Learning and collaborating towards improved living standards: A field experiment with Nigerian smallholder vegetable farmers

2023-03-01

2024-06-30

1. Technology adoption:
- Vegetable production
- Adoption of improved vegetable varieties
- Adoption of good agricultural practices
- Vegetable productivity

2. Livelihood outcomes:
- Home garden production
- Nutrition security
- Crop income

3. Intrahousehold collaboration:
- Project-level Women Empowerment in Agriculture Index (Pro-WEAI)

1. Technology adoption:
- Vegetable production: a) Share of farmers that cultivated vegetables, b) Total revenue from vegetables cultivated.
- Adoption of improved vegetable varieties: a) Share of farmers that bought improved vegetable seeds, b) Share of farmers that bought government-certified vegetable seeds, c) Share of vegetable crops of improved variety.
- Adoption of good agricultural practices: a) Share of farmers that adopted at least 5 GAP, b) Number of GAP adopted.
- Vegetable productivity: a) Total value of production for each of the three most commonly grown vegetables (NGN), b) Total value of vegetable production (NGN).

2. Livelihood outcomes:
- Home garden production: a) Total value of vegetable production from the home garden (NGN).
- Nutrition security: a) Household Food Insecurity Access Prevalence (HFIAP), b) Household Food Consumption Score (FCS).
- Crop income Total crop income (value of production of all crops cultivated minus costs of purchased inputs (NGN)).

3. Intrahousehold collaboration:
- Pro-WEAI: a) Self-efficacy, b) Autonomy in income use, c) Input in productive decision making, d) Control over the use of income.

The research studies smallholder farmers in northern Nigeria who cultivate vegetables for local and national markets. The East-West Seed Knowledge Transfer (EWS-KT) foundation assists smallholder vegetable farmers through the provision of agricultural extension services (AES) with the aim of improved farm productivity. EWS-KT works through a train-the-trainer model and first identifies key farmers in target communities.

The EWS-KT model distinguishes between three types of farmers: key, core, and peer/neighboring farmers. Key farmers are selected for being ‘high potential’ farmers who are very influential in their communities, but also based on practical aspects like access to irrigation, and proximity to the tarmac road. Key farmers cultivate an irrigated, centrally located plot and have an entrepreneurial mindset (among other characteristics). A demonstration plot is established on the key farmer’s premises where EWS-KT will train both the key farmer and other motivated ‘core’ farmers from the community. Key farmers manage demonstration plots during the training period. During the season, each key farmer, together with a group of 10-20 core farmers from the same community receives direct training from EWS-KT technical field officers at their demonstration plot. Neighboring or ‘untrained’ farmers receive no direct training but may get exposed through the various facets of the intervention (e.g., presence of a demonstration lot in the village, open training events) or through exchanges with other farmers. Afterwards, farmers are graduated publicly, making them recognizable to others.

This research focuses on three interventions:
• AES training: EWS-KT provides extension training on vegetable production to selected key and core farmers, with separate groups for men and women. Importantly, the same training content is provided to both groups.
• Branding: A training certificate is introduced for successfully trained farmers, awarded in a public ceremony in which the ability of these farmers to share their knowledge within the community is explicitly announced. The hypothesis is that this signal will result in more active interactions and learning from certified farmers. Moreover, these farmers might also benefit from certification through improved market access resulting in better livelihood outcomes.
• Gender sensitization training: This intervention includes spouse training on vegetable production and sensitization messages and discussions about the relevance of intrahousehold collaboration and working together as a family. An extra training session will be included for women-only group on the preparation of bio-pesticides, to increase women’s participation and generate additional income.

This research employs a randomized control trial in 150 communities from eight local government areas in Kano and Kaduna states in northern Nigeria. Treatment is assigned at the community level to account for spillover effects within communities, while data will be collected at the household level. Communities were randomly assigned to one of two main treatment groups or the control group:
• Treatment group 1 (T1): AES training
• Treatment group 2 (T2): AES training and branding
• Control group (C): No AES training and no branding

Orthogonal to the initial randomization, we randomly formed two other groups from the pool of treatment communities only:
• Sub-group 1 (S1): Spouse training and intrahousehold collaboration sensitization
• Sub-group 2 (S2): No spouse training

Not available

Computer

This is a cluster-randomized design. The randomization is at the community level.

Yes

150 communities.

In total, 2,562 individuals and one of their spouse (for married farmers) are targeted.

102 treated communities (T1=52, T2=50; orthogonal from the treatment arm: S1=51, S2=51) and 48 control communities.

In each of the 150 community, farmers are selected in the following way:
• 9 core farmers in each of the 102 treatment communities
• 9 Neighbouring farmers in each of the 102 treatment communities
• 12 Neighbouring farmers in each of the 48 control communities
• 150 key farmers; 1 per community

Using Stata software, power calculations were computed to obtain the minimum detectable effect sizes (MDES) for the study outcomes. The calculation is based on the priors of the baseline data collection from 3,476 observations. In brief, the study is well-powered and can detect small effect sizes for most outcome indicators (MDES range between 0.001 - 0.116), with perhaps an exception of revenues and incomes. Whereas MDES is less than 30% of the standard deviation for each indicator, it is high compared to the mean. For total vegetable revenue, we can only detect a 30% increase, for total crop income this is even 61%.