School feeding interventions linked to small-holder agriculture can have multiple goals in the following areas:
•Food security: supporting incomes of recipient households (those consuming food) and farmer households (those providing the food)
•Education: increasing school enrolment, attendance and reducing drop-out, and improving cognition and learning achievement
•Health: improving nutritional status of school age children
The impact of the intervention in each of the above areas occurs through a number of complex pathways. This section describes the pathways through which the programme is expected to operate. Though the evaluation of all potential effects of the intervention is beyond the scope of this study, a subset of outcome indicators for the evaluation will be selected based on: a) a knowledge gap specific to school feeding in Mali; b) knowledge gaps in the school feeding literature in general; c) feedback from peer reviewers and technical partners; and d) budget constraints.
illustrates in very broad terms the impact theory of school feeding on food security, education and health. School feeding affects educational outcomes directly by increasing enrolment, attendance and completion (line ‘a’ in the figure). It affects health directly by improving nutritional status (line ‘b’), this in turn has an indirect impact on education, as improving nutritional status has a positive impact on learning outcomes (line ‘d’). The intervention can also affect income directly by increasing households’ food security (line ‘c’). Finally, there are effects running through increased income, health and nutrition, and vice versa, as richer families are investing more in human capital and more educated and healthier adults are more economically productive. However, these latter effects only occur in the long term and certainly not before children have left school; therefore, we will not discuss them in the following design.
Overall programme theory of school feeding interventions.
It must be emphasised that the ability of the school feeding intervention to deliver the effects depicted in Figure
critically depends on the appropriate implementation of the programme. The management and implementation of the intervention involves several actors and scoping visits. A preliminary study undertaken by PCD [9
] has shown that in Mali there are several problems of communication, supervision and monitoring among these different stakeholders. Programme success will also depend on the ability of communities to actively engage in the programme and in the strengthening of the public institutions involved. The issues of social accountability, ‘good governance’ and the links between accountability and programme effectiveness are important areas that this impact evaluation will explore in more detail.
Impact on food security and smallholder agriculture
The intervention is designed to stimulate the economy at a community level by purchasing food from small-holder farmers. Food for the school feeding programme is currently purchased from traders by mayors of the Communes of intervention. Traders in turn purchase food from small-holder producers, though these do not need to be residents of the villages targeted by the project. The capacity building or HGSF+ component supported by PCD is intended to confer the community-based organisation more decision power and will increase their ability to purchase from individual farmers or farmers’ associations residing in the project villages. On the small-holder farmer production side the programme can have three main effects that are summarised in Table
and schematically in Figure
, including output effects, distributional effects and stabilisation effects. In addition to these effects, the programme can also have some wider effects on the local economy by generating employment.
Programme impact on smallholder farmers
Programme theory of school feeding and smallholder agriculture.
The programme introduces additional demand on the market. In Mali, the school feeding programme purchases food for schools at the Commune level, which includes several villages. Most Communes have food markets. Mayors purchase from traders in the commune. Traders in turn purchase food from, among other sources, small-holder farmers. The effect of the food purchases is a shift of the demand curve for food (say rice) to the right (movement from D0
), as illustrated in Figure
Potential impact of HGSF on commune food demand.
The size of the shift depends on the extent of the substitution effects and on the size of the market considered. Substitution occurs when households reduce the domestic consumption of food because children are fed in school. At one extreme, there is full substitution: families do not provide any additional food to children fed in school or any other household member in excess of the school food ration. Food provided in school entirely substitutes for food normally consumed in the home. The result is an increase in household savings and consumption of non-food items. In this case, there is no shift of the demand function (D0) and HGSF does not have price or output effects. There is still a distributional effect of the intervention if food is purchased from small-holders rather than large farmers. Profits of large farmers will in this case decrease whilst profits of smallholder farmers increase. Full substitution, however, is unlikely to occur. The largest substitution is likely to occur when households interpret the school food ration as a cash transfer. In this, theoretical, case the income equivalent of the ration is spent following income elasticities. Considering that the areas of intervention are very poor, assuming an income elasticity of food of 0.6 to 0.8 and a food share of 0.8, we understand that only about half of the HGSF transfer would eventually be spent on food and the food demand function shifts to D*. However, studies show that rarely do people interpret food transfers as cash transfers and that people tend to attach some preference to the food received and thus consuming food beyond what the income elasticities would suggest. Therefore, the final shift in the demand curve is likely to happen somewhere between the curves D* and D1.
