School cheating and social capital

In this paper we propose and validate cheating in standardized tests as a new indirect measure of social capital. Given the low-stakes nature of most of the tests examined here, we interpret the widespread presence of cheating as a signal of limited trustin central education authorities. Cheating is negatively correlated with several social capital proxies in the local environment where a school is located (the municipality or the province), even controlling for area-wide differences in social capital and for a number of features of the local environment. When distinguishing between different kinds of social capital � contrasting universalistic and particularistic social values (along the lines of de Blasio, Scalise and Sestito, forthcoming) � cheating appears to be negatively correlated only with measures of universalistic social values (while the correlation of cheating with particularistic social values, if any, is positive). We also document a number of empirical regularities in cheating behavior: (i) within classes student homogeneity is associated with higher cheating (Lucifora and Tonello, 2012); (ii) the presence of external inspectors greatly reduces cheating (Bertoni, Brunello and Rocco, 2013), and to a greater extent in low social capital environments; (iii) in primary schools, cheating is more pervasive in smaller classes; (iv) and a larger share of �local� teachers, or of teachers with a permanent contract, is generally associated with higher levels of cheating.