Scoring rules in experimental procurement

Abstract

We report the results of an experiment where subjects compete for procurement contracts to be awarded by means of a scoring auction. Two experimental conditions are considered, depending on the relative weight of quality vs price in the scoring rule. We show that different quality-price weights dramatically alter the strategic environment and affect efficiency. Our evidence shows that each weighting better delivers against a matching objective function than using a scoring rule which misrepresents the buyer's objective function. Nonetheless, there are large deviations in how each performs, with the higher weight on quality delivering much greater efficiency evaluated against its own objective function than a low weight on quality evaluated against its own objective function, despite the higher quality weight inducing higher deviations from equilibrium. We propose a “mediation analysis” to show that the “direct effect” (due to the different strategic properties of the induced game-forms) outweighs the “indirect” one (how the different game-forms affect out-of-equilibrium behaviour). We also perform a structural estimation of the Quantal Response Equilibrium induced by subjects’ behavior, where we find that subjects are risk averse and noisy play affects behavior in the direction of underbidding.