The size of the demand shift also depends on the size of the market considered. If we are considering the national rice market, the shift is extremely small and the price and output effects are likely to be negligible [10
]. The effect is larger at the commune level if food is procured locally. However, some of the food may be purchased outside the commune and, therefore, the shift of the demand function at the commune level would be smaller than the one depicted in Figure
The impact of HGSF on output and prices at the commune level will depend on the slopes of the demand and supply functions (see Figure
). Welfare effects on producers and consumers can be calculated using changes in consumer and producer surpluses. We do not know the values of the supply and demand elasticities of food items, which need to be estimated econometrically. For convenience, we consider constant elasticities around the equilibrium point (a logarithmic form provides this type of elasticity ln(q
) = a
) + cln
Illustrative view of the commune food demand market and HGSF.
Consumer surplus is the area between the market equilibrium price line (Pe) and the demand curve (D0). Producer surplus is the area between the market equilibrium price line (Pe) and the supply curve (S). Producers’ and consumers’ surpluses can be calculated provided we know the initial equilibrium price level (Pe), the quantities of food produced and consumed at this price, and the own price elasticity of food demand (the shape of the demand and supply curves).
Two extreme cases of small and large supply elasticities are shown in Figure
. The size of supply elasticity will depend on three main factors, such as yield risk, market failures and rigidity of fixed factors. High income risk and missing markets are likely to be present in the communes of intervention, thus reducing the size of supply elasticity. Farmers are not likely to respond promptly to price changes. In addition, while farmers may be able to vary the amount of variable inputs used (labour and fertiliser, for example), they might not be able to change the amount of fixed input in the short run (such as, equipment, land and livestock).
Potential impacts of HGSF with small and large supply elasticities.
The two supply curves in Figure
can be seen as short- and long-term supply curves or elastic and inelastic with respect to risk and market factors. In any case, they illustrate the differential impact of HGSF on prices and output.
•If farmers are not able to provide the additional food demanded by HGSF (small supply elasticity – curve Ss), then most of the effect of HGSF goes into prices and little impact on output. From a welfare perspective, producer surplus increases (farmers win), while consumer surplus may decrease (consumers may lose).
•If farmers provide any additional food demanded by using current input more intensively and by quickly changing use of fixed inputs (large supply elasticity – curve Sl), then HGSF would have a large impact on output and a negligible impact on prices. From a welfare perspective producer surplus increases (farmers win) and consumer surplus increases as well (consumers win). For the benefit of both producers and consumers, therefore, a high supply elasticity is needed.
In general, we should expect the supply elasticity to be between the two extremes depicted and, therefore, the programme to have an impact on both prices and output. The impact on prices depends on the level of spatial market integration. In principle, if markets are efficient, prices for the same food items should be the same everywhere after an adjustment for transport costs. However, the literature on market integration suggests that transfer costs may create a wedge between prices at different locations, which allow prices in the two locations to vary in an uncorrelated way within a band [11
]. In other words, if transport costs for an isolated commune are very high, food prices may increase up to a point when trade between the two locations takes place and prices are equalised. There is, therefore, a real possibility that food prices increase at the commune level and that these price effects are transmitted to consumers.
Based on this theoretical model, the programme will have a positive impact on farmers’ income via an increase in prices and food quantities produced. The impact on consumers is less obvious. Depending on the size of the increase in prices, some households may have their welfare reduced as a result of the intervention. This observation suggests that the evaluation in addition to assessing impact on farmers’ income should also monitor price levels and model, by simulations, if not observations, their impact on consumers. Note also that the intervention will have other minor positive effects at the village level by creating additional employment (cooks, treasurers and stock keepers) and demand. This suggests that a micro-simulation at the village level should be conducted in order to assess the potential impact through general equilibrium effects.
The programme also has a distributional impact as it shifts demand from large to small farmers. As described in earlier sections, even if households are fully substituting the school meal, the programme generates a demand shift from large to small farmers. While small local farmers see an increase in their income, larger farmers suffer a reduction. This effect can be observed to the extent that the evaluation will be able to collect income data from a large number of farmers, with and without the programme, large and small.
