Résumé

Reverse (procurement) auctions are widely used across many sectors to allow buyers to acquire goods or services at competitive prices from multiple sellers. Experimental studies typically examine bidding behavior using a direct-response method (« hot » procedure), in which bidders submit one bid conditional on a randomly drawn induced cost value. Because this approach generates only one observation per period, multiple auction periods are usually conducted with different cost values, introducing potential learning and wealth effects. An alternative is to use the strategy method (« cold » procedure), which elicits multiple bids per subject conditional on several possible cost values, within a single auction period. The latter procedure facilitates online implementation, as it does not require simultaneous participation and allows auction outcomes to be computed ex post for incentive payments. Despite extensive discussions in the experimental literature about how hot versus cold procedures may affect behavior, direct evidence remains scarce in reverse auction settings.

We develop a laboratory experiment to disentangle the effect of the elicitation method on bidding behavior from those driven by repetition. While preliminary results indicate that bids elicited through the strategy method (in a one-shot setting) do not differ significantly from bids obtained under the direct-response method in repeated auctions. However, when both methods are implemented in a multiperiod setting and and we controll for periods, we find that strategy-method bids are, on average, more aggressive, that is, closer to costs.

Informations pratiques