10 - Generalizing the winner's curse hypothesis: The case of the Belgian IPO market

Abstract

This chapter examines the winner's curse hypothesis on a sample of initial public offerings (IPOs) on the Belgian stock market from 1989 to 2004. The investor is assumed to be uninformed and to systematically allocate a lump sum to an initial order. Based on the sample of IPOs performed on the First Market of Euronext Belgium, evidence of underpricing is found that can be quite severe on average. However, taking into account the prevailing allocation systems, which can be very different from one issue to another, there is no significant evidence of the possibility of escaping the winner's curse phenomenon for an uninformed investor. Economically significant returns can be obtained for very low levels of investment, but the high volatility of observed returns over a 4-month horizon does not enable one to extract scientifically valid evidence of positive initial returns. Given the very heterogeneous set of allocation rules for the various IPOs under study, the payoff structure for this uninformed investor appears to be very nonlinear. Irrespective of the reserved amount and the holding period, it finds no evidence of positive abnormal return for this strategy over the period, thus confirming the winner's curse hypothesis. However, conditioning the subscription to the direction of the stock market returns can provide positive abnormal returns under some circumstances.