rural Employment Guarantee Scheme, an India-wide workfare program to provide 100 days of guaranteed employment to the poor. In Southern India successful workfare projects have been implemented by NGOs, such as those by OXFAM, in response to the tsunami.
 In Bangladesh, workfare is routinely part of the response to disasters. For example, following the 2004 floods, permanent workfare programs such as Vulnerable Group Development, Vulnerable Group Feeding, Food-for-Work, Test Relief, and Gratuitous Relief distributed a total of 0.74 million metric tons of food grains in addition to some cash assistance, corrugated iron sheets, and clothing.
 Pakistan has implemented three labor-intensive workfare projects targeted to the Afghan refugee population during 1984-1996. Although their overall impact on poverty is not known, the projects created 22.6 million person days of employment over a 12-year period and transferred vocational skills to participants.
 In Sri Lanka, in response to the Tsunami, a cash-for-work program was introduced and implemented through INGOs, NGOs and community-based organizations. The works ranged from cleaning of debris to irrigation infrastructure management. Most of the resources (about Rs. 750 million) were spent on wages and the rest on tools and materials.
 Many UN agencies and NGOs regularly use workfare on a small or medium scale in disasters, for example to maintain or improve camps for the displaced. 15  The Bank has also supported workfare programs in response to emergencies. The World Bank has supported very large, national-scale workfare programs in Korea and Argentina 16 in response to unemployment caused by the financial crisis.
Workfare has many advantages. It helps affected communities return to normal life, offering hope and dignity. Offering manual labor with a low wage tends to attract near-poor and poor workers, allowing for self-targeting. If the wage is set very high, it tends to attract the non-poor and distort labor markets. Workfare is also justified when the assets constructed are of value to poor households and communities. Project selection is therefore important. Works that do not require much planning and capital such as clearing debris are best suited for workfare programs, but almost any job that requires unskilled labor can be offered as workfare. 17 However, large-scale implementation of workfare can be complex and not easy to safeguard from fiduciary problems, especially if there is no prior implementation infrastructure. One of the most frequently raised criticism of workfare is that too many projects are useless, merely digging holes and filling them up again, or constructing roads from nowhere to nowhere. 18 Thus, capacity building for workfare programs, including their roll out in case of disasters, needs to be developed exante in designated social assistance and local government agencies as part of the planning for disasters.

Social Welfare/Care Services
Some very vulnerable groups such as widows, orphans, and persons with disabilities have special needs in disaster situations-beyond cash, food, housing and employment support. For example, orphans require care and shelter, while persons with disabilities need rehabilitation, and many affected groups require psychosocial support. Social welfare and care services that link these groups to essential services, e.g., family placement or community based rehabilitation are an essential compliment to cash transfer programs in normal situations, and often an important need in disaster (and post-conflict) situations. 19 While not financed by the Bank, several types of social welfare and care services were an essential component of emergency response in Sri Lanka for family placement of orphans, providing psycho-social support, and reducing abuse in camps. These services can also be useful in post-conflict situations.
The Bank has only recently initiated support for social care services for disabled in emergency operations in South Asia. In Pakistan, the Bank is supporting rehabilitation and independent living of disabled persons, including those who suffered a disability during the earthquake through the Pakistan Earthquake Disability Project two linked grants to a disabled persons' organization and an international disability NGO. This suite of projects will help the Bank understand better what role it can play to promote rehabilitation of persons with disabilities in disaster affected areas in environments of low capacity. Still, there is a major gap in efforts and a concomitant gap in knowledge of what works and how to deliver such services in disaster situations.

FACTORS AFFECTING LIVING STANDARDS
In South Asia, the World Bank recently supported large-scale government implemented cash transfer interventions as part of its overall disaster response following major disasters in Sri Lanka and Maldives (2004), and Pakistan (2005), building on the successful example of the cash transfer program implemented to help earthquake victims in Turkey. Following on basic relief efforts, and rehabilitation and reconstruction efforts, the Bank used the deployment of cash grants to affected households on a remarkably large scale. While these cash grants are referred to in Bank documents as livelihood grants, suggesting conditionality on income generation, they were for the most standard social assistance/cash transfers that allowed households to meet basic needs.
