Culture as a Random Treatment

Abstract

In a 2015 ASR article, I introduced SISTER, a new method to estimate the causal effects of culture using migrant populations. Chou raises significant concerns about SISTER and concludes that the method is flawed. I contend that this conclusion is incorrect because it is based on a mischaracterization of the method’s identification assumptions. Specifically, Chou disregards that SISTER exploits cultural variation across multiple countries of origin/ancestry as the main source of identification. I argue that SISTER can comply with the IV assumptions precisely because it is a multiple-origin method that conceptualizes culture of birth as a random treatment. I discuss potential threats to the exogeneity condition and offer several recommendations for future applications of the method.