We study the effects of different information structures (full information, supply uncertainty and demand uncertainty) on equilibrium prices, allocative efficiency and bidding behavior in a (supply-side) uniform-price multi-unit auction, using supply function competition and a novel experimental design. Our setup integrates different types of market power and a varying level of competition. We empirically find that average prices tend to be higher under full information compared to the cases where bidders either have limited information about about the demand level or rivals’ technologies or; the latter even leading to strictly lower average prices as the exertion of market power and bid shading is strongly reduced. We explain this finding with a behavioral equilibrium concept, where bidders behave as if competing against the average market situation. Further, we address the problem of multiplicity of equilibria by exploiting the equilibrium conditions to obtain an empirical selection of the average equilibrium supply function. The respective predictions of the average prices exceed those by standard OLS in all information treatments.