Abstract: | Recent research has been motivated by the fact that the outcome function in implementation may not be “credible”. On the one hand, the players may try to renegotiate the final outcome, if there is another outcome which they prefer. On the other hand, the “social planner” may have an incentive not to implement the finaloutcome, if there is another outcome which he prefers. I show that in the exchange economy, the first problem is not serious. a planner with unlimited commitment power can design a “collusion-proof” mechanism, which is stable against all sorts of group devciation, including the ex post trade of goods among the agents. I will, However, argue tha hte second problem (the planner's commitment problem) can be very serious. |