Script reform, human capital and social change: evidence from Uzbekistan’s Latin alphabet transition

Summary

Script reforms are common nation-building tools, yet rigorous evidence on their effects is scarce. This paper examines Uzbekistan's transition from Cyrillic to Latin script, mandated in 1993 and implemented in schools from 1996. I study heterogeneous effects across initial conditions using a difference-in-differences design that exploits variation in exposure across birth cohorts and pre-reform district characteristics. The reform had no significant average effect on literacy, but it generated substantial heterogeneity: districts with lower baseline literacy gained 0.09 standard deviations per standard deviation of the baseline literacy gap, whereas districts with more Russian speakers experienced relative losses of about 0.05 standard deviations. These patterns reveal the redistributive politics of nation-building. The reform also reshaped labour markets: Latin-exposed cohorts moved away from state employment and, less sharply, from entrepreneurship. The same cohorts became more likely to express intolerance towards speakers of other languages, an effect specific to the linguistic margin. The findings suggest that script reforms do not deliver aggregate literacy improvements but have significant distributional consequences and may heighten ethnolinguistic divisions.