Comparison of two parameterizations of the winner's curse model in a sample of Swedish Treasury auctions

Purpose – The purpose of this paper is to examine the underpricing effect in Treasury auctions. Design/methodology/approach – The paper compares two winner's curse models using a dataset on multi-unit auctions. The dataset is from Swedish Treasury auctions, which is under a discriminatory auction mechanism. One model is a single-unit equilibrium model assuming that each bidder bids for 100 percent of the auctioned securities, which is described by Wilson and solved by Levin and Smith. The other model is a multi-unit model calibrated by Goldreich using the US Treasury auctions data and assumes that each bidder bids for one unit of the auctioned securities. Findings – The empirical results show that, although both models work well in predicting the bid-shading, the multi-unit model fits the Swedish Treasury auctions data better than the single-unit model. Research limitations/implications – The evidence implies that bidders rationally adjust their bids due to the winner's curse/champion's plague. Originality/value – This study provides close quantitative predictions of the amount of bid-shading using both single-unit model of Wilson and multi-unit model of Goldreich, and indicates that winner's curse or champion's plague worries bidders in countries other than the USA.