Bidding on price and quality: An experiment on the complexity of scoring auctions

We run an experiment on procurement auctions in a setting where both quality and price matter. We compare two unidimensional treatments in which the buyer fixes one dimension (quality or price) and sellers compete on the other, with three bidimensional treatments (with different strategy spaces) in which sellers submit a price-quality bid and the winner is determined by a score that linearly combines the two offers. We find that, with respect to the theoretical predictions, the bidimensional treatments significantly underperform, both in terms of efficiency and buyer's utility. We attribute this result to the higher strategic complexity of these treatments and test this intuition by fitting a structural Quantal Response Equilibrium model with risk aversion to our experimental data. We find very similar estimates for the risk aversion parameter across all treatments; instead, the error parameter, which captures deviations between the observed bids and the payoff-maximizing ones, is larger in the bidimensional treatments than in the unidimensional ones. Our evidence suggests that increasing the dimensionality and the size of the suppliers' strategy space increases their tendency to make suboptimal offers, thus undermining the theoretical superiority of more complex mechanisms.