An assessment of the Gelmini school reform in Italy: a synthetic control approach

I attempt to gauge the impact of the 2008-2010 Gelmini school reform on Italian students’ educational achievement. The reform aimed at cutting on educational spending by targeting teaching staff expenditures, as well as boosting the overall eciency of the education system. I apply Synthetic Control methods (SCM) to a panel dataset of six PISA international assessments for 25 countries, and carry out a case study of the reform. I find large effects on mathematics performance, but no statistically significant evidence of an impact on reading scores. The inferential strategy based on placebo runs sets the p-value for math treatment effect about the 10% threshold, meaning weak statistical significance. The observed positive effect on math scores may simply be the result of training for Invalsi tests and teaching to the test. However, a conservative and economically relevant conclusion can be drawn from my results: the Gelmini school reform did not negatively affect Italian students’ achievement in international tests. Contextual questionnaires allow me to provide an interpretation for these results. In robustness analysis, I experiment with changing the matching period, the predictors, and the donor pool units. Moreover, I directly address the scarcity of pre-intervention observations by merging TIMSS data, and applying a recently developed penalized Synthetic Control method. This is one of the few studies that attempted to quantitatively assess the outcome of the Gelmini reform. Moreover, to the best of my knowledge, my analysis constitutes the first attempt to apply SCM to a merged PISA-TIMSS database, and the first application of the penalized SCM to international assessments data.