Abstract/Summary

This research project represents the first baseline socioeconomic survey that CDRI has conducted for the Greater Mekong Sub-region Transmission Project of Asian Development Bank. The study was designed to develop a set of comprehensive baseline demographic, social and economic as well as energy consumption indicators for the project. The tools for data collection were village and household surveys in 27 villages, of which 17 villages are potentially project beneficiaries (treatment group) and 10 are non-beneficiaries (control group). It then used propensity score matching method to predict the probability of household participation (propensity score) in the two groups.

From the matching operation, it produces favourable results for consumption, male adults’ study time, and family and children’s teaching time, while total income differences of households in the two groups, and income from sources such as crops, common property resources, labour wage and remittance remain statistically significant, albeit reduced. This indicates that after controlling for household and village characteristics, households in the control and treatment groups have similar consumption behaviour, total spending and time allocation for reading and teaching at home. Although the matching operation could not produce better results in terms of household income from different sources, the median bias of the matching operation has been reduced substantially.

The propensity score matching analysis enables us to see if good matches for the treatment households can be found in the comparison sub-sample, and to consider the possibility of dropping some of the comparison households (those with highly different characteristics) from any follow-up survey. After matching, 35 households, which were off-support, were dropped creating the new sample of 709 reduced from 744. For follow-up evaluation study, the research team should refer to after-match sample villages as the study site and it is hoped that the survey costs could be reduced.