Agribusiness Centers Impact Evaluation (Ghana)

Abstract

The Millennium Development Authority (MiDA) financed the construction of 10 Agribusiness Centers (ABCs) to provide services for the initial processing, storage, and marketing of grain crops produced by farmer-based organizations (FBOs) within their respective intervention areas. This data contains baseline information for the impact evaluation of this activity.

The treatment group for each ABC is comprised of households interviewed in the baseline survey that belong to MIDA trained FBOs within approximately 20-kilometer radius around the ABC. The comparison group for each ABC is selected from those MiDA-trained FBOs that are located outside the approximately 20-kilometer radius around the ABC. Therefore, those FBOs forming the comparison group will be loosely located between two concentric circles centered on the ABC; the inner circle will have a radius of approximately 20 kilometers, and the outer circle will have a radius of approximately 30 kilometers.

The three main research hypotheses this evaluation will try to answer (once endline data is collected), presented in form of hypotheses to be tested are: 1) access to ABCs will reduce post-harvest losses; 2) access to ABCs will imply higher market prices; and 3) access to ABCs will lead to higher crop yields. This, in turn, increases total production, total profits and, therefore, farmers' income from crop harvest.

To estimate the casual impact of this program, once endline data is collected, NORC proposed to use a double-difference estimator with matching or covariate controls as appropriate, or a combined regression analysis/matching approach to assess changes in farmers income, production, crop revenue and post-harvest losses related to access to ABCs.

We document some differences between households in the treatment and the comparison group. In particular, household heads in the treatment group are less likely to be female and have less education than in the comparison group. In agricultural practices too, we see some important differences. Households in the treatment group are more likely to grow kernel maize than households in the comparison group, which are more likely to cultivate fresh maize and rice. Treatment households cultivate a larger proportion of their land than do comparison households; conversely, comparison households irrigate a higher fraction of their land. While we see these and some other differences between the two study groups, they are similar in many other important characteristics such as number of household members, the fraction of children currently in school, and the likelihood of having experienced hunger. We also do not find major differences for agricultural or total income. This is important because it suggests that treatment and comparison groups are relatively comparable.