Comparing cash and food transfers: findings from a pilot project in Sri Lanka

By Lili Mohiddin (Oxfam GB), Manohar Sharma (IFPRI), Anette Haller (WFP Rome)

Lili Mohiddin, Manohar Sharma & Anette Haller

Lili Mohiddin has been an Emergency Food Security and Livelihoods Advisor with OXFAM GB since September 2005, based in the UK and supporting the teams in southern Asia and the Middle East, and the Commonwealth of Independent States (CIS) countries.

Manohar Sharma has a PhD from Cornell University and is currently a research fellow at the International Food Policy Research Institute (IFPRI), Washington, USA where he conducts research in areas such as poverty reduction policies, agriculture, rural finance, and evaluation of social programmes.

Anette Haller is a programme advisor in the Emergency Needs Assessment Branch of the UN World Food Programme (WFP).

The authors would like to acknowledge the valuable contributions of the following people to this article: Jonathan Campbell (WFP Sri Lanka) and Agnes Dhur (WFP Rome), Susanne Jaspars (Emergency Food Security Specialist), and Pantaleo Creti and Chris Leather (Oxfam GB). The article is based on preliminary research results and has not undergone formal peer review. The views expressed in this article only reflect those of the authors.

Food for work programme, Mullaitivu, Sri Lanka

This field article shares the findings of a pilot cash transfer project that gives valuable insight into household expenditure and consumption patterns in cash-receiving and food-receiving households over a three-month period in Sri Lanka.

Thematic research on cash and food transfers has increased significantly since the late nineties. However, comparative research on cash and food transfers remains limited and claims about the relative advantages of one or the other are often based on comparisons of programmes implemented in different contexts.

In 2006, the World Food Programme (WFP), Oxfam (GB) and the International Food Policy Research Institute (IFPRI), through a unique collaboration, implemented a Cash Transfer Pilot Project (CTPP) in Sri Lanka in the aftermath of the tsunami. The key objective was to compare the impact of cash and food transfers on beneficiary households' food and livelihood security and on the local economy. A wider objective was to learn how best to determine the feasibility and appropriateness of cash-based programmes in emergency food-security assessments.

The role of WFP was to supply the resources necessary for the implementation of the pilot. Oxfam GB provided technical implementation expertise and IFPRI undertook project impact analysis¹. The collaboration drew on the relative expertise of each agency and included the joint design of the project and the Monitoring and Evaluation (M&E) approach.²

Background

The eastern, northern, and southern coastlines of Sri Lanka suffered heavily from the tsunami in 2004, with at least 38,000 deaths, approximately 5,000 people missing, and 500,000 people displaced. Many of the worst affected areas on the east coast were fully or partially under the control of the Tamil Independence Movement (LTTE), and had been subjected to years of conflict and, in some areas, frequent displacement. World Bank led assessments indicated damage of around $1.5 bn.³ Resulting unemployment was substantial, with estimates of one million job losses. WFP started emergency food distribution within days after the tsunami struck. Initially this consisted mainly of imported food, as it was prior to the harvest and local prices were not internationally competitive at that time. However, following predictions of a good 'Maha' rice harvest, the Government of Sri Lanka discouraged rice importation. Furthermore, various non-governmental organisations (NGOs) (such as Oxfam, Save the Children UK, and the British Red Cross) raised concerns that import-based food aid, especially rice, would distort local markets and made strong arguments for cash-based interventions.⁴

Pilot cash transfer project (CTPP)

Mother and Child Nutrition (MCN) Clinic, Mullaitivu, Sri Lanka

WFP Sri Lanka undertook a series of Emergency Needs Assessments (ENAs), which considered the scale of the response from organisations and the Government of Sri Lanka, and the ability of households to re-establish their livelihoods. The last ENA assessment in May 2005 advised that there should be a reduction in food assistance from 900,000 to 350,000 people, i.e. a shift to a more targeted intervention (Vulnerable Group Feeding [VGF] programme) starting mid August 2005, following the completion of the General Food Distribution⁵. Based on a feasibility assessment, a pilot cash transfer project (CTPP) was proposed as part of the VGF response. An additional study commissioned by WFP⁶ in May 2005 identified key issues in assessing the appropriateness of cash transfers and the feasibility of such interventions in Sri Lanka, as well as providing inputs into the design of the CTPP. Delays in the completion of the General Food Distribution, due to setbacks with the local procurement of rice, supported the case for a three months VGF programme, in which 12,000 out of 312,000 beneficiaries⁷ received cash in lieu of a full food ration between October/November 2005 and January 2006⁸.