The emergency cash transfer package to help households meet basic needs for each country is described in detail below. Cash grants (of much higher amounts) were also provided to households with damaged housing to aid reconstruction. 20 Pakistan: About 250,000 households received Bank-financed cash grants in six monthly payments of Rs. 3,000 (US$50) per household. The Government itself also financed payments to compensate for death and injury. Each family that suffered a death of a next of kin received Rs. 100,000 (US$1,660); families who suffered multiple deaths also received Rs. 100,000. Compensation for injuries ranged from Rs. 15,000 to 50,000, with the highest payments going to amputees.
Sri Lanka: Sri Lanka gave tsunami-affected households four rounds of cash grants of LKR 5,000 ($50) per round. The first and second round of cash transfers covered 250,000 households, but the number of households was reduced in subsequent rounds, e.g., 160,000 households received benefit in the third round. 21 Maldives: Maldives gave a one-time cash transfer for basic needs to 10,000 households affected by the Tsunami. The transfer amount was Rf. 500, 1000, or 1500 (US$39-117) per household member depending on damage (corresponding to 2-6 weeks of average consumption). For example, a family of six with a completely damaged house received Rf. 9,000 (US$700). 22 20 These packages also included cash grants for housing, rehabilitation and reconstruction components outside the scope of this paper. 21 Sri Lanka: Post Tsunami Recovery and Reconstruction (2005)

Similarities and Differences in Approaches
The cases reviewed above show many similarities but also a few differences in the Bank's use of post-disaster social protection support. 23 We have included Turkey as an interesting comparator country where cash grants were also given for basic needs, disability and death of main household earner. 24 The vulnerability criteria selected households with no senior government employee that were either headed by women not currently married, contained a disabled member, or contained five or more children. 25 It appears that local officials had considerable discretion in determining eligibility and criteria may have varied between areas.
The similarities include:  Bank support was provided in exceptionally devastating rapid-onset disasters in which there was political will for generous support to affected households; some of the lessons of these disasters may therefore not apply to less visible disasters or to slow-onset disasters.
 The Bank-financed cash support in all three operations, as there was a clear need and client demand for income support to affected households.
 The Bank decided, designed, deployed, and funded the assistance under great time pressure. This meant, among other things, that the design and implementation of Bank-funded income support did not always benefit as fully as possible from timely technical assistance.
 Given emergency conditions, formal impact evaluations were not part of the operations at the time of approval. However, surveys or spot-checks were undertaken (e.g., Maldives, Pakistan, Sri Lanka) as part of the process to assess whether transfers were delivered to target groups.
There are also interesting differences in design and implementation arrangements:  Benefit amounts and frequency of payments differed significantly. For example, Pakistan, although the poorest among the countries compared here, offered much more generous support (especially when compared to GDP) than Maldives, Sri Lanka, or Turkey (Table 1). In Pakistan, affected families 26 received between US$300 27 (if eligible for the cash grants as per the vulnerability criteria 28 ) and US$2,000 (if they also suffered a death), the equivalent of 7-48 percent of per capita GDP per member. In contrast, the income support per person was 3 percent of per capita GDP in Sri Lanka, and 1.5-4.5 percent of per capita GDP in Maldives. 29 In all cases, the benefit amounts were much higher than the (usually very low 30 ) regular benefit levels offered by these countries to social assistance beneficiaries (such as the chronic poor or those affected by smaller disasters where no foreign assistance is involved).
 The frequency of payments varied as well. Benefits were paid out in one lumpsum payment in Maldives (as in Turkey), in four tranches in Sri Lanka, and six 26 The Pakistan program used the unit of family (husband, wife, children) rather than household. As a result, there were instances of multiple beneficiary families under the same roof. 27 The poverty line was used to calculate the amount of basic needs/livelihood grant. 29 Cash transfers were 3-9 percent of per capita GDP in Turkey. 30  The Government's criteria were broader and included loss of any household member (disabled or missing), loss of business assets and loss of income: wage/salary employment or income from selfemployment and disability of any household member. Vishwanath, T and Narayan, A (2005) "Insights from a beneficiary assessment survey: Implications for the re-imbursement of cash grants". Memo. an area with a large informal economy. 32 In Sri Lanka, the first cash transfer to all beneficiaries was delivered in March/April, three months after the Tsunami.