Finally, the programme potentially reduces household risk. The programme can stabilise small farmers’ incomes by offering a stable demand and price. A number of positive effects follow from risk reduction, including an increase in expected utility, a reduction in the use of inefficient mitigating and coping strategies, (such as lower yielding crops and precautionary savings), and an increase in productive investments. This impact can be observed indirectly by observing farmers’ risk-mitigating and coping behaviours. However, it is quite possible that yield risk dominates price risk. In addition, whatever the price effects, these may take a long time before having an impact on farmers’ expectations. The programme’s impact on risk behaviour, therefore, is unlikely to be large.
It is less obvious that a stabilisation of income variability at the aggregate level is needed or is an obstacle to the implementation of the programme. According to an USDA report [6
], the success of HGSF initiatives in Mali is potentially compromised by the insufficiency and instability of food production in the aggregate, by the inability of vulnerable regions and areas to produce food in the desired quantity and at the desired time, and by the inability of farmers to respond to the incentives provided by the project. Our analysis of the Food and Agriculture Organisation (FAO) data on aggregate cereal production in Mali over the period 1961 to 2007 suggests that, at current growth rates, by the year 2017 agricultural production in the country will be sufficient to bring malnutrition below 5% of the total population (see Figures
and ). It is also clear that cereal production has not only dramatically increased from the mid-80s but has also become more stable, accompanied with reduction in undernutrition rates during the same period.
Cereal production in Mali 1960 to 2007; Source: Data from FAO.
Malnutrition in Mali from 1961 to 2007; Source: Calculated from FAO data.
The sufficiency and stability of the food supply in the aggregate may hide seasonal stress, regional differences and the presence of chronically poor groups. However, in each village, the pool of farmers from which the project will purchase food is rather small and the ultimate sellers will be traditionally surplus farmers who in the absence of the programme would sell to traders. Two other factors may facilitate a stable supply of food at the community level. First, the involvement of the community-based organisation in the management of the intervention will help the identification of farmers able to provide food at the desired time and in the desired amounts. Second, following a model adopted by the Purchase for Progress programme of the WFP, purchases may be organised through contracts with farmers’ associations rather than with individual farmers, thus increasing the likelihood of providing a stable supply.
Impact on education
We formulate hypotheses regarding the impact of school feeding on child schooling and learning starting from an economic model of parental educational choices in developing countries adapted from [12
below illustrates the determinants of schooling and learning. Schooling produces learning, which in turn has welfare effects. Schooling can be thought of as enrolment, attendance, drop-out or school completion. Learning is the acquiring of basic skills, such as language and mathematics. These skills are valued in the markets and educated children are expected to generate higher incomes and wages. In addition, more educated individuals may conduct healthier lives.
Figure 10 Programme theory of impact on education. The main determinants of schooling and learning are child characteristics, schooling costs, households’ characteristics and school quality. Children do not participate in schooling for different reasons; (more ...)
In this model, the main determinants of schooling and learning are child characteristics, schooling costs, households’ characteristics and school quality. Cognitive ability and motivation facilitate learning and encourage families to send children to school. Children do not participate in schooling for various reasons; at the household level, it is often a trade-off between the costs and benefits of schooling that determine whether a child will go to school or not. Costs are not only direct, such as school fees. For example, the opportunity cost of sending a child to school would mean foregoing the benefits of any work that the child could be doing instead of attending school. Often, the opportunity costs follow seasonal patterns, or increase with age, meaning that older children might need stronger incentives than younger children in order to stay in school. The opportunity costs of schooling may also be higher for girls - girls are often kept at home to look after siblings, help with other work, or simply for cultural reasons. These costs have a direct effect on schooling but should not affect learning. Household characteristics such as income and preferences (including attitudes towards education, and time discounting) affect schooling directly, while other characteristics may affect learning directly (for example, more educated parents may improve learning by helping children with their homework). School quality affects learning directly through the quality of the teaching, the teaching environment (supplies and facilities) and schooling, by affecting household schooling decisions.
An initial outcome that drives increased school participation is the incentive to households to send children to school. Generally, this incentive is achieved through an income transfer offsetting the financial and opportunity costs of schooling, and also through an enhancement of the services provided at school. School feeding may also have an incentive effect on pupils actually wanting to go to school to receive food rather than staying at home and missing out. In theory, both of these effects will contribute to shift short-term household decisions towards increased schooling. The specific effect of the incentive will very much depend on the context in which school feeding is operating. Conceptually, the health and nutrition improvements from school feeding can also reinforce the impact on education. Addressing micronutrient deficiencies, in particular iron and iodine, has been shown to have a positive impact on learning (see Section
Impact on nutrition), as has the systematic deworming of school-age children in areas of high prevalence of intestinal helminths [13
]. The income transfer incentive and the improved health and nutrition status resulting from school feeding service provision would then lead to improved access and learning outcomes.
