This study examines the impact of the Rural Education Project (PER), a government intervention aimed at improving the efficiency and quality of education in rural Colombia. Low educational efficiency and quality contribute to pervasive poverty and underdevelopment in rural areas of many developing countries. To address this issue, the PER provides funding, training and material for the implementation of flexible education models designed to meet the specific needs of rural students. From its initiation through 2005, the PER was implemented in 4,485, or 14 per cent, of rural schools in Colombia, including preschools, primary and secondary schools and adult education programmes. This study evaluates the programme's impact at the school level on enrolment, educational efficiency as measured by passing, failure, and dropout rates and educational quality as measured by standardised test performance.
The authors use panel data on 21,207 rural schools for 2000 through 2005 from the Colombian national census and Ministry of Education reports. Using these data, the authors estimate the average impact of treatment on the treated using three different strategies. First, they estimate a series of difference-in-differences models comparing changes in outcomes over time between treated and nontreated rural schools. Second, to correct for potential bias arising from nonparallel trends across time between the treated and control groups, the authors estimate propensity scores for participation in the programme. Then, using these scores, the authors estimate separate difference-in-differences models for the subsample of schools whose scores lie within a common support. Finally, to correct for bias arising because the distribution of characteristics between schools within the common support could be significantly different, the authors use a nonparametric difference-in-differences matching estimator first devised by Heckman, Ichimura and Todd (1998).
The authors find that the PER led to significant improvements in educational efficiency in rural Colombia. The nonparametric matching estimates indicate that schools that received the PER had a 4.7 percentage point greater increase in grade passing rate than other schools in the sample, a 1.4-point greater reduction in failure rate and a 3.2-point greater reduction in dropout rate. These results are robust across the other estimators employed by the authors. With respect to educational quality, however, the results are not robust across the three sets of estimators; however, the authors' preferred estimator, the nonparametric matching technique, indicates that schools that received the PER recorded an improvement in standardised language test scores of 0.03 standard deviations greater than schools that did not receive the PER. This study also indicates that the PER's positive effects are driven by programmes targeted toward primary and secondary education, rather than preschool or adult education. The preferred nonparametric matching model indicates that PER programmes targeted toward primary schools increased passing rates by 8 percentage points, decreased failure rates by 2.3 points and reduced dropout rates by 5.7 points. PER programmes directed toward secondary schools increased passing rates by 5.1 points and reduced dropout rates by 4.4 points. However, PER programmes targeted toward preschool and adult education programmes had no significant impact on educational efficiency or quality.
From a policy perspective, the findings suggest that supply-side interventions focused on providing educational inputs can have significant and positive effects on educational efficiency and quality. At the same time, this study also suggests that specially tailored programmes that address the particular needs of rural areas can be an effective strategy for improving educational outcomes in these underdeveloped regions.