Rethinking “Distance from”: The Examples of Wittenberg and Mainz
本文指出经济史研究中用距离变量作为历史干预代理变量的做法存在历史和经济计量问题，通过复制和模拟发现其易产生虚假显著估计，提醒学者谨慎使用。
Abstract An influential economic history literature uses distance-based variables as proxies for exposure, or access, to historical “treatment interventions” with potential economic impacts. This paper examines the assumptions underlying this practice and find s historical and econometric problems that engender misleading conclusions. Historical evidence challenges the assumption that physical proximity determined access to ideas or technology even in pre-industrial settings. Replication exercises and statistical simulations reveal that “distance from” variables frequently produce falsely significant estimates in first-stage instrumental variable (IV) and Ordinary Least Squares (OLS) regressions, and potentially violate the exclusion restriction in the IV second stage. My findings suggest caution in using distance-based measures as proxies for the diffusion of information and resources.