Beschreibung:
Studies of long-run stock price abnormal performance aftercorporate events are plagued by difficulties in statistical inference and the inevitable joint hypothesis problem in tests of market efficