Home-grown school feeding: promoting local production systems diversification through nutrition sensitive agriculture.

The consumption of some non-staple crops such as legumes and dark, green leafy vegetables can address common deficiencies in key nutrients such as vitamin A and iron; however, limited markets and supply chain development impede their production and accessibility to consumers. This study investigates the pathways to promote agricultural production and dietary diversity for a local market intervention called Home-Grown School Feeding (HGSF). School feeding menus from 24 districts across 10 regions in Ghana during the 2014-15 school year were analysed in terms of food groups and several individual foods. The menus were then compared with food groups produced by households during the past year or consumed in the past seven days using data collected from a household survey. Greater inter-food group diversity in the menus was associated with higher production levels for tubers and dark, leafy green vegetables in the South and cereals in the North. A correspondence between the frequency in which a food group appeared in a menu and the share of households who consumed foods from the food group was also noted. Key issues, such as optimizing supply chains, enabling farm linkages and supporting diverse nutrient rich food groups, that underlie the success of Home-Grown School Feeding and other agricultural policies with similar goals of promoting production and dietary diversity are highlighted through commodity specific examples. The findings of this study may help strengthen operational linkages between agriculture production and nutrition for HGSF and other similar interventions.

Introduction

Evidence suggests that the mono-culturization of agriculture production towards staple foods-based systems in developing countries has crowded out the production of other food groups such as legumes and vegetables (Kataki 2002; Pingali 2015). This may have serious implications for the availability and consumption of non-staple food groups and thereby common deficiencies in nutrients such as protein, iron and vitamin A (Negin et al. 2009; DeClerck et al. 2011). Policies that promote diversity of food supply may help to rectify the imbalance in such food systems (Hawkes 2007; Remans et al. 2014; Pingali 2015).

A growing literature highlights the pathways among agricultural production, dietary diversity and nutritional outcomes which are multiple and context-specific (Jones et al. 2014; Fanzo et al. 2013). Gender is one critical pathway as women exhibit a high level of participation in the production and marketing of some non-staple, nutrient-rich crops, including neglected and underutilised species (Mayes et al. 2011; Malapit and Quisumbing 2015). Women are also more likely to undertake regular management activities such as weeding, cleaning and grading which are more common for vegetables (Joshi et al. 2006). Other studies have found that smallholder farmers are more likely to be engaged in production and marketing of non-staple foods including vegetables and legumes (Joshi et al. 2006; IFAD 2014). However, significant challenges adversely affect market conditions and integration for non-staple food production (Pingali et al. 2005; Barett 2008). These include high transaction costs linking production with value chains, limited incentives for farmers in terms of prices (Joshi et al. 2003), and the perishable nature of some non-staple food groups such as fruits and vegetables (FAO 1981; McKee 2012).

Realigning macro-level agricultural policies, such as those linked to output and input support, is important to create a more balanced incentive structure. However, this alone is not sufficient. Over time, the bias favouring the production of staple crops has, as cause and effect, deeply influenced household level production and consumption behaviour with implications for diets and nutrition (Ecker and Quaim 2011). Diversifying production to non-staple foods would require a systematic and sustained localized approach to incrementally build viable market conditions. In community-based peasant farming which is predominant in most parts of sub-Saharan Africa, market systems are primarily atomistic based around a few communities. Thus, conventional broad based market support instruments which are premised on well-integrated markets, and incentive responses designed around staples are likely to have limited impact. Localized market interventions can potentially address some of these challenges through strengthened commodity-specific value chains. One such intervention is ‘Home Grown School Feeding’ (HGSF) through which locally-sourced meals are provided daily to children attending schools. In 2014, at least 47 countries in sub-Saharan Africa were implementing school feeding programmes, of which at least 20 were HGSF or similar models.

Key principles of HGSF include local food procurement, smallholder engagement, nutrient-rich and diverse foods, and regularity in meal provision (Gelli et al. 2010). When appropriately designed, school meal menus can help ensure an effective application of these principles. Several studies have illustrated the potential impact of HGSF to create demand for cereals and other staple foods, leading to increased farmer incomes and improved livelihoods (Masset and Gelli 2013; Sumberg and Sabates-Wheeler 2011; Gelli et al. 2010). However, the potential for HGSF to support diversified food systems has been less explored.

This study investigates how HGSF may support diversified food systems in the context of Ghana. The Ghana School Feeding Programme (GSFP) provided daily meals to an estimated 1.7 million children, or about one out of every three primary school children in 216 districts of the country during the 2014–15 school year (GSFP Secretariat 2015). Meals were based on weekly menus that districts develop each school year and that may be adapted to the local context in terms of agricultural production, food culture and preferences (Parish and Gelli 2015). Specifically, this study presents a framework, relating school meals to local production and consumption patterns, and applies it to a set of 24 districts located across Ghana. The investigation considers food groups as well as some specific commodities to illustrate how HGSF may support diversified food systems through different pathways including women’s empowerment, market integration and supply chain management.

