This paper investigates the heterogeneous impact of school closures during Covid-19 pandemic in Italy on academic performance across different schools, grades, subjects and groups of students. Our analysis utilises an innovative dataset that combines administrative data on standardised tests in grades 5 and 8 with a specifically-designed survey that collects information about teachers’ practices between February and June 2020. Firstly, by employing a multilevel (mixed-effects) model, we estimate the extent of learning loss and examine its variability across schools, for students in primary and middle levels during the school year 2020/21. The findings confirm that learning loss has been considerable (between 0.05 and 0.27 SD) although heterogeneity across disciplines and grades exists – higher in English in grade 5, and in mathematics and reading in grade 8. Secondly, as a main contribution of the paper, we explore the mechanisms behind the substantial differences observed across schools, which can be explained by the ability of teachers in using digital tools and evaluating their students, as well as by the leadership role exerted by school principals.

本文调查了意大利 Covid-19 大流行期间学校停课对不同学校、年级、科目和学生群体学业成绩的异质影响。我们的分析利用了一个创新的数据集，该数据集将 5 年级和 8 年级标准化考试的管理数据与专门设计的调查相结合，该调查收集 2020 年 2 月至 6 月期间有关教师实践的信息。首先，通过采用多层次（混合效应）模型，我们估计了 2020/21 学年中小学学生的学习损失程度，并研究了其在学校之间的差异。研究结果证实，尽管不同学科和年级之间存在异质性，但学习损失相当大（0.05 至 0.27 SD 之间）——5 年级的英语较高，8 年级的数学和阅读较高。其次，作为本文的主要贡献，我们探讨了不同学校之间观察到的巨大差异背后的机制，这可以通过教师使用数字工具和评估学生的能力以及校长发挥的领导作用来解释。