Multi-Component Cash Transfer Programs Evidence from Mali’s Social Safety Net Program (Jigisémèjiri)

Despite falling rates of poverty and child undernutrition in Africa over the last two decades, the absolute number of people living in poverty and the absolute number of undernourished children continue to rise due to population growth (Beegle et al., 2018; Black et al., 2013). Global evidence suggests that cash transfer programs can reduce poverty and food insecurity and can build resilience for the poor. When cash transfer programs are combined with nutrition interventions, they also have the potential to accelerate improvements in child nutrition, especially when targeted to the critical window of opportunity for nutrition, the first one thousand days of a child’s life (Ruel et al., 2013). In West Africa, many cash transfer programs are combined with accompanying measures such as promotion sessions that aim to improve knowledge and increase adoption of recommended behaviors—including those related to child nutrition (Beegle et al., 2018, see Box 1). However, the extent to which such multicomponent programs lead to changes in behavior and improvements in outcomes related to children’s nutrition and health is still not well-understood.


Multi-Component Cash Transfer Programs
Evidence from Mali's Social Safety Net Program (Jigisémèjiri) Melissa Hidrobo, Lieven Huybregts, Naureen Karachiwalla, and Shalini Roy Background Despite falling rates of poverty and child undernutrition in Africa over the last two decades, the absolute number of people living in poverty and the absolute number of undernourished children continue to rise due to population growth (Beegle et al., 2018;Black et al., 2013). Global evidence suggests that cash transfer programs can reduce poverty and food insecurity and can build resilience for the poor. When cash transfer programs are combined with nutrition interventions, they also have the potential to accelerate improvements in child nutrition, especially when targeted to the critical window of opportunity for nutrition, the first one thousand days of a child's life (Ruel et al., 2013). In West Africa, many cash transfer programs are combined with accompanying measures such as promotion sessions that aim to improve knowledge and increase adoption of recommended behaviors-including those related to child nutrition (Beegle et al., 2018, see Box 1). However, the extent to which such multicomponent programs lead to changes in behavior and improvements in outcomes related to children's nutrition and health is still not well-understood.

Mali's Jigisémèjiri Program and the impact evaluation
The Government of Mali's social safety net program (Jigisémèjiri), initiated in 2013, aims to reduce poverty and improve human capital accumulation through targeted cash transfers (CT), accompanying PROJECT NOTE JANUARY 2020 Box 1: Examples of West African countries that combine cash transfers with promotion sessions  Togo measures (AM), and preventive nutrition packages (PNP). The CT consisted of unconditional transfers of 30,000 FCFA paid every three months to the head of the household. For the AM, nongovernmental organizations (NGOs) organized communication activities on priority themes defined by national authorities covering social protection, education, maternal and child health, and nutrition including infant and young child feeding. The PNP distributions, provided by the World Food Program, consisted of two types of fortified flour (Supercereals CSB+ and CSB++) for children 6-59 months old and pregnant and lactating women.
In order to obtain a rigorous and independent assessment of the impacts of the program, an impact evaluation was conducted by the International Food Policy Research Institute (IFPRI) and the Institut de recherché pour le développement (IRD), in five regions of Mali (Sikasso, Koulikoro, Kayes, Mopti, Segou). Communes were randomly assigned to start receiving the CT and AM components either in 2014-2015 (early treatment) or in 2016-2017 (late treatment). The two components were implemented for three years in both the early treatment and the late treatment communes. Then, within a subset of the early treatment communes deemed more nutritionally vulnerable, villages were randomly assigned to either receive the PNP component for one year or not to receive it (see Figure 1).
The study included three rounds of data collection: baseline (before the start of program activities), midline (before the late treatment group began receiving the program components), and endline (after approximately one year of implementation of the distribution of PNP). The two-stage randomized design meant that, at midline, the early and late treatment groups could be compared in order to rigorously estimate the combined impact of the CT and AM components; at endline, villages receiving PNP and those not receiving PNP within the nutritionally vulnerable communes could be compared to rigorously estimate the impact of PNP.

Figure 1: Project Timeline
Two samples were used to evaluate the impact of the program: i. A panel sample that consisted of households with a child 6-23 months of age at baseline that were revisited at midline and at endline. This sample served primarily to assess impacts on household-level outcomes such as food security and asset holdings; ii.
A repeated cross-section sample that included the households with a child 6-23 months of age at baseline, as well as new draws of households with children 6-23 months at midline and at endline. This sample served primarily to assess impacts on child nutrition outcomes within the window of opportunity of the first thousand days.
Two additional analyses were also conducted. A process evaluation was undertaken in April 2017 to better understand "how" and "why" the program did or did not achieve intended impacts. Furthermore, to complement the analyses of household-and child-level impacts, a separate analysis was conducted on impacts on relationships between spouses, specifically in terms of intimate partner violence.

Program delivery and take-up
The CT component of the program was implemented largely as designed. The household head-almost always male-was generally the main beneficiary on the beneficiary card, as was intended. The household head was also the person who most frequently collected the transfer and was the main decisionmaker about how to use the transfer. Most households reported spending the entire the transfer amount within a week of receiving it, even though transfers were distributed every three months and were intended to cover those three months. Beneficiaries largely reported using the CT as the program intended: for food, health, livestock, and education-related expenses.
The AM sessions of the program were implemented on a regular basis. Adherence to the AM component was moderately high but weaker than the CT component, likely because it was not mandatory to attend the AM sessions. At endline, between 62-70 percent of households reported having attended at least one AM session. One person usually attended these sessions, most often the head of household, although other household members were slightly more likely to attend among households in the late treatment group as well as on days when no distribution of transfers was organized. However, there were significant barriers to attendance such as lack of information or awareness about the date, time, and location of sessions or who should attend, particularly among women, which led to lower participation of women in AM sessions.
The PNP component was implemented as designed and was well-accepted by beneficiaries. The distribution of PNP was accompanied by a cooking demonstration. Given that the theme of child feeding was carried out during the AM sessions, the PNP component was limited to the distribution of PNP and awareness-raising on the proper use and conservation of the fortified flour.

Impacts of CT and AM on consumption, assets, and resilience of households
Comparing early and late treatment groups at midline, the CT+AM intervention led to significant improvements in household welfare. It increased the proportion of households classified as food secure or having little to no hunger, improved household-level dietary diversity, and improved diet quality. The CT + AM intervention significantly increased the probability of a household purchasing cereals and maize in bulk.
Although we find no impacts on the value of food consumption, the Jigisémèjiri program reduced determinants of poverty and increased resilience by strengthening households' savings and asset base. Early treatment households had higher savings and higher asset values at midline. The increases in asset holdings emerged particularly in livestock and transport assets, which are productive assets that evidence shows can generate new income, serve as stores of wealth or collateral, and can be sold for income if needed (Moser, 1998).

Impacts of CT and AM on child-level and caregiver outcomes
The combined CT and AM intervention at midline had no effect on maternal knowledge of child nutrition and health. However, there were two impacts on infant and child nutrition practices that included improvements in the timely introduction of semi-solid foods and the prevalence of a minimally acceptable diet, but no impacts on children's dietary diversity or breastfeeding practices. This is perhaps explained by the weaker than anticipated participation of mothers and caregivers to the AM sessions. The intervention led to a small reduction in the prevalence of diarrhea in children 6-23 months but no other reduction in child morbidity. An impact of the combined CT and AM intervention on child morbidity is more likely to occur if the availability and access to health services and the quality of care are guaranteed, aspects which were not directly targeted by the intervention. The small impact on child diarrhea is not associated with improvements in caregiver treatment practices in the case of child diarrhea, in terms of dietary mitigation strategies, use of oral rehydration solution or zinc, or provision of more liquids and foods. The combined CT and AM intervention also did not result in any impact on child nutritional status and growth. Child anemia, stunting, and wasting remained very high at midline.

Impact of PNP on children's and caregivers' nutritional status
The PNP intervention improved children's nutritional status. Children had improved length-for-age zscores, which reflects a reduction in chronic child undernutrition. The PNP intervention also reduced severe anemia. Child morbidity and breastfeeding and complementary feeding practices were not affected, nor was child dietary diversity, implying that consuming the PNP flour did not lead to a less diverse diet by replacing, for instance, noncereal food groups. Blanket feeding programs-such as the PNP intervention-can be effective in reducing undernutrition, but also carry the risk of providing additional calories to subgroups of the population who do not need it. The PNP intervention led to higher rates of caregiver overweight (Body Mass index ≥ 25 kg.m -2 [BMI]); however, no meaningful impact of PNP was found on the proportion of caregivers who were underweight (BMI< 18 kg.m -2 ).

Impact of Jigisémèjiri on women's well-being
Jigisémèjiri had no objectives explicitly related to gender. Most women did not report that the program changed their responsibilities or how they saw themselves. However, both women and men reported being happier as a result of the CT component and reported learning new things as a result of the AM component. Women also experienced less intimate partner violence as a result of the program, but only in polygamous households. Evidence on mechanisms for reduced violence suggests that the program led to significant decreases in men's stress and anxiety among polygamous households, and larger reductions in disputes in polygamous households compared with monogamous households. There were no impacts overall in measures of women's empowerment.

Conclusion and Recommendations
 The objective of the Jigisémèjiri program to reduce the vulnerability of poor households has been met. However, this study reinforces lessons from other studies in concluding that cash transfers alone or combined with relatively low-intensity accompanying measures have limited impacts on child nutrition.
 If an objective of the program is to create more "continuous" increases in food consumption, a potential option would be to make the transfers more frequent rather than quarterly: e.g., monthly or bi-monthly such as many other transfer programs globally.
 It may be worth reconsidering targeting the CT exclusively to the household head. At minimum, it may be useful to target both the household head and spouse and encourage household heads to engage other household members, such as their spouses. This would allow women to feel more involved, reduce confusion about who should attend the AM sessions, and potentially improve women's selfefficacy and decision-making power.
 Reducing barriers to attending the AM sessions-for example, by improving communication about the timing and location of each session to all household members (e.g., including spouses and other members of the family) could increase participation, particularly among women. Combining the AM session with the PNP as an incentive for caregivers to attend could also make the AM component more effective and strengthen impacts on child-level outcomes.
 The AM component aimed at improving child nutrition could be further strengthened to a behavior change intervention using multiple delivery channels including mass communication, interpersonal and group counselling, and community mobilization. Evidence from other studies suggests that behavior change communication interventions-implemented at high coverage-are effective in improving caregiver nutrition knowledge and improve child feeding practices.