2. EXECUTIVE SUMMARY
The World Food Programme (WFP) plays a pivotal role in the food security of Syrian refugees within the Syria +5 region. WFP periodically reviews its operational approach, so it is timely to consider what the best modality for the next phase of the response might be, including the pros and cons of delivering assistance through unrestricted cash as opposed to food-restricted value vouchers electronically redeemed at designated WFP-contracted retailers.
Emerging humanitarian research suggests that unrestricted cash transfers offer an effective and efficient alternative to value vouchers for refugees in host communities. Recently, several international agencies delivered cash assistance in Jordan and Lebanon and reported positive results. Some of WFP’s key donors have expressed a strong inclination toward food assistance in the form of unrestricted cash, whereas others favor food-restricted value vouchers.
This study focused on beneficiaries living in host communities. Those living in refugee camps were excluded. Therefore, results and conclusions reported here reflect this sampling decision. They are representative for Syrian refugees in host community settings in Jordan and Lebanon but may not be fully representative of refugees in camps. The findings may well be applicable in comparable contexts and middle-income countries with functioning markets, but they may not necessarily apply in a very different refugee-host combination, such as in locations with limited market functionality, failed states, or situations where there are food shortages.
This study’s objective was to compare the impact of WFP’s assistance-delivery modality, whether unrestricted cash or food-restricted value voucher, on (1) food security and other basic needs of beneficiaries and (2) the program’s costeffectiveness for WFP. In this study, use of cash was unrestricted; the beneficiaries could spend it freely. In contrast, the voucher option was restricted to food items at the voucher-accepting WFP-affiliated retail outlets.e better with contextual changes.
2.1. Impact of modality switch on food security and other basic needs of beneficiaries
The study was designed as a randomized controlled trial involving 3,123 communitybased beneficiary cases. The sample’s representativeness was ensured by design, with statistical significance at 95%. In both Jordan and Lebanon, a random sample of WFP beneficiaries was selected and then allocated randomly into three different representative groups: (1) a voucher control group; (2) an unrestricted cash group; and (3) a choice group with the ongoing option to use vouchers, cash, or a combination of both.
In Lebanon data was collected at two predefined post-distribution monitoring (PDM) points spanning five months. In Jordan three PDM rounds were conducted spanning eight months. Each PDM included an extensive quantitative survey of each participating case plus qualitative analysis through focus groups.
Use of cash enabled food security that is higher than or equal to that made possible by vouchers
Cash produced food security results superior or equal to those of vouchers. Measured using WFP’s standard Consolidated Approach for Reporting Indicators (CARI) methodology, food security outcomes were better with cash in three out of five PDM rounds and equal in the remaining two rounds. This trend was seen consistently while using multiple food security indicators, including the food security index (FSI), Food Consumption Score (FCS), and dietary diversity index. In Lebanon, the average FCS was significantly higher in the cash group than in the voucher group, as was the percentage of beneficiaries having an “acceptable” FCS.
Similar to Lebanon’s, Jordan’s first round of monitoring (PDM1) showed the cash group to have a higher average FCS than the voucher group. However, the second monitoring round in Jordan (PDM2 in May) saw a significant increase in FCS in both the cash and the voucher groups, resulting in more than 93% of participating households having an acceptable FCS. This trend largely persisted until PDM3 in October. Since the considerable FCS increase in Jordan since PDM2, food security results were statistically similar between the cash and voucher groups. Dietary diversity and nutrition results echoed FCS trends.
Modality did not affect beneficiary expenditure on food. On average, both the cash and voucher groups spent approximately twice the WFP assistance value on food. This trend was consistent throughout the study. Thus, giving assistance as unrestricted cash did not reduce total food expenditure. It is important to note that both groups spent the total value of the WFP assistance on food—that is, the cash group did not reduce its spending on food despite having the flexibility to spend on other needs.
The cash group’s improved food consumption outcomes were not achieved through increased reliance on coping strategies. Results demonstrated that both groups used consumption-based and livelihood coping strategies equally. Cash buyers did not buy more food on credit than voucher buyers did, nor did they rely on less expensive or less preferred food.
Cash advantage augmented in more challenging contexts
Over the course of the five PDMs, cash proved particularly advantageous when food security was lower. When the context improved, cash performed as well as vouchers but its additional benefit to food security became less clear because of a ceiling effect on food quantity. Once the quantity ceiling was reached, beneficiaries used the cash advantage to buy what they perceived to be better quality food and took preferences into greater consideration. It is notable that cash did not perform worse than vouchers in any of the five PDMs. Therefore, cash offered a normalizing, shock-absorber effect that helped beneficiaries cope better with contextual changes.
Modality greatly affects shopping channels, with cash group shopping mostly outside of WFP network
Although voucher recipients spent about 60% of their total food expenditure in WFP shops, cash beneficiaries spent only around 20% there. Two key factors shaped this behavior: cost saving and convenience. Lifting the restriction on the chosen retail channel (through unrestricted cash) allowed beneficiaries to hunt for bargains and take transportation costs and convenience into account. Typically, in this context, unrestricted cash raised purchasing power by 15% to 20% over that of vouchers restricted to WFP shops. The cash group used the greater purchasing power of cash, without a reduction in total expenditure, to boost the quantity (or the perceived quality or both) of food purchased. Hence, rational optimization on the free market enabled the better food security outcomes for this group.
Similar spending on non-food basic needs, and comparable results
Beneficiary households in both Jordan and Lebanon typically spend 38% to 45% on food, 24% to 30% on rent, and 30% to 33% on other non-food items. In both Jordan and Lebanon, spending patterns in the voucher and cash groups were similar and not impacted by modality. As food and accommodation represent the most important basic needs, and food spending exceeds WFP transfer value by nearly 100%, spending on other categories is expected—and observed—to be unaffected by WFP modality. The modality switch did not change spending behavior, not even on temptation goods (for example, tobacco). These trends remained consistently similar over time (up to eight months in Jordan). Consequently, switching WFP modality over the course of the study did not influence beneficiaries’ fulfillment of basic needs (including access to housing, health care, and education).
Cash does not harm
At the experimental scale and over the study period of eight months, use of cash did not show any disadvantage over the use of vouchers. Cash beneficiaries did not face greater debt levels, repayment demands or acceleration, or greater incidence of theft or mistreatment.
The modality switch from vouchers to cash did not precipitate household disagreements or harmful dynamics over time in either Jordan or Lebanon. Women continued to hold considerable decision-making power in the household. Women living in male-headed households confirmed the findings, and there were no reports that the modality change diminished their role in the household. Women continued to make decisions about food spending in 60% to 70% of households. No differences in household dynamics were seen between the voucher and cash group except in PDM3 in Jordan, where the percentage of households with women as the only decision maker on food spending decreased in the cash group from 67% in May to 58% in October, whereas it was 65% in the voucher group.
Beneficiaries strongly prefer cash to vouchers
In both Jordan and Lebanon, more than 75% of households favored cash assistance and only 15% to 20% favored vouchers. Both genders reported a strong preference for cash. These trends persisted throughout the study period and were seen in all PDMs. Preference for cash was even stronger among those already in the cash group (roughly 90%); personal experience clearly drove the preference.
The behavior of beneficiaries in the choice group supports these results: more than 70% of them chose to access their assistance as cash, whereas only about 20% chose vouchers. Mixed usage of both cash and vouchers was about 10%.
Beneficiaries cited three key reasons for preferring cash to vouchers: higher purchasing power, flexibility and the capacity to manage cash flow, and dignity and empowerment. Reasons for preferring vouchers over cash centered on logistics related to ATM location, and on the perception that the food-restricted nature of vouchers helps people discipline their spending on food.
2.2. Cost-effectiveness of switching the voucher program to cash
Cost-effectiveness was calculated using an adapted Omega+ methodology to compare the relative cost-benefit of the cash and voucher modalities.
A move to cash assistance would imply a change of business model. Under the e-voucher program, the partner bank generates revenues from the contracted retailers, who pay a transaction fee to the bank whenever a WFP beneficiary uses the e-card. Whereas the WFP shops possibly pass some of these bank fees to the beneficiaries through higher prices, changing to cash entails WFP absorbing the fees directly. If a cash program was implemented for all beneficiaries under the existing agreements with the banks (as of May 2016), WFP would be expected to pay up to $1.5 million in Jordan and up to $4.6 million in Lebanon. This analysis did not consider set-up costs, as capabilities were already in place. Expected cost savings from the voucher-to-cash switch are relatively small. Hence, running an unrestricted cash program would raise WFP total costs by 0.8% in Jordan and 1.6% in Lebanon. However, this increase is driven by bank fees negotiated in the context of a low volume of cash withdrawals. In the meantime, bank fees in both countries have been renegotiated, with positive implications for the cash business case.
Still, the cash option is conceivably more cost-effective (Omega value of 0.95 in Jordan, 0.93 in Lebanon). Despite the higher costs for WFP, the FCS has the potential to be significantly higher in the cash group (+6% in Jordan, +8% in Lebanon; PDM1). Although this dimension by itself does not fully tip the scale toward one modality or the other, from an operational cost perspective there are no red flags to argue against cash.
Excursus: Effect on the local economy
The macroeconomic impact of changing the assistance-delivery mechanism was not studied experimentally. It is clear, however, that switching the modality from vouchers to cash will likely result in some shifts in the food products purchased.
Under both modalities, WFP assistance is spent entirely on food. Spending patterns— or, more broadly, income allocation—are relatively similar when beneficiaries move from vouchers to cash. A switch to cash would redistribute some spending away from WFP shops. On a very large scale, this could have either negative effects on the local economy (for example, lower tax collection due to purchases in informal retail channels) or positive effects (for example, the purchase of perishable locally produced food commodities, or generation of income for the local host communities rather than large retail chains). However, given that only a fraction of WFP assistance would be spent differently, the net effect is not expected to be important.
By extension, this means that if WFP shifts to a full-scale cash program, the impact on the local economy is not expected to change significantly.
This study found the delivery of food assistance in the form of unrestricted cash to be cost-effective. The benefit of cash over the food-restricted value voucher was particularly pronounced when food security was low. This feature can increase beneficiary resilience in the face of some external shocks. Although study participants were Syrian refugees living across Jordan and Lebanon, our findings may well be applicable to comparable contexts where refugee populations are familiar with a cash economy and live in host country settings with relatively large and functional markets. Since delivery modalities are most effective when tailored to the context, these findings may not necessarily apply in a very different refugeehost combination or in locations with limited market functionality, failed states and situations where there are food shortages.
We recommend considering unrestricted cash as an effective modality to deliver food assistance, especially at the outset of an assistance program in similar contexts. In the specific context of Jordan and Lebanon, e-voucher programs are well established and set-up costs have already been incurred. Additionally, ATM network coverage and unrestricted debit-card payment facilities are still limited in some localities in both countries. In such scenarios, assistance could be optimally delivered through the modality of choice, whereby beneficiaries can freely choose their assistance as unrestricted cash, value vouchers, or a mix of both. Rational optimization by individual beneficiaries is expected to result in better overall outcomes, as we observed in this study.