Forward-selected panel data approach for program evaluation

Policy evaluation is central to economic data analysis, but economists mostly work with observational in view of limited opportunities carry out controlled experiments. In the potential outcome framework, panel approach (Hsiao et al., 2012) constructs counterfactual by exploiting correlation between cross-sectional units data. The choice control units, a key step its implementation, nevertheless unresolved data-rich environments when many possible controls are at researcher’s disposal. We propose forward selection method choose and establish validity post-selection inference. Our asymptotic framework allows number grow much faster than time dimension. easy-to-implement algorithms their theoretical guarantee extend big settings.