Impact Evaluation for El Salvador’s Rural Electrification Sub-Activity
The Rural Electrification Sub-Activity of the El Salvador Compact will extend electricity to households in the Northern Zone that currently are not connected to local power distribution networks. Service will be provided to these households through, as appropriate for the household, investments in the extension of distribution networks, in individual household connections to the network, and in the supply of off-grid solar photovoltaic systems.
The impact evaluation seeks to determine the impact of electrification on the cost of energy, energy consumption, time allocation, and household income. Because the new electric lines will come from the existing power grid, an experimental design is not feasible for the overall impact evaluation. Therefore, the evaluators will use a non-experimental design taking advantage of the timeline of the rollout of the project combined with propensity score matching to identify treatment (households that receive the new electrical service) and control groups (households that do not receive new service).
Using specialized household surveys for both the household head and his/her spouse with an intra-household time allocation module, the evaluators will estimate the differences in energy consumption, household income, and time use between the treatment and control groups. A difference-in-difference estimation method will control for changes in non-observable variables, and instrumental variables estimation will control for any remaining potential sources of selection bias.
For a subsample of towns and households, the evaluators will select an additional control and treatment group. The treatment group will be randomly assigned vouchers of varying amounts that would help cover costs for the installation of the connection that the households will need to pay in order to access the electricity once the cable reaches their household. Vouchers will be assigned randomly to between 10 and 50 percent of eligible survey respondents. The vouchers would not only encourage a sufficient level of demand for electricity access in the early stages of intervention, but would also provide for this subsample of households a basis for experimental evaluation of accessibility to electricity by serving as an instrumental variable for electricity access. The randomized selected control towns and households will serve as an appropriate control group given they will receive no vouchers.
The impact evaluation will be based on a survey of 1,532 total households in the departments of Chalatenango and San Miguel. These departments are proposed because, according to program plans, they include the largest numbers of cantons that will benefit from the electrification program. In addition, these districts include a number of cantons that will benefit from both the road improvement and the electrification projects, so data collection could be done at the same time for both evaluations.
The questionnaire includes two sections – one that will be answered by a male in the household (including household income and agricultural productivity) and will be interviewed by a male survey taker, and the second which will be answered by a female in the household (including household demographics and expenses) and interviewed by a female survey taker. The separate interviews are intended to get more accurate data on topics that are more familiar to certain people in the household. The current data collection plan anticipates that each household will be surveyed three times, 1) baseline in November 2008, 2) follow up in November 2010, and 3) final in November 2011. This may change however if there are delays in the construction schedule.