which studies a collusion-proof contract when, unlike the current setting, agents can collude only after they participate in the contract. This latter assumption may not be appropriate in many auctions where bidders are intimately familiar with their opponents even before participating. In fact, an allegedly predominant form of collusion involves bidders coordinating on their participation decisions: Colluders either refuse to participate or with- draw their bids to let a designated cartel member win without facing competition. Further, the idea of “selling the firm” to potential colluders, featured in Che and Kim (2007), relies on the agents’ inability to collude on their participation decision. The current paper relaxes the assumption by allowing the bidders to collude on their participation decision.