Bids and values

Abstract: This thesis consists of three self-contained essays on auctions and one essay on the voluntarycontribution to public goods.Essay I analyzes bidding data from real-estate auctions within the theoretical framework of the independent private-values model. The data is collected from executive auctions of owner-occupied flats in the Stockholm metropolitan area during 1995-1997. The flats are sold through an English auction where the seller, prior to bidding, sets a secret reservation price, which is never disclosed. An equilibrium bid function is derived and mapped into two econometric specifications. By structural estimation the distributions of bidders' private valuesand seller's reservation price are estimated. The estimated average bidders' value of the flats is fairly low, about one-fifth of the officially assessed value. Using the estimated distributions, the seller's expected profit from moving to a mechanism where the reservation price is announced is quantified. Simulations indicate that the seller's revenue is slightly higher if a reservation price is announced. The average sale price is, however, higher if the reservation is kept secret.Essay II and Essay III study bidding data from procurement auctions of railway embankments (Essay II) and of road markings (Essay III). The observed bids in both data sets have been generated by a first-price sealed bid auction. The bids are analyzed within the theoretical framework of the private-values and the common-value model. The two auction models are mapped into various empirical specifications. The purpose is to examine whether the bidding behavior in these procurement auctions can be explained by one of the two models. This is done by estimating the underlying distribution of bidders' values or signals. In Essay II the distribution of values is estimated with the maximum likelihood method, using all bids. In Essay III both maximum likelihood and nonlinear least square methods are adopted, using all bids as well as just the winning bids. In Essay II both auction models fail to explain observed behavior, whereas the common-value model seems to fit data fairly well in Essay III.Essay IV describes a laboratory experiment in which nearly 300 students are asked to state their willingness to pay for the preview of a movie. Two elicitation methods are used: the open-ended (OE) and the dichotomous-choice (DC) formats. The results indicate that individuals respond to incentives in the predicted direction. Further, when an individual had an incentive to overstate his true WTP the DC format yielded higher estimates of WTP than the OE format. When an individual is subjected to opposite incentives the tests suggest no behavioral differences between the two formats.