 Delivery agency/systems. In Pakistan, the national cash transfer agency, the Baitul-Mal, was not used to deliver Bank financed cash transfers. The Earthquake Reconstruction and Rehabilitation Authority under the Prime Minister's Secretariat oversaw cash grant delivery, with participation from local governments and communities. Targeting teams comprising local government officials, working with communities, collected information using an open process and information campaigns, verified the targeting process, while appeals were handled by union councils. A monitoring system to track payments, manage appeals and cross check errors was developed NADRA, while a third party cross checking was instituted. 33 In Sri Lanka, the local governments, together with communities identified beneficiaries in affected communities and benefits were delivered through bank accounts. In Maldives, army and civil servants teams delivered social assistance in collaboration with the main social assistance program. Task teams visited all affected islands; the eligibility criteria were damage to the house of the individual. This was visually confirmed by the task teams on their first day on the island. The following day, task teams gathered the population and made the payment in cash, maintaining a basic list of beneficiaries  Mix of social protection instruments. The Bank focused on cash support in the short and medium term, except in Pakistan, where for the first time the Bank also financed disability and social care services (community-based rehabilitation). However, donors such as ILO (Sri Lanka) and Oxfam (India) also supported workfare programs to provide temporary employment for beneficiaries after the Tsunami.

Evidence on Impact of Disaster Income Support in South Asia
These variations in basic design of social protection support raise the question of whether interventions actually increased household income and how cash transfers and other assistance could be designed for greater impact. To answer these questions would require further much needed research that goes beyond the scope of this paper. However, some initial results from surveys on cash transfer recipients from Sri Lanka and Maldives are summarized below. An impact evaluation of the emergency cash transfers in Pakistan is underway. 34 32 Earthquake reconstruction-and the livelihood cash grants-are handled by the Earthquake Reconstruction and Rehabilitation Authority, which was created after the earthquake in the Prime Minister's Secretariat. 33 Vishwanath, T and Narayan, A. "Livelihood Cash Grants in the Aftermath of Disasters: Sri Lanka and Pakistan" (2006). 34 Vishwanath, T and Narayan, A. "Livelihood Cash Grants in the Aftermath of Disasters: Sri Lanka and Pakistan" (2006).
 In Maldives, government records show that more than 53,000 people were helped by the cash grant within one month of the tsunami, and a further 5,000 people who were missed out in the first round got help later. A survey was carried out of socioeconomic conditions before the tsunami and six months after. 35 The surveys shows that the support was well targeted: more than three-quarters of the affected population received cash grants, the share being higher for those most directly affected. The grants were also well targeted geographically. Islands hit more severely received more in-kind and financial aid. 36  A beneficiary assessment of the Sri Lanka cash grants program, completed after the first two rounds had been disbursed, reviewed the targeting and implementation of the cash transfers, and made recommendations for the final rounds. Key findings from the survey were:  Targeting: The assessment showed that few affected households were excluded from receiving cash transfers. However, many households who should not have received the grant did so. About 25 percent of households who received the transfer were ineligible according to the Government's inclusion criteria and 65 percent according to the Bank's more restrictive criteria (see targeting criteria above). Some of the inclusion errors were the result of pressures on the local authorities to include everyone. Based on this finding, targeting was made more restrictive, with a much smaller number of households qualifying for assistance in subsequent rounds, though phasing in more restrictive targeting proved difficult. 37  Incentives: The termination of the cash grants can be politically difficult because it means revoking a perceived entitlement. There was considerable demand in Sri Lanka (also in Pakistan) to continue the grants, with potentially adverse work incentives and upward pressure on wages and pricesparticularly given large amount of benefit (especially in Pakistan). This pressure to increase benefit might have been all the more great if affected areas continued to be economically depressed, particularly given the finding of the survey that many of the beneficiaries were workers or self-employed whose loss of income and assets from the tsunami was significant. Phasing out cash grants, under these conditions might require stimulating income generation, either through workfare or access to subsidized micro-credit. 35 "Tsunami Impact Assessment Survey 2005: A socio-economic countrywide assessment at household level, six months after the tsunami", Ministry of Planning and National Development, Maldives. 36 The food rations were largely adequate: around 80 percent of highly affected households answered that the amounts of rice, sugar, and bottled water provided were sufficient to meet their needs. The incidence of food shortage fell rapidly after peaking in late December, 2004 and by February, 2005 rates of food insecurity were back to pre-tsunami levels. Moreover, six months after the tsunami, employment had recovered and household income was already higher than before the disaster. 37 More than 61 percent of the recipients reported no damage to their houses (partial or full) in the survey questionnaire, while 33 percent reported some damage to their housing. The fact that the proportion of applicants with damaged housing is quite similar between those who were approved and those who were rejected, indicates arbitrariness in targeting this program.

