Experiences with multiple propensity score matching

This presentation shows a somewhat complex automatization scheme in Stata that was developed during preparation of two papers using firm-level data and applying the propensity score matching techniques to distill the direct effects of the presence of foreign investors on various indicators from selection effects. The problem involved running multiple propensity score matching estimation procedures on different group of firms and on different efficiency measures. The solution involves 1) multiple nested loops to provide standardized output for several combinations of measures and groups; 2) correction of the standard PSM procedures to provide all required standard errors in the “returns†; and 3) use of postfiles to create user-friendly results sets that permit both reporting tables and generating publishable figures. The presentation also discusses alternative approaches that may be used to tackle similar problems.