Most recent regulations establish that resolution of global banking groups shall be done according to bail-in procedures and following a Single Point of Entry (SPE) as opposed to a Multiple Point of Entry (MPE) approach. The latter requires parent holding of global groups to put up front the equity capital needed to absorb losses possibly emerging in foreign subsidiaries-branches. No model rationalized so far such resolution regime. We build a model of optimal design of resolution regimes and compare three regimes: SPE with cooperative authorities, SPE with non-cooperative authorities and MPE (ring-fencing). We find that the costs for bondholders of bail-inable instruments is generally higher under noncooperative regimes and ring-fencing. We also find that in those cases banks have ex ante incentives to reduce their exposure in foreign assets. We also examine recent case studies that help us rationalize the model results.