Lessons Learned: Design and Implementation
Whichever type of benefit is offered, success requires careful attention to details of design and implementation-how will support be targeted, distributed, to whom, how much, and for how long? Since it takes time to design support instruments, approve their funding, and especially to set up implementation arrangements, there are advantages of drawing up plans for disaster response and building implementation capacity before disaster strikes, and working with existing agencies that deliver cash transfers to the poor. This section discusses some of the design parameters for successful cash transfer (social assistance) support. Although a lot remains to be learned in this area, some tentative lessons of best-practice design of emergency income support can be distilled from the reviewed experience:  Preference for Cash. The form of income support should be cash unless there is evidence that food markets have been disrupted. In rapid-onset disasters, food and other in-kind income support can often play a useful role in the immediate postdisaster period because the disaster often resulted in disturbance of markets. However, in-kind support should eventually be discontinued and converted to cash support, except if food markets remain interrupted. In slow-onset disasters (other than conflicts), food markets are much less likely to be disturbed and the argument for cash is therefore stronger.
 Eligibility Conditions. Targeting can be either universal-all households in affected areas-or, if the number of victims is large relative to resources available-targeted to those most in need. The best approach depends on the purpose of the assistance (short-term recovery and food security or combating long-term vulnerability); the scope of the disaster (small and localized or large); and on what indicators can best be verified.
o Targeting all those affected by the disaster, using simple, transparent and monitorable criteria has merit in small to medium disasters, e.g., targeting those who lost asset or house has merit if the objective is to help all the affected regardless of their pre-disaster income status. It can be a relatively simple, cheap, quick, and transparent method, as the Maldives example illustrates, although it runs the risk of excluding the homeless and people renting or without secure tenure. It also runs the risk of including people not very directly affected as the Sri Lanka example illustrates.
o Targeting the poor has greater merit in very large disasters where budget constraints may lead to a need to target, and in the medium term where the objective shifts to helping only that subset of the population that is at risk of becoming chronically poor without help. 38 Despite the most careful targeting, while exclusion errors should be minimal, a larger error of inclusion in times of disasters is to be expected. Strengthening the capacity of the main social assistance agency to deliver cash transfers in times of emergency is therefore essential (see more on this below). Experience from both Pakistan and Sri Lanka shows that a targeting system for disaster victims at the time of the event itself can be costly and difficult to implement, with huge data needs, is very demanding of scarce managerial resources, and can result in delays in benefit delivery and compromised targeting effectiveness.
 Providing transfers using death and disability as eligibility criteria may have additional moral hazard implications, particularly where administration and record keeping are weak. For example, providing death benefits can run the risk of overstated disability or casualty figures because of false claims. One way around this, at least in the case of disability, is for grants to disabled to be provided to the most obvious disabilities-loss of limb, for example, given difficulties of ascertaining and certifying less visible disabilities.
 Level of Benefit. The amount of support should be adequate for subsistence but not so much as to jeopardize work incentives. These amounts might be higher than provided by the main social assistance program (if these are very low) but only initially, and then should decline over time, to avoid adverse incentives, promote equity with other beneficiaries and contain fiscal costs.
 Duration and Frequency of Benefit. Timely cash support should follow for a period that would be very limited for most households, but that could stretch into the medium term for the more vulnerable, with these individuals being absorbed into the main cash transfer/social assistance system if poverty persists. Ascertaining the correct amount of emergency income support requires more research. However, a possibility would be, in small-to-medium sized emergencies, to provide a first larger transfer to all affected, followed by a second somewhat smaller transfer to all households after perhaps 3 months, with transfers narrowly targeted to vulnerable households after this period through the main social assistance system. In very large emergencies, targeting of households in affected areas may be required with the most vulnerable requiring longer duration of support (through the main social assistance system). The frequency of benefit could be staggered over time if good banking channels exist, otherwise infrequent or one-off transfers may be better in order to reduce transaction costs for program administrators and beneficiaries and to provide timely transfers.