Materials and methods

The analysis draws from two sources of data. The first source was household survey data collected from the same 24 districts that included information on household dietary consumption and agricultural production. These data were collected as part of a baseline survey for an impact evaluation of the GSFP (Gelli et al. 2016). The impact evaluation sought to assess the impact of the programme on a wide range of child and community outcomes including education, nutrition and agriculture (Gelli et al. 2016). In total, the evaluation focused on a set of 60 districts in Ghana out of a total of 216 districts. HGSF was introduced in 30 of these districts while the other 30 districts served as a control group. Two comparable public primary schools and the surrounding communities were selected from each district.[1] Household listings were compiled in each enumeration area (EA) by the survey team supervisors assisted by community leaders. The list of all households with a child aged five to 17 years of age in each EA constituted the sampling frame from which participating households were selected at random for the household questionnaire. About 20 to 25 households were selected from each community for the survey. The questionnaires were administered by teams of enumerators from the University of Ghana in the local language. The responses were input and cleaned at the University of Ghana.

The second source of data was a set of school feeding menus from 24 districts from the HGSF arm of the impact evaluation. Menus were developed using a tool known as the School Meals Planner (Partnership for Child Development 2014; Fernandes et al. 2016). The tool supports the design of menus that link dishes with the types and quantities (in grams) of each individual food. Menus were used in district schools that participated in the Ghana School Feeding Programme (GSFP) during the 2014–15 school year. But menus for the remaining districts in the home-grown school feeding arm of the impact evaluation could not be obtained for several reasons. For example, the school sampled for the evaluation may not have offered school feeding or a menu was not developed for the district with the School Meals Planner tool. In the HGSF context ‘local’ can be defined in different ways depending on the context (Sumberg and Sabates-Wheeler 2011). For the analysis presented in this study, ‘local’ is defined as the district according to governance structure in Ghana. For each menu, the number of individual foods from each food group, as well as the quantity of each food group in grams was calculated. Inter-food group diversity was defined as the number of food groups. Intra-food group diversity was defined as the number of unique individual foods in a food group.

Figure 1 illustrates the distribution of the 24 study districts across the ten regions of Ghana. The country encompasses six distinct agro-ecological zones in addition to more than 100 ethnic groups, reflecting diverse food cultures and agricultural production practices. Northern Ghana, which includes the three regions of Upper West, Northern and Upper East in the north, is typically distinguished from Southern Ghana, which includes the remaining seven regions of the country. Poverty and food insecurity are concentrated in Northern Ghana, while the national capital Accra is located in Southern Ghana (GSS 2012). Thirteen of the 24 study districts were located in Northern Ghana.

Fig. 1

Distribution of agro-ecological zones and location of the 24 study districts. Notes: Agro-ecological zones are delineated by the dotted lines and include: Rainforest, Coastal savannah, Deciduous forest, Transitional zone, Guinea savannah, and Savannah. The red dots indicate the districts represented in the study. The map is based on a figure from Kemausuor et al. 2013

GSFP menus

Individual foods in the menus were classified into the following five food groups: 1) cereals; 2) tubers; 3) legumes (including nuts and seeds); 4) dark, green leafy vegetables and 5) other vegetables (Kennedy et al. 2013). Dark, green leafy vegetables, including individual foods such as cassava and cocoyam leaves, were defined as a separate food group apart from vegetables, given their nutritional value, in particular non-haem iron and vitamin A, as well as their prominence in the Ghanaian diet. Fruits were not analysed due to their minimal inclusion in the menus. Common cereals included rice and maize, while cassava and cocoyam were tubers that featured prominently. Legumes that appeared frequently in the menus included cowpeas,[2] groundnut[3] and melon seeds, while tomatoes and okra were common ‘other vegetables’.

Household survey data

Survey data were collected from a sample of households in the same 24 districts between June and August 2013 (Gelli et al. 2016). The data included a total of 1009 households who reported their dietary consumption in the past seven days. Of these, 621 were farming households who reported undertaking any agricultural production in the past year. Survey enumerators asked households to specify which individual foods they produced and consumed using a predefined list. There were differences among the individual foods included in the production and consumption list. The agricultural production list included 42 individual foods as compared to 59 individual foods for household food consumption. The responses were classified into the same five food groups defined for the menu analysis in Section 2.1. Households could also provide open responses for both the production and consumption questions. These responses were reviewed and classified into the five food groups when applicable. In most cases the individual foods were classified into one food group. However, cassava and cocoyam were included in both the tubers and dark, green leafy vegetables food groups even though they may have only been cultivated for the tuber.

