Click here for search results


The study analyzed data from a range of sources, including a portfolio review, IEG’s own analysis of existing data sets for a dozen countries (three energy surveys, nine Demographic Health Surveys, and two income and expenditure surveys), and a review of Bank and external studies. The economic analysis unpacks the causal chain from the provision of electricity to the various benefits which it is claimed to bring, and quantifies these benefits where possible so as to address the balance of costs and benefits. The data were used to test the impact of rural electrification on several variables, such as the quantity of lighting used, opening hours of clinics, female health knowledge and income from home businesses.

First, a portfolio review was conducted to identify Bank lending (including IBRD, IDA, GEF, and Prototype Carbon Fund) for rural electrification since 1980.This allowed for the quantification and comparison of the project design features, the scale of rural electrification support in both monetary terms and the number and type of beneficiaries, and how these indicators changed over time. Second, 10 country case studies were compiled based on a desk review of Bank documentation and other documents on RE to capture the variety of experiences in different settings. [i]Third, a review was carried out of existing literature on the impacts of RE. Fourth, analysis was made of Demographic and Health Survey (DHS) data for 9 countries to examine the impacts of RE on health and family planning outcomes. [ii]In addition, household income and expenditure surveys for two countries (Ghana and Peru) were analyzed to examine impacts on rural income generation. And, fifth, RE-specific data sets were examined for Lao PDR, the Philippines, and Sri Lanka. [iii]

The report combines an overview of the Bank’s rural electrification portfolio with an analysis of the impact of this lending. It does so using a theory-based approach, identifying inputs and outputs, and then the outcomes (benefits) from those outputs and who receives them. The impact analysis is carried out on several levels, relying on the various survey data mentioned in the preceding paragraph. Most of this analysis is on single survey cross-sectional data, although panel data are available for two countries (Ghana and Peru). The challenge for most impact evaluations is to overcome possible selection bias. In the case of electrification, selection is very clearly on the basis of observables, most notably income and location. [iv]When the selection criteria are observable, as in this case, then the regression-based approach adopted in this study overcomes selection bias.  Hence the regression based approach is largely used to capture the impact of electrification compared to the counterfactual of no electrification. The study is first to unpack some of the possible benefits of rural electrification – for example, through media access to increased health knowledge and improved health and fertility outcomes. The report acknowledges weaknesses in the available data, calling, where appropriate, for more data collection specifically designed to examine these impacts.

[i] These countries were Bangladesh, Ghana, Lao PDR, Morocco, Nepal, Nicaragua, Peru, Philippines, Senegal, and Sri Lanka.
[ii] These countries were Bangladesh, Ghana, Indonesia, Morocco, Nepal, Nicaragua, Peru, Philippines, and Senegal.
[iii] These datasets were of variable quality. The ESMAP survey of the Philippines has been the best designed to date, though unfortunately the complete data set is no longer available, which restricted some of the analysis. The Lao PDR survey was based on the Philippines survey, but problems in survey design and execution also limited the analysis (e.g. the agricultural production and time use modules). The Sri Lanka survey had a rather different focus, and so limited usefulness for impact analysis. A recommendation of this report is for more tailor-made surveys to assess the impact of rural electrification programs.
iv]  Income may be regarded as endogenous with respect to electrification, creating a problem if data prior to electrification are not available. However, income may be instrumented either with assets, or if these are also thought to be endogenous, then fixed household characteristics such as education and sex of household head.

Permanent URL for this page: