We investigate the efficiency of Coasean bargaining when transfers between agents are capped. We model a two-stage Coasean environment where, in the first stage, property rights are costly to attribute. After the attribution stage, agents voluntarily exchange over the level of harm. If property rights are attributed via an all-pay auction, then the introduction of a cap is Pareto improving. Using a Tullock contest we find a cap is not Pareto superior, but may increase Kaldor-Hicks efficiency. Our analysis provides additional insights into the establishment/enforcement of property rights, such as indigenous land rights, and highlights when restrictions on transfers are desirable.