 Managing Expectations. Managing expectations of beneficiaries is important in most entitlement programs. Providing cash grants to beneficiaries increases their expectations of continuing this entitlement. Ensuring that the public is aware about the one off nature of the grants and enforcing strict limits on grants duration (and potentially keeping benefit frequency low-one or two rounds as most) would be important to mitigate these pressures. Those who continue to be poor should be folded into the main cash transfer program (with adequate financing to the program provided). If employment does not rebound at the time of termination of cash transfers, some time-bound form of conditional grant might be introduced, for example workfare and/or micro-credit.
 Payment Mechanism. For reasons of security and fraud prevention, well-developed banking systems should deliver cash transfers, with uncovered beneficiaries assisted to open bank accounts. However, determining the best approach for delivering cash transfers in low-income settings-private agents, post offices--requires more research. Cash-in-hand by public servants or army officials may also avoid the delays of opening bank accounts, where opening accounts may take some time. Depending on the risks of fraud, insecurity, and corruption, it may prove possible to safeguard direct cash disbursement by the public agency to a reasonable extent, especially for a limited period.
 Monitoring and Evaluation Surveys and other tools for gauging the effectiveness of the targeting and other administrative processes for delivering cash transfers or other social protection instruments used in emergencies should be instituted where possible. Given enormous data requirements for administrating cash transfer programs, MIS system to, inter alia, register beneficiaries and monitor payments is also essential for effective delivery of social protection to affected households. These systems should be already in place in national social protection agencies, with options to expand when disasters unfold.
 Strengthening the Capacity of the Main Social Assistance agency. Ideally, the main safety net agency should coordinate or deliver cash grants to affected areas in partnership with local agencies. For this to be possible, the regular safety net programs would need budgetary and logistic mechanisms for scaling up in disasters, including provisions for taking on additional staff to handle the logistics of emergency support in disaster situations. Building program capacity for timely and efficient delivery of support to affected households within existing cash transfer agencies, even if these capacities do not exist, should be an important component of ex-ante disaster planning.
 Mix of instruments. Cash transfers can help to flexibly address the basic needs of households where food markets are functioning well. However, if employment does not recover as the duration of cash transfers ends, other (time-bound) instruments such as workfare or micro-credit may be needed in combination with cash transfers to help families avoid poverty.

POLICY IMPLICATIONS AND SCOPE FOR FUTURE RESEARCH
Summing up the above, social protection support, particularly cash transfers, has been an important and well-performing part of the disaster response in several major recent disasters and will be needed in future disasters. As noted in this paper, these instruments have also been used to help households cope with income loss in post-conflict situations and in times of economic crisis To prepare for better design and more timely and efficient delivery in future disasters, this section lists implications and possible next steps for the client countries and for the Bank.

Implications for client countries
As part of government disaster planning efforts, countries could plan social protection responses aiming for swift, equitable, and consistent deployment of support. These plans would include emergency income support as an essential function of the country's main cash transfer program, with implementation ensuring that this support is timely, efficient, and equitable for those affected by disasters. This means that countries would need to start investing in their main cash transfer programs to improve their administrative efficiency to respond to natural disasters Ex-ante design and planning for emergency social assistance would provide answers to the key questions of disaster intervention design: what are the form, amount, and duration of emergency social protection support. What should be the type (or mix) of social protection instruments, e.g., cash, workfare and social care services? Who should be eligible for support and how targeted? Which agency should be responsible for its delivery? Plans and implementation arrangements would vary by country and type of disaster, but there are many basic principles that need discussion and agreement ex-ante. Perhaps most importantly, the implementation agency should be designated and start building the necessary response as part of the Government's overall emergency response and social protection system.
