We know that culture influences people’s behavior. Yet estimating the exact extent of this influence poses a formidable methodological challenge. This is because preferences and beliefs are endogenous, that is, they are shaped by individuals’ own experiences and affected by the same macro-structural conditions that constrain their actions. In a 2015 ASR article, Javier Polavieja introduces SISTER (Survey-based Imputation of Synthetic Traits used as Exogenous Regressors), a new method to address endogeneity problems in the estimation of cultural effects by using migrant populations. This innovative method uses imputed traits, generated from non-migrating equivalents observed at the country of origin, as instruments for immigrants’ own cultural traits measured at the country of destination. When migrants from multiple origins are cosnidered, SISTER can yield exogenous estimates. Polavieja used this innovative method to estimate the net exogenous impact of traditional values on female labor-force participation in Europe and found that this impact is much larger than standard regression methods would suggest. The article prompted a heated methodological debate in ASR bewteen the author and Winston Chou. Here you can download the original 2015 ASR article, Chou's comment and Polavieja's reply, together with supplementary material, including an annotated guide to implementation comands (in Stata).
The SISTER method
A unique statistical method developed at the D-Lab