Nutrition incentives in dairy contract faming in northern Senegal

Summary of research¹

Contract-farming arrangements are agreements between farmers and buyers, usually for high-value foods, wherein farmers agree to produce a given product and buyers agree to purchase it. Contract farming has the potential to solve many constraints faced by small farmers related to lack of information, market opportunities and credit. This paper documents the impact of a contract-farming scheme in northern Senegal that seeks to introduce a preventive health product in agricultural contracts. In particular, this study tests whether providing a micronutrient-fortified food product can lead to a more regular supply of milk delivered to a dairy processing factory in northern Senegal.

Milk production in the northern part of Senegal is dominated by the Fulani, who have a long nomadic pastoralist history. Livestock provide the Fulani with a source of income through milk and meat production, a source of prestige and social status, and an informal insurance device through which they sell their cattle during negative-income shocks. Seasonality plays an important role in milk production in the Sahel due to its semi-arid climate. Pastoralists and herds move daily and seasonally in search of water and pasture. Senegal’s livestock sector contributes 37% of agricultural value added and 5.5% to national GDP (FAO, 2005). However, milk production cannot keep up with the growing demand for dairy products and two thirds of the demand for dairy products is met by imported milk powder.

Intervention

In 2006, the Laiterie du Berger (LDB), a Senegalese social business, began to collect milk from pastoralist families near the town of Richard Toll in the Senegal River Valley. LDB’s business model is to produce high-end products (mostly yogurts) for urban populations, based on fresh milk that it collects itself twice a day from households living within 50 km of its processing plant. The business has a social orientation and the long-term company objective is to contribute to generating additional milk-related earnings for mostly poor, semi-nomadic households in the area. In this setting, three potential factors that lead to irregular and costly milk delivery to the processing factory are: seasonality of milk production; credit constraints which lead to sub-optimal investment in inputs and risk coping strategies; and not directly compensating women for their labour. 

In order to reduce the variability in milk production and improve the nutritional status of children in supplier households, a nutrition incentive was offered to milk suppliers of LDB. The nutrition incentive was delivery of free, micronutrient-fortified yogurt (MNFY) for children if suppliers delivered a certain amount of milk to the LDB over a certain number of days. The daily sachets of MNFY were manufactured by the dairy company using the milk collected from the dairy farmers. The yogurt was mixed with cereal and fortified with 2.1 mg of EDTA iron to address malnutrition of the study population and in particular the high rates of anaemia, which were 89% for children 6-23 months and 79% for children 24-59 months at baseline (Hidrobo, Quinone, Le Port et al, 2013). 

The MNFY was delivered daily to milk collection points easily accessible to milk suppliers, and in particular women. A behavioural change campaign (BCC) was also conducted with all households in the study. Designed and implemented in partnership with the Cellule Contre la Malnutrition (a unit in the Prime Minister’s office coordinating all malnutrition-related interventions in the country), the BCC focused on optimal infant and young child nutrition, the importance of micronutrients, and the role that diverse diets and fortified food can play in delivering them, although the specific MNFY used in the study was not discussed (as advised by the ethics committee to avoid what could be considered product advertisement).

Research method

A cluster RCT was implemented to test whether this nutrition incentive improved children’s nutritional status and increased milk deliveries. A total of 320 concessions (groups of three to seven households, usually related to each other), across four existing milk routes (Dagana 1, Dagana 2, Mouda, and Rosso) were randomised via a public lottery to either an incentive group or control group. During public lottery events, all suppliers were invited to sign a container-level contract ahead of the random draw committing them to deliver 0.5 litre/lactating cow/day accounting for the lactating cows of all households routinely delivering milk to this container. Suppliers also declared the number of children aged two to five years living in their household and contributing container households. For the incentive group, fulfilment of the contract in a given week triggered daily home delivery of MNFY for each child aged two to five years in the household the following week. In the control group, no incentive or penalty was given.

The authors collected household-level data before the start of the intervention in January 2013 (baseline) and again after the intervention in January-February 2014 (endline). A household questionnaire was administered to the head of household or spouse and contained information on the composition and demographics of the household, dwelling characteristics, food security, assets and preferences of the head of household. A questionnaire was administered to all mothers with children aged 24-59 months at baseline and included information on nutrition knowledge; knowledge about the incentive design; involvement and decision-making power with respect to milk production; and usage of milk within the household. Anthropometric measures and haemoglobin were also collected for all children aged 24-59 months at baseline; the incentive had a significant impact on increasing children’s haemoglobin levels (full results will be available in a future publication). The total number of households included in the baseline survey was 445 (213 from the treatment arm and 232 from the control arm) and comprised all households that are milk producers for LDB and who agreed to be part of the study. Of these households, 98% or 437 were resurveyed at endline. Administrative data were collected by LDB and GRET (partner non-governmental organisation) on weekly milk delivery by container. This dataset includes information on the number of days the household delivered milk during that week and the quantity delivered each day, as well as the number of  sachets of MNFY it received in that specific week.