Possible lead agencies in South Asia for emergency income support could be:  In Bangladesh, the Ministry of Food and Disaster Management already runs safety net programs that are triggered periodically by disasters, and would probably be well-placed to take the lead on this in coordination with others agencies, e.g., local government.  In India, the Disaster Management Authorities at national and state level (who report to the chief ministers) could coordinate efforts, working with relevant safety net agencies, together with the Ministry of Social Welfare that delivers cash transfers (social pensions, disability payments, etc.)  In Nepal, the Ministry of Home Affairs' Disaster Relief Section is the apex body for disaster management responsible for rescue and relief to disaster victims, and together with the Ministry of Social Welfare could perhaps lead these efforts.  In Pakistan, the main federal cash transfer agency, Bait-ul-Mal, could take the lead in coordination with other federal and provincial authorities.  In Sri Lanka, the main safety net program Samurdhi is one agency that could lead the efforts.

Implications for the Bank
Given recent experience and the likelihood that Bank support for disasters will be required in the future, it is important to consider how best to provide timely and high quality social protection support. Recent experience has shown a remarkable variation in the design and execution of Bank-supported emergency income support. It has shown that even if financing is timely, procurement and technical delays can still adversely affect the efficiency of Bank support in emergencies.
The Bank could enhance its operational effectiveness in the following ways:  Integrate social protection as a key component of the ex-ante planning already ongoing as part of the countries' general hazard management work and supported by the new Bank-led Global Facility for Disaster Reduction and Recovery, and the South Asia Regional Hazard Risk Management Program. Efforts will need to be country-specific. Each country's hazard management plan or strategy would contain a section elaborating on deployment of social protection in case of an emergency.
 Develop a best-practice toolkit that would help task teams and client countries design emergency income support programs, including monitoring systems, in accordance with international benchmarks. This would enable more consistent advice and streamlined project delivery.
 Provide capacity building of designated safety net agencies to respond to disasters, e.g., advance operational planning, Management Information Systems, and mechanisms for expanding cash transfers and other programs, as well as requisitioning additional staff and budgets in future disasters. This support could be provided as part of the safety net capacity building on-going in several countries, or could be initiated as part of the planning for natural disasters in the country's main social assistance agencies. In cases of emergency in countries where this capacity is weak, utilizing the main cash transfer/social protection agency at the time of the disaster to deliver cash transfers would be beneficial for two reasons: (i) it could help strengthen the regular safety net program and increase its capacity to deliver emergency cash transfers; and (ii) working with the main agency, can help disaster-affected individuals that are long-term poor integrate with the main safety net program.
 Consider on-time post-disaster technical assistance and implementation support to enable task teams to support best-practice implementation from the onset of the disaster, even before loans are processed. The Bank could award contingent standby contracts to firms with the required expertise to become effective immediately upon a disaster, anywhere in the world. Work under such contracts could be financed either from a trust fund or from some kind of revolving fund (to be replenished by borrowing countries after loan processing). The Bank and other donors could even go one-step further and establish ex-ante funding for the cash transfers and other social protection support through contingency financing instruments that would become effective upon the occurrence of an emergency event as the Bank's new policy on Rapid Response to Crises and Emergencies (BP/OP8.0) opens for. Given that South Asia is prone to disasters, it could well spearhead such innovations.
The social protection policy options outlined above are intended to help the countries cope better with natural disasters and other emergencies. Ultimately, in concert with other measures, rapid and high quality social protection support could assist affected populations to meet their basic needs, restore their income, and avoid irreversible losses of physical and human capital that would lead them deep into poverty.

ANNEX 2: CASH VERSUS IN-KIND
The debate on cash versus food and other in-kind support is an old theme in development economics. Compared to food, cash is often superior because of its versatility, the dignity and choice it offers to beneficiaries, and its lower transaction costs thanks to easier transport, storage, and accounting. Further, many beneficiaries of food, clothing, blankets, or buckets have been known to resell these items in local markets, at a discount.
Were recipients offered cash instead, they would often be able to purchase these items easily in local markets. Moreover, large-scale imports of food grains can depress local market prices and adversely affect producer incentives for local farmers. Further, cash is often surprisingly safe to move and distribute, in part because it is less visible than food transports. In fact, cash has been used successfully in complex emergencies, in countries with weak banking systems, and in countries in conflict; cash was even used in lawless Somalia, relying on traditional hawala banking channels. In contrast, moving in-kind supplies, coordinating their distribution, and safeguarding them from theft and abuse is hard, even in normal times, and more so in disaster situations.
