This paper presents the outcome of an experiment where the standard one shot sealed bid procurement auction for two identical goods provides the benchmark. Inducing scale economies a combinatorial auction is applied on the situation with non-linear costs. The mechanisms are first run without, and then with the possibility for subjects to communicate prior to bidding. There are two human and one computer bidder in each period. It is demonstrated that the combinatorial mechanism is able to enhance efficiency and that subjects are less inclined to cooperate under the combinatorial auction than under the standard bidding format. [JEL Classification: D44, D78]