Results

Descriptive evidence on the evolution of milk production across treatment and control groups over time is presented in Figure 1 using four different outcome indicators: (1) whether the container delivered milk at least once in the past week; (2) whether the container fulfilled its contract that week; (3) the number of days it delivered milk that week; and (4) the total amount of milk it delivered that week.

These results show significant increases in the probability and frequency of delivery during the dry season, but only limited impacts on total amount of milk delivered. Baseline and endline household survey data support these results and link it to households postponing when lactating cows are sent on migration, thus resulting in more lactating cows staying near the home for milking and delivery. The incentive did not lead to increases in daily milking effort per cow; nor did it affect consumption patterns of milk within the household.

The authors also found considerable heterogeneity of impact. The effect was time-sensitive and mostly effective during the dry season, when households must decide when to go on seasonal migration and which cows to take. The impact on amount of milk delivered is also limited to those households where women hold the contractual relationship with the milk company and have greater control of milk-related decisions.  For this sub-sample the impact is large, with the incentive leading to a 42% increase in milk deliveries over the year. This effect is largely mediated by having twice as many lactating cows present in the area near the home at the height of the dry season.

Results show that the ability to fulfil the contract and the treatment effect were largely seasonal, with over 80% of containers fulfilling the contract in December 2012 and dropping drastically to approximately 30% at the height of the dry season in July 2013. The effect of the incentive follows a similar pattern, with a higher probability of contract fulfilment in treatment than in control group from February 2013 through June 2013. Once the rains began in August and the contracts were no longer binding (both treatment and control groups are easily able to fulfil the contracts), the treatment effect disappears.

Results also show that the percentage of active containers decreased from nearly 100% being active at the start of the study to less than 50% being active by the end of June. Here also, the incentive increased the probability of a container being active, but only until the first rains. The average number of days of milk deliveries in a week also decreases from approximately six before the start of the study to approximately two at the height of the dry season, with those in the treatment group delivering more days from January through July than the control group. Consistent with seasonal patterns, mean milk production before the study is approximately 20 litres, drops to approximately 10 litres at the height of the dry season, and then spikes to over 40 litres during the rainy season.

Discussion

Innovations that help overcome the inherent difficulties of contracting with a large pool of small farmers in a poor-country setting can provide important avenues towards securing a viable contract-farming scheme. Results of this RCT show that the nutrition incentive increased regularity of milk deliveries, albeit limited to the dry season and to those contracts headed by a woman. The impact on milk deliveries is mainly driven by delayed cow migration in the dry season and thus longer lactation period of lactating cows. The authors find no extra effort per lactating cow as a result of the incentive, nor do they observe an impact in terms of milk usage.

Results are consistent with a household-bargaining model, where households decide whether to migrate with cows or stay delivering milk to LDB. The introduction of the incentive not only increased a household’s utility from staying but, by targeting women, also increased women’s bargaining power, particularly with respect to decisions over migration. Impacts are concentrated in households where women are in control of the milk contract, which is further consistent with bargaining models. In households where men are in charge of the contract and women have more limited control over milk production, women’s response to the incentive is limited. What remains a puzzle is why men had a limited reaction to the incentive. A likely explanation is that men and women have similar preferences over children’s health but have different preferences on migration. Given migration patterns and social norms, it is likely that preferences over migration differ, with more drawbacks to women migrating with cows.

Conclusions

Several conclusions are derived from the results. Firstly, health-related incentives to reward effort or commitment, which are commonplace in many professional contracts throughout the world, can also trigger important behavioural responses in poor and remote settings. Second, intra-household dynamics play an important role in the effectiveness of these incentives. While targeting women may lead to larger impacts, this is only the case in households where women have control and power over decisions.

Lastly, value chain logistics may be cost-effective in increasing access to preventive health services to remote rural populations in poor countries. In sub-Saharan Africa in particular, inhabitants of rural areas are at a large disadvantage in terms of both access to preventive health services and quality of the services. Distance either strongly increases the cost of health campaigns that directly reach out to the targeted populations or strongly reduces household willingness to visit doctors. This is particularly so for preventive health. The results suggest that, from the LDB standpoint, the increase in milk delivery generated by the incentive is not sufficient to cover the full cost of the incentive itself. This is largely due to the limited complier population, seasonal effectiveness and unit cost. From a public policy perspective, however, such a contract may prove efficient at reaching remote populations and opening avenues for public-private partnerships in preventive health and/or nutrition services within agricultural value chains in poor countries.

Footnotes