Calculation

A descriptive analysis by food groups and the selected individual food items was then undertaken for the menu and household survey data from the set of districts in Ghana. The share of households in the district that were producing or consuming foods from each food groups was calculated. Separate indicators were constructed for four individual, non-staple foods that were commonly featured in the menus and that were also included in both the agricultural production and diet modules of the household survey. These individual food items were groundnut and cowpeas (classified in the legumes food group), okra and tomatoes (both classified in the other vegetables food group). The statistical significance of differences in production and consumption between Northern and Southern Ghana were assessed for analyses using the household survey data, but not the menu data due to the limited sample size.

In addition, a simulation was undertaken to estimate the overall potential national market demand from the GSFP for each of the five food groups based on information from the 24 district menus. The simulation assumed that all schools participating in the GSFP followed a menu from one of the 24 study districts. GSFP schools located in regions that included only one of the study districts were assumed to follow the menu from that study district. In other regions with multiple study districts, the minimum and maximum per-child quantity of each food group from the set of menus provided the basis for lower and upper bound estimates used, respectively. These estimates for the weekly quantities of each food group per child were multiplied by the number of children enrolled in schools participating in the GSFP in that region, and the number of school weeks in the 2014–15 school year, which was provided by the head office for the programme (GFSP Secretariat 2015).

Results

Menu analysis

The menus included the names of dishes prepared each school day (Monday through Friday) for children attending the school as well as the individual foods and quantities needed for each dish. For example, in one district, children may receive bean stew and plain rice on Monday. The menu would list this dish as well as each of the individual foods in the dish such as beans, rice, onion and tomatoes as well as the quantity in grams per child. Caterers were instructed to purchase sufficient amounts of the individual foods every week based on the quantities provided in the menu and the number of children attending the school. In addition to purchasing the individual foods, the caterers prepared the meals every day for the school children. The meals were prepared on the school premises or brought to the school from another locale.

All 24 menus included all five of the food groups defined by the study. Figure 2a presents the intra-food group diversity of the menus overall, while differences between Northern and Southern Ghana are presented in Fig. 2b. The greatest intragroup diversity was observed for other vegetables. Eight menus had four to six individual foods from this group. Legumes had the next highest intra-group diversity. Two menus had four to six individual foods from this group. Most menus (88%) included two cereals, and one or two tubers. More than half of menus (58%) did not include dark, leafy green vegetables. More individual foods from the cereals and legumes food groups were included in menus from Northern Ghana, while more individual foods from the tubers, dark, leafy green vegetables and other vegetable food categories were included in menus from Southern Ghana.

Fig. 2

a Intra-food diversity in menus, overall. Notes: Sample was 24 district menus from Ghana. The definition of food groups follows Kennedy et al. 2013. b Inter-food diversity in menus, by geographic region. Notes: Sample was 24 district menus from Ghana. The definition of food groups follows Kennedy et al. 2013

Household survey analysis

Table 1 presents findings from the household survey analysis regarding agricultural production. Cereals were the most commonly grown with 88% of farming households producing at least one individual food in the food group. Other vegetables were the least commonly reported (10%). Cereals and legumes were more likely to be produced in Northern Ghana (p < 0.001). Tubers and dark, green leafy vegetables were more commonly grown in Southern Ghana (p <0.001).

Table 1

Share of households producing individual foods in the five food groups during the past year, by geographic region

Food group:	Overall (n = 630)		North[a] (n = 452)		South[b] (n = 178)
	Mean	SD	Mean	SD	Mean	SD
Cereals	0.88	0.32	0.95	0.21	0.70	0.46 ***
Tubers	0.41	0.49	0.26	0.44	0.77	0.42 ***
Legumes	0.49	0.50	0.64	0.48	0.12	0.33 ***
Dark, green leafy vegetables	0.32	0.47	0.17	0.38	0.71	0.46 ***
Other vegetables	0.10	0.30	0.10	0.30	0.10	0.30

Sample limited to farming households in the 24 study districts of Ghana. Survey data collected in Summer 2013. Food groups defined as per Kennedy (2013). SD = standard deviation. T-tests used to assess North/South differences: * p <0.05, **,

Includes Northern, Upper East and Upper West regions

All other seven regions of Ghana

Sample limited to farming households in the 24 study districts of Ghana. Survey data collected in Summer 2013. Food groups defined as per Kennedy (2013). SD = standard deviation. T-tests used to assess North/South differences: * p <0.05, **,