The same household targeting mechanisms were used for cash and food beneficiaries. These included:

• Families whose houses were completely damaged by the tsunami
• Families from partially damaged households in the buffer zone9
• Families in the buffer zone with no damage to their houses but who had lost their main livelihoods
• Families considered as destitute.

The amount of cash disbursed was equal to the local market value of the WFP food ration, taking into consideration average seasonal market price fluctuations in the area. The total transfer value amounted to 150 Sri Lankan rupees per beneficiary per week, or US$1.5.¹⁰

The cash was distributed on a fortnightly basis to targeted households from randomly selected communities in three districts of Sri Lanka (Batticaloa in the east, Galle and Hambantota in the south).¹¹ The Samhurdhi community bank network, normally used for social welfare payments, community-based savings schemes, and more recently, large-scale government tsunami cash payments, provided the cash-delivery mechanism.

WFP transcribed the cash-beneficiary household head's name, identity card number, and household size (provided by local government officials) onto coupons that indicated the amount due per person, total household entitlement, and a calendar for indicating cash collection days. All coupons were 'signed-off' by the officials, with any unclaimed cards returned to WFP. The cash was redeemable from banks on presentation of the coupon and the identification card. Failure to redeem cash within the collection period, spanning three consecutive days, would result in non-payment.

For the food-beneficiary households, the food-delivery mechanism was not modified for this pilot. Local officials submitted lists of targeted household names to WFP, and in return received an appropriate number of blank coupons for completion. On receipt of the coupon at the distribution point, households received allocated food. Forms indicating receipt of food would normally be signed, as would coupons. When there were inadequate amounts of all or some foods for distribution, households would be informed. As there were no fixed days for collection, beneficiaries were not penalised for non-attendance.

Key findings

The CTPP has allowed comparative analysis of household expenditure and consumption patterns in cash-receiving and food-receiving households over a three-month period. However, these patterns should not be extrapolated over a longer period of time, or expected in other contexts, for which they may not be representative.

The appropriateness of cash depends on a secure environment where markets are functioning

Communities in the more densely populated and less conflict-affected south of the island had very good market and bank access. This resulted in less time spent collecting cash and accessing markets, and as market prices were stable, the 'value' of the cash transfer did not depreciate. There were no restrictions or limitations on consumer and trader movements or products sold in the area. Almost all cash-receiving households in this area preferred receiving cash to food.

In the more poverty-stricken LTTE-controlled east, access to banks and markets was inferior. At the time of project planning, the availability of public transport and traders within the target area led to the assumption that market access was adequate and that traders would respond to any increase in demand induced by the cash transfer.
However, an unforeseeable deterioration in security within the area resulted in more vigorous controls at roadside government checkpoints. Traders were restricted in the amount and types of produce that were allowed into the area, and consequently food prices increased and the actual value of the cash transfer eroded proportionately. Further, movement control and higher transportation costs also imposed higher transactions for consumers in purchasing food. For these reasons, all households in this area indicated preference for food rather than cash transfers.

This highlights the importance of not only an initial assessment but also ongoing monitoring of both the security and market environments, so programmes can be adapted if and when necessary.

Cash beneficiaries diversified their diet and bought non-food items

Overall, when households did receive cash, they diversified their diet. They spent more on dairy products, meat, packaged foods, and nonfood essentials such as clothing and footwear, and they bought cereals with a higher market value than the ones supplied by WFP. These increased expenditures were financed by reductions in the consumption of key staples.

For both household groups (cash-receiving and food-receiving), calorie consumption declined over the project period, partly because important Hindu and Muslim festivities took place during the baseline survey. In this period, households would have incurred additional festival-related costs and in the instance of Muslim households, there would have been changes both in types of food consumed and consumption patterns. In the poorer, more remote and conflict-ridden communities in the east, the decline in per capita daily calorie intake was significantly steeper for cash-receiving households (a decline of 535 calories) than for food-receiving households (a decline of 290 calories), suggesting that cash transfers there had a net negative effect on household calorie intake. This was partly due to the higher level of unmet non-staple and non-food needs of the poorer households in this area and also due to higher liquidity afforded by cash (cash gave households the opportunity to purchase goods which would otherwise have required them to save up over a period of time). Also, as already mentioned, transaction costs imposed by remoteness and conflict had the effect of eroding the value of cash transfers relative to food transfers.

