Get access

A, B or C? Experimental Tests of IPO Mechanisms


  • The authors acknowledge financial support from SDA Bocconi Research Division 2005. This paper has been completed while Stefano Bonini was Visiting Associate Professor at NYU–Stern Department of Finance. We are indebted to Alessandro Previtero, SDA Bocconi and UCLA for invaluable support in experiment structuring and comments. We thank Tim Jenkinson and Howard Jones at Said Business School, the University of Oxford, for helpful discussions and support. We are indebted to Peter Bossaerts, Stefano Gatti, Sandro Shelegia, Alessandro de Cillis, seminar participants at Bocconi University and the EFMA Conference 2007 for their comments. We owe special gratitude to the Editor, John Doukas, and two anonymous reviewers for invaluable comments and suggestions. The ideas expressed in the paper do not necessarily reflect the position of the authors' respective institutions. Any errors remain our own. Correspondence: Stefano Bonini.


Empirical research has provided extensive evidence on the inefficiency of bookbuilding in controlling underpricing. Both academics and practitioners have investigated this phenomenon proposing innovative offering methodologies. In this paper we explore the information revelation and underpricing properties of two baseline models, uniform auctions and bookbuilding, and two newly proposed mechanisms, Ausubel auction and Competitive IPO. Our findings confirm the empirical weaknesses of bookbuilding and provide hints that standard auctions may stimulate less bidding. However, (a) the Competitive IPO features increase competition both among banks and among investors resulting in more information revelation and less underpricing than standard bookbuilding; and (b) the Ausubel auction yields superior price discovery and underpricing outcomes than uniform clock auction and bookbuilding. Our experimental results provide novel insights into the ongoing debate on optimal equity offering mechanisms, suggesting that the solution to current issuing methodologies shortcomings may require the development of a ‘hybrid’ procedure blending properties of existing and new mechanisms.