The intervention aimed to stimulate agricultural productivity to ensure food security in several districts of Tanzania by implementing a multi-component intervention, which included developing farmer organisations, improving access to inputs through agro-dealer networks, strengthening extension services, and expanding access to output markets through contracting with processors. The intervention sought to reach 45,000 smallholder farmers, including 27,000 women. The ultimate objective was to enable men and women smallholder farmers to benefit from improved technologies, agronomy and efficient markets necessary to improve their food security and increase household incomes. The evaluation aimed to provide robust evidence about the impact of an integrated agricultural development programme, with the ultimate goal of informing agricultural policy in Tanzania and elsewhere. The evaluation design was based on randomisation, and that was not possible. Overall, results from this study must be interpreted with significant caution.
With a history of state control in agriculture, Tanzania is struggling with low agricultural productivity, inadequate capital, poor infrastructure, and uncoordinated farmers’ organisations.
The study asked two questions:
- Does a market-based approach targeted at food crop production stimulate agricultural productivity?
- To what extent does the programme contribute towards obtaining the targets of food and nutrition security?
The primary outcomes of interest were uptake of technology, agricultural productivity and production, producer prices, and the organisational strength of farmers’ organisations and agro-dealer networks.
The programme being evaluated aimed to increase the production of selected commodities (maize, rice, soy and beans) and to increase farmers’ ability to market these commodities at good prices. A consortium of organisations coordinated implementation of the intervention’s activities, all of which centred on working with farmers’ groups to promote adoption of recommended agricultural practices, access to quality inputs and access to markets. These activities included the creation of agro-dealer networks, trainings for both extension workers and farmers’ groups, demonstration plots, field days and warehouse renovations.
Theory of Change
The intervention is based on two sets of assumptions: 1) better access to markets and technology will lead to increased agricultural productivity of food crops; and 2) improvements in food crop productivity will lead to higher incomes and improved food security.
The intended design was randomisation. However, the team could not achieve a balance between the treatment and control groups at baseline. Consequently, the researchers used various statistical techniques to try to adjust for the lack of balance. These techniques include doubly-robust estimation, which combines propensity score matching and ordinary least squares. The authors acknowledge that the validity of these techniques depends on strong assumptions, and the results should be interpreted with significant caution.
As a supplement to the main impact evaluation, the study included an artefactual field experiment on production and intra-household bargaining. Participants completed sorting tasks, with some completing the task individually and others working jointly with their spouse. The participants chose rewards upon completing the task, and the study examined whether the rewards chosen were more in line with the husband’s preferences or the wife’s (a proxy for bargaining power), depending on whether they completed the task separately or together.
The randomisation failure limits the ability of the study to detect changes attributable to the programme. The analysis did not identify any significant effects of the intervention. However, it cannot be determined whether this is because the intervention had no effects, or because the study’s methods were unable to detect them.
The researchers did observe significant effects in the artefactual field experiment. When couples completed the task together, bargaining power did not change based on the distribution of work between the husband and wife. However, when they performed the task separately, men’s bargaining power increased. The authors suggest that if this pattern holds for real-world labour, men have incentives to work alone on income-generating activities.