Can cohort data be treated as genuine panel data?