In-kind support is justified under specific circumstances: when food and basic goods are unavailable in local markets, or where benefits are to be targeted to specific household members as with supplementary feeding of infants or school meals to students. These circumstances sometimes exist in the immediate aftermath of a disaster, especially if normal roads and supply channels have been interrupted. In such circumstances, cash could cause inflation in the price of key basic goods and there is a strong case for food and other in-kind relief.
The time dimension is important: emergency supplies of food, water, blankets, tents, roofing sheets, and water containers have saved many lives immediately after severe disasters. Yet most countries have active private trading channels that bounce back quickly after disasters that can often supply essential goods at lower cost than donor operations. South Asian countries, for example, have opened to international trade in food and have vibrant private trading channels. When donors or governments prolong their inkind support, as they often do, they undermine private traders, market recovery, and local producer incentives. They also play to the sometimes well-founded suspicion that their choice of support is more motivated by donors' interest in reducing excess food stocks than by the need of the recipients.
A major new study of cash support in the humanitarian sector finds that its use is growing but lacks much behind potential and should be expanded. The study notes that: "Cash should not be seen as a sector in its own right, but as a mechanism that needs to be considered across all sectors of humanitarian response. As such, all relief practitioners should see cash as one of the options available to them. This implies that information on how to implement cash based responses should be included in generic policies and guidelines, in induction training for new staff, in technical training courses, for instance on assessment methodologies, and in disaster preparedness and contingency planning processes……. ……Donors will also need to develop the skills and capacity to make informed decisions about whether to fund cash responses. The central role played by national governments in providing cash aid in Pakistan and following the Indian Ocean tsunami suggests that, where governments have the capacity, they are the most appropriate delivery channels." 39 39 Paul Harvey: "Cash-based responses in emergencies", ODI Humanitarian Policy Group (HPG) Report 24, 2007.

ANNEX 3: TURKEY 1999 EARTHQUAKE -OVERVIEW OF EFFORTS AND IMPACTS
The lessons from this natural disaster seem to have inspired efforts after later South Asian events, and are therefore worth dwelling upon. Probably because more time has lapsed, more evidence is available on the performance of assistance given in response to this natural disaster, which is briefly summarized here.
The Government rolled out a range of support programs in response to the Earthquake. Accommodation assistance was offered in cash to households that had suffered medium or heavy damage to their homes and that choose not to avail of free housing in public facilities. A one-time grant was also given to repair damaged housing. This was complemented by social assistance in the form of a one-time lump-sum payment for each death and permanent disability. The benefit levels were TL 750 million per death, TL500 million per first degree disability and TL 300 million per second degree disability (US$653-1632). In addition, survivor and disability pensions were introduced.
Efforts were successful. Impacts have been reviewed, for example in the Investment Completion Report (ICR) and in an IEG evaluation report. 40 The cash allowances were viewed as an important step in helping people cope with the effects of the earthquake, overcome trauma, and reintegrate into normal life. The provision of cash allowances was a tangible sign of the government's efforts to help the victims, and had a positive impact not only on the material living conditions of recipients, but on their psychological orientation as well. The cash nature of the benefits was important, both in terms of meeting the needs of the recipients (the applicants revealed a strong preference for assistance in cash as preferable to aid in kind) and in the logistics of getting assistance to victims quickly, which could not have been done if benefits were in kind. In conclusion, cash transfer projects can make an important contribution to recovery. Some lessons from Turkey's experience for other efforts are:  It was possible to effectively and accurately administer emergency cash transfers on a large scale.  Using institutions with cash transfer experience (that already have the necessary human resources and management systems in place) was critical to smooth delivery of the assistance program.  Getting cash support to victims quickly had positive effects on the economy and on people's sense of safety and security. It was a prominent first sign of the government's support in a time of acute need.  The process was streamlined by the agency delivering the cash, who worked together with volunteers, other government agencies, and NGOs. In one project site, application processing, bank account creation (for allowances), and result notification were combined into one procedure.  Because cash transfer runs the risk of creating dependency, public awareness campaigns are required that make it clear up front the number of months that payments will be made. When beneficiaries know this, they can make informed resource allocation decisions. Otherwise, going month to month without clear information on the duration of payments can lead to dependency beyond the short term.