Estimation of a Dynamic Auction with Participation

We study participation decisions and bidding behaviour in Michigan Department of Transportation procurement auctions. Patterns in the bidding data suggest that bidders’ participation behaviour has a forward looking component. To fully understand the extent of these effects on auction outcomes, we construct and estimate a dynamic asymmetric auction model with endogenous participation. We develop an estimation approach which builds on Guerre, Perrigne and Vuong (2000) and recently developed dynamic discrete game estimators. We then quantify the level of inefficiencies under the current auction rules and then consider how alternative auction rules affect efficiency. We also analyse the effect of ignoring dynamics in this market by estimating a static version of our model. This approach results in misleading conclusions concerning auction efficiency.