Econometric solutions vs. substantive results: a crucial trade-off in the time-series-cross-section analysis