View article

Targets and tools for the monitoring and evaluation of educational policies and interventions in the economics of education have changed rapidly in the last 20 years. Most previous works in this field have focused on running multivariate analysis models to find statistical associations between variables, controlling for the presence of other covariates and factors that also influence the dependent variables. Behind this traditional approach lies the strong assumption that all covariates related to the dependent variable are exogenously determined. In other words, we would say that unobserved variables are equally distributed among the population to be analysed. 1 Given modern estimation technology it is no longer reasonable to make this assumption. Parents’ decisions regarding the education of their children (for example, choice of school, pre-primary education attendance, extra-curricular activities, support at home …