Multidimensionality and Complexity in Scoring Rule Auctions: Experimental Evidence
Abstract: We experimentally investigate the problem of a procurer who needs to procure a good or service for which both price and quality matter. To this end, we compare two unidimensional treatments where one element (price or quality) is fixed by the auctioneer and sellers compete on the other dimension, with three bi-dimensional treatments where sellers submit both a price and a quality offer and the contract is awarded on the basis of a "score" that (linearly) combines the two offers. Our results show that, differently from the unidimensional treatments, the bi-dimensional treatments significantly underperform relative to their theoretical predictions, both in terms of efficiency and in terms of procurer's surplus. We argue that this might be related to the higher complexity of the bi-dimensional treatments. To corroborate this intuition, we fit a structural Quantal Response Equilibrium to our data: we obtain very similar estimates for the risk aversion parameter across treatments; on the other hand, the rationality parameter, which measures how close are observed bids to the payoff-maximizing ones, is lower in the bi-dimensional treatments, suggesting that adding a dimension to the strategic decision problem faced by bidders increases the likelihood that they play suboptimally, thus greatly reducing the theoretical superiority of more complex mechanisms.