Learning increases as a result of schooling. In addition, the same factors affecting parents’ decisions also affect learning directly. In this model, school feeding affects schooling and learning in two ways:
•School meals reduce financial and opportunity costs of schooling and, therefore, increase schooling directly, which in turn affects learning positively.
•School meals increase cognitive ability. This in turn increases child learning in school and affects parents’ schooling decisions (learning increases expected income and, therefore, parents’ interest in schooling).
Note that the reduction in school costs can be partially outweighed by additional programme costs. In particular, the programme may require community participation in two ways [1
]. First, communities are sometimes required to provide fire-wood for cooking and other items such as fresh fruit, vegetables and condiments [14
]. In addition, they are expected to provide cooks and storekeepers (though participants in these activities often receive compensation in the form of a daily meal). Second, the CGS may collect contributions from parents either in monetary form or in-kind. All these contributions increase the costs of schooling.5
There may also be feedbacks from increased schooling and learning to school quality. First, learning in school may increase because the average cognitive ability of pupils has increased (peer effect) or because teachers become more motivated to teach. Second, cost reduction may bring to school children of poorer background thus reducing the average cognitive ability and reducing the overall performance via the same peer effects. Third, schools may become overcrowded because of increased attendance, though the effects of crowded classrooms on learning are still unclear [15
Impact on nutrition
School feeding interventions can potentially have an impact on nutritional status of school children and their younger siblings, as summarised in Figure
Figure 11 Programme theory of impact on nutrition. The nutritional impact is mediated by the extent of food substitution effects within the household, and the use of the energy intake by the child and her siblings. The reduction in malnutrition via diet diversification (more ...)
a) The nutritional impact is mediated by the extent of food substitution effects within the household, and the use of the energy intake by the child and her siblings.
b) The reduction in malnutrition via diet diversification and the absorption of micronutrients in the body can have direct effects on cognition.
c) Better nourished children may be best positioned to learn while in class and outside class.
These three issues will now be discussed:
a. Substitution effects
Building on [17
], we provide a simplified programme theory of the nutritional impact of school feeding. The school meal can be shared by children with other household members or can substitute (at least partly) for food normally consumed in the home. This is obvious in the case of take-home-rations, whereby children take home a quantity of food on a regular basis, but also applies to any school feeding programme, because households may in principle use the school meal as a substitute for food normally consumed and spend the monetary equivalent otherwise. Evaluations of fortified biscuit programmes in Bangladesh and Indonesia found that gains in nutritional intake were not limited to the children actually receiving the biscuits at school. The two studies found significant evidence that school children shared the biscuits with their younger sisters or brothers at home. A recent randomised control trial (RCT) in Burkina Faso also found that take-home-ration programmes led to an improvement of the nutritional status of younger siblings in beneficiary households [19
]. In Uganda, an RCT also found significant improvements for pre-schooler siblings of children receiving school feeding [20
]. This provides emerging evidence of a spill-over effect and a window of opportunity to also affect children during a critical developmental stage when nutritional interventions can have the strongest impact.
Ingested foods contribute to three outcomes, of which physical growth is only one:
••Physical growth. Food can help physical growth in terms of height and weight. It is normally believed that catching-up by stunted children after the age of five is limited. However, food intake should increase storage of fat and, therefore, weight.
••Physical Activity Level (PAL). Energy intake is spent in work after school or in more activity and play.
••Basal Metabolic Rate (BMR). Energy is required to maintain the healthy functioning of the body while at rest.
