In this study we identify necessary and sufficient conditions for sufficient statistics to strictly (Pareto) dominate all nonsufficient statistics as information for contracting in agencies with moral hazard. We first observe that strict dominance requires that an optimal compensation scheme...