LEARNING INEQUALITIES BETWEEN PRIMARY AND SECONDARY SCHOOL. DIFFERENCE-IN-DIFFERENCE WITH INTERNATIONAL ASSESSMENTS

Evaluating the effect of institutional features by exploiting cross-country variability with crosssectional data is difficult. Difference-in-difference strategies are sometimes employed to reach identification. In this paper, we discuss the difference-in-difference strategies adopted in the literature to evaluate the effect of early tracking on learning inequalities using surveys administered to children of different grades. In their seminal paper: “Does educational tracking affect performance and inequality? Differences-in-differences evidence across countries” Economic Journal (2006), Hanushek, and Woessmann analyze the effect of early tracking on inequalities with two-step analysis. Other scholars, instead, focus on the social background regression coefficient, using individual-level models applied to pooled data from all countries. We demonstrate that since test scores are measured on different scales at different surveys, pooled data strategies may yield to completely uninformative results. Against this background, we use data on reading literacy in PIRLS 2006 and PISA 2012 and carry out two-step difference-in-difference analyses on the effect of early tracking on social background learning inequalities.