Catch-up growth in children and adolescent may be possible though the process is slower than catch-up in weight and it is not certain up to what age it takes place (probably up to the end of the adolescent growth spurt) [21
]. All malnutrition indicators could change after the intervention (stunting, wasting and underweight), though the impact will depend on the extent of substitution effects and on whether children are increasing the use of energy for PAL and BMR. A child may have normal height and weight and still be undernourished because he does not expend enough energy in activity and play to maintain health and develop his cognitive abilities. Assessment of malnutrition should also measure PAL, particularly in adolescents who engage in considerable work and play. Unfortunately, there is no accepted theory, nor evidence, on whether children adapt to nutritional stress by reducing weight or PAL. There is also uncertainty on the definition of a minimum acceptable level of PAL (an arbitrary factor of 1.5 of BMR is often used by FAO, for example). Finally, there is no standardised way to measure PAL. Observation of behaviour in class and questionnaires for parents and teachers could be used to measure PAL.
Finally, highly deprived children may use additional energy intake from school meals simply to restore the original BMR. In addition, higher weight requires more energy; therefore, BMR is a function of body weight and the BMR requirement increases as weight increases.
Because of the complex pathways described in this section, we should not expect a strong impact of the programme on the nutritional status of children. However, we might expect an improvement in children’s activity and play and an improvement in nutritional status of siblings (if substitution effects are strong).
b. Diet diversification, micronutrients and cognition
Micronutrients may have a direct impact on cognitive abilities. It is not well understood how iron affects brain functioning and the central nervous system, but there is ample evidence that reduction in iron deficiency improves mental functions across all age groups [22
]. Iron interventions were found to have a positive impact on infant development scales, intelligence quotient (IQ) tests and school achievement.
c. Indirect effects of better nutrition on cognition
Restoration of micronutrient requirements and energy intake can also have an impact on attention and motivation. Energy intake [24
] and iron intake [22
] can have an impact on hyperactivity, withdrawal, nervousness, hostile behaviour and happiness. The emotional status of children affects the attention span and has other spill-over effects. The quality of teaching in class is likely to be affected as teacher may become more motivated and as the quality of students performance in class improves (think of the different impact on learning of improving attention of 10%, 50% or 100% of students in class).
Role of social accountability
The effectiveness of the programme could be considerably improved by improving the communication mechanisms between the actors involved, by strengthening the monitoring system, and by introducing elements of social accountability. On the institutional side, the introduction of a monitoring system and the creation of communication mechanisms between the different actors would likely have the effect of improving the programme performance (see Figure
). Similar effects could be expected from a stronger engagement of the CGSs. The delegation of certain responsibilities to the CGSs could increase motivation and awareness of the programme among beneficiaries while at the same time ensuring a better implementation of activities, (such as monitoring of public officers’ performance and food procurement from small holders in the community). These effects would have a positive impact on all the intermediate and final outcomes of the intervention.
Impact of institutional and community strengthening on programme performance.
Main hypotheses and outcome indicators
We summarise here the expected impact of the intervention on education, nutrition and social protection discussed in Section
Programme theory of the intervention.
•The intervention will have an impact on a small number of farmers in the intervention villages. Other persons in the village may benefit either directly or indirectly via an increase in income.
•The intervention will have a positive impact on enrolment, attendance and drop-out rates.
•The intervention will have an impact on cognitive abilities and class behaviour, including attention.
•The impact on learning (test scores) will be moderate as school quality is unlikely to change in the short term.
•The intervention will have a limited impact on physical growth of children because of the increase in PAL, substitution effects and the age range (6 to 17 years) of the targeted population. An impact on siblings of school-going children is possible if substitution effects are strong.
•The intervention will have a moderate impact on the diet because food purchases by communities and mayors do not follow nutritional guidelines and nutrition education is absent.
•The intervention will have little or no impact on micronutrient status as the food provision is not fortified and only moderate effects on diet diversity are expected.
•The overall impact of the programme will increase through the introduction of social accountability mechanisms and the strengthening of the monitoring and communication system.
includes a list of the main outcome indicators of the study. Section
Methods of analysis will describe how data will be collected using different survey instruments. All outcomes, including school attendance and test scores, will be obtained at the household level.
Main outcome indicators of the intervention
Note that in addition to outcome indicators we will also observe the programme impact on intermediate indicators, particularly for those outcomes that are more difficult to observe directly: income and social accountability. In the case of income, we will look at intermediate outcome, such as input use (labour, land, seeds and fertiliser), investments (farm capital such as tools and machinery), and market access (marketed surplus, prices and markets). In the case of social accountability, we will observe the impact of the programme on knowledge and practices of mayors and CGSs members as they result from the training activities. The quantity, quality and timely preparation and delivery of food in school will also be examined.