The decrease in food intake could also have been due to the fact that there remained some scope for reduction in the consumption of staples without compromising basic calorie intake. Even at the end of the cash distribution, average per capita daily energy intake was slightly above 2,100 kcals for both cash-receiving and food-receiving households.

Little apparent impact on livelihoods

Although cash-receiving households cited investment in home improvement and businesses as priorities, there was no significant difference in actual expenditure in these areas between cash-receiving and food-receiving households. This could perhaps be attributed to both having knowledge of government/NGO plans to rebuild houses, and expectations of NGO livelihood projects continuing in the future.

Another possible explanation could be that the amount of cash was insufficient for the purchase of livelihood assets or investment in small-scale business. The cash transfer was worked out on the basis of the value of the food ration rather than the value of the basic livelihood assets that needed to be rebuilt.

In the east, the impact evaluation also found that cash beneficiaries reduced their engagement in casual labour over the implementation period. However, further understanding of household motivation to seek casual labour during this period is required before a definitive conclusion on the effects of cash transfer on work -related decisions is made.

Working through local banks was effective

The Samhurdhi banks were suitable partners for distributing cash because of previous experience in large-scale cash distributions, their extensive geographical coverage, and their knowledge of the targeted communities. Bank staff members were paid according to the number of days worked, they were trained, and most importantly, they were involved in the design of the disbursement system and coupons. They were efficient, had low logistical costs, and accounted for all the cash transferred from the WFP bank account. Nearly all cash beneficiaries involved in monitoring the process expressed satisfaction with the bank services. In contrast, over a third of food beneficiaries were dissatisfied with the food distribution system, reporting long queues. Spoiled food and under-scooping were also reported by a small percentage of beneficiaries. Food-related logistical costs were significantly higher and unlike cash, a percentage is expected to be lost in 'losses and leakages'.

Cash was more cost-efficient, but costeffectiveness depends on context

Cost-effectiveness compares expenditure (costs) and outcomes (effects) associated with an action. Cost-efficiency relates to the cost of delivering the transfer (transportation, administration, delivery costs) compared to value generated.

The cash disbursement system was more cost-efficient (5 per cent cheaper) than the food system in all geographical areas. This cost-efficiency calculation considered the costs to WFP of providing cash and food assistance (calculated as a cost per beneficiary), and the value of the cash and food assistance to the beneficiaries. Expenses related to logistics included external and local transportation, handling and storage, but not WFP human resources. The lower cost of delivering cash was largely due to the existence of a well functioning bank network compared to the relatively high-costs of moving food. It is likely that when staff costs are considered, the cost-efficiency of cash will be more pronounced. However, this should not be taken as a valid statement on cost-efficiency in other contexts, where delivering cash might be more expensive due to factors like insecurity and lack of financial infrastructure.

The cost-effectiveness of food was higher in the east in areas where the security was deteriorating, markets and banks were difficult to get to, and where market prices were prone to higher fluctuations. Conversely, the cash was more cost-effective in areas with integrated and competitive markets, better bank access and lower market-price fluctuation. Neither food nor cash transfers reflected beneficiary transportation costs,¹² nor did cash transfer take into account price fluctuations. Both are cost elements that affect cost-efficiency and effectiveness calculations.

Fears that cash expenditure would lead to adverse social impact and gender inequity in resource decision-making were not realised

At the project planning stage, there were reservations that replacing food with cash would have a negative impact on food-security and decision-making. It was feared that the male member of the household would control the cash entitlement and would be more likely to purchase items like alcohol. The results show that these assumptions were unfounded. The level of joint decision-making between husbands and wives was slightly higher in cashreceiving households compared to food-receiving households. Alcohol consumption, although it increased marginally in both household types over the implementation period, increased less for cash households. In households with a high level of female control,¹³ there was a slightly higher expenditure on diverse food items and packaged goods, and a reduction in the purchase of alcoholic beverages compared to other households.

The findings of the Sri Lanka project are similar to those of projects elsewhere

The pilot project findings are similar to those of projects elsewhere, for example in the Horn of Africa (Kenya, Somaliland, Uganda) and Asia (Pakistan, Afghanistan, Bangladesh).¹⁴ These projects also found that a large proportion of cash is spent on basic food and non-food items, particularly when small amounts of money are received on a frequent basis. Other studies showed that expenditure on livelihood assets is more likely to result from projects of longer duration, where larger amounts of cash are disbursed and where cash is a complement to food aid (rather than replacing food aid).

Conclusions

In areas where markets were functioning and accessible, cash transfer was more cost-effective and preferred by beneficiaries. In those areas where markets were less functional or accessible, food assistance was more cost-effective and preferred by beneficiaries.

The appropriateness of cash programming depends on market access and functioning (whether they are competitive and integrated), and security. Food aid is more appropriate in contexts where markets are not working well, where security conditions impose higher market transactions costs for consumers, and in situations of high and unpredictable inflation. Opportunities exist for using both interventions in parallel or in a phased approach depending on seasonal and contextual changes over time and space. This is especially relevant to emergencies, where market access could be a limiting factor. In the immediate aftermath of a shock, food intervention may be more appropriate. Cash-based interventions may be gradually introduced as markets recover and could potentially be used as an exit strategy.

When livelihood recovery, protection, or support are project objectives, analysis should include the identification of livelihood groups and the seasonality of their livelihood activities. The size of the transfer needs to be based on the value of assets that need to be rebuilt, which in turn should determine the scale and duration of the project.¹⁵

Analysis of both cost-effectiveness and costefficiency of interventions should be standard practice in food and cash interventions in order to get a wider body of evidence on the relative costs and impact of these two approaches.

For further information, contact: Anette Haller, ODAN, WFP, Via Vola 68, Parco de' Medici 00148 Roma, Italia. tel: +39 (0)665 132528 email: anette.haller@wfp.org

Manohar Sharma, IFPRI, 2033 K St., NW., Washington D.C.2006, USA. tel: +202 862 8142 fax: 202 467 4439 email: m.sharm@cgiar.org

Lili Mohiddin, Emergency Food Security & Livelihoods, Humanitarian Dept., Oxfam House, John Smith Drive, Oxford, OX4 2JY, UK tel: +44 (0)1865 473849 email: LMohiddin@Oxfam.org.uk

¹ Oxfam seconded a Programme Manager with cash implementation experience to WFP for the duration of the project, and an Oxford-based food security and livelihoods adviser provided regular technical inputs. Oxfam's participation in the project should not be perceived as endorsing WFP involvement in cash programming, but rather as assisting with the analysis of the relevance of cash assistance to WFP's mandate.

² N. Edirisinghe (2005). Anticipation of the effects and comparative advantages and limits of proposed cash transfers in lieu of food in Sri Lanka's tsunami-related emergency through the use of specific assessment tools. Prepared by the WFP Strengthening Emergency Needs Assessment Capacity (SENAC) project, supported by ECHO. http://www.wfp.org/odan

³ Data taken from WFP Needs Assessment, January 2005. http://www.wfp.org/odan

⁴ Oxfam International (2005). Making the case for cash: Humanitarian food aid under scrutiny. Oxfam Briefing Note, available at: http://www.oxfam.org.uk/what_we_do/issues/ conflict_disasters/bn_cash.htm.

⁵ Tsunami Recovery Programming & Needs Assessment Follow-Up Mission Sri Lanka, WFP May/June 2005. Available at: http://www.humanitarianinfo.org/srilanka/infocentre/ assessments/others/doc/WFP_NA_May/WFP_NA_May.pdf

⁶ See footnote 2

⁷ This is equivalent to 3,276 households. On average, there are five members per household.

⁸ The food ration (per person per day) included 400g of cereals (rice/wheat flour), 60g lentils, 20g oil, 20g sugar, and 40g Corn Soya Blend, providing some 2000 kcal per day.

⁹ A zone defined by the government, up to 200 metres from the sea, where rebuilding houses is prohibited for safety reasons.

¹⁰ The exchange rate at the end of 2005 was 1US$ =Rs.100.

¹¹ As ethnic diversity in cash recipient households was desired, careful district selection was required.

¹² Transport costs were significant for food beneficiaries, as the majority received a single bulk delivery of 12 weeks of food ration.

¹³ These are households in which the women control the money to purchase food from the market.

¹⁴ S. Jaspars (2006). From Food Crisis to Fair Trade - Livelihoods analysis, protection and support in emergencies. ENN Special Supplement No. 3. http://www.enonnline.net See ODI website for additional publications on cash programming: http://www.odi.org.uk/.

¹⁵ This is clearly highlighted in P. Creti and S. Jaspars (2006). Cash Transfer Programming in Emergencies. Oxfam, http:// publications.oxfam.org.uk/oxfam/display.asp?isbn=0855985631.