Implementation with renegotiation when preferences and feasible sets are state dependent