Panel Paper: Millennium Challenge Corporation's Electricity Project in Tanzania: Evaluation Design and Early Impacts

Saturday, November 9, 2013 : 1:45 PM
Boardroom (Ritz Carlton)

*Names in bold indicate Presenter

Duncan Chaplin and Arif Mamun, Mathematica Policy Research
In this paper we will describe the evaluation design and early findings from a rigorous evaluation of an electricity development project. To help promote economic growth and reduce poverty, the Millennium Challenge Corporation (MCC) is funding a $207 million energy sector project in Tanzania. The two major components of the project we are evaluating are: (1) an electricity line-extension activity designed to provide new electricity lines to over 300 communities, and (2) a customer connection financing-scheme initiative that will provide low-cost connections to about 5,800 households in 29 of these communities that were randomly chosen. 

We developed rigorous evaluation designs to estimate impacts of these two components of the electricity project. To evaluate the line-extension activity we are using a quasi-experimental difference-in-differences design with a matched comparison group. To estimate the impacts of the financing-scheme initiative we are using a random assignment evaluation design, where communities randomly assigned to the treatment group will pay about one-sixth of the standard connection fee, while those randomly assigned to the control group will pay the usual amount.

As the first step of the evaluation, baseline surveys were conducted with over 10,000 households in 2011, before the new lines were built. These households were spread across 182 of the intervention communities receiving the line-extension activity and another 182 carefully matched comparison communities. The treatment and control communities for the financing-scheme initiative were randomly chosen from the 182 surveyed intervention communities. Hence, the same survey data are used to evaluate both components.

The data allowed us to demonstrate baseline equivalence. After matching at the household level, we find no evidence of systematic baseline differences between the intervention and matched comparison groups used to evaluate the line-extension activity, or between the treatment and control groups used to evaluate the financing-scheme initiative. Thus, we have the platforms for estimating unbiased impacts using follow-up survey data for these two components of the energy sector project.

The line extension work began in 2012 and will continue during 2013. We will collect connection rate data from the local electricity company soon after the lines are built and follow up survey data on a larger set of outcomes in 2015. In this paper we will present short-run estimated impacts on connection rates for both the line-extension activity and financing-scheme initiative. In future we plan to use the follow-up survey data to estimate impacts on energy use, health, education, economic activity, income, and poverty.