1step,rule,n_before,n_after,n_removed,rationale2availability,target month s+1 <= t-2 and regressors observed at origin t,345,337,8,"last estimation sample (origin 2026-04-01, h=1); rows = candidate pseudo-origins s, one per input month"3lags_in_sample,target and lagged regressors present in file (drops first months lacking y[s-2]/e[s-1] and GDP-empty months),337,335,2,"last estimation sample (origin 2026-04-01, h=1); rows = candidate pseudo-origins s, one per input month"4covid_window,drop rows whose target or regressor month lies in 2020-03..2021-12,335,310,25,"last estimation sample (origin 2026-04-01, h=1); rows = candidate pseudo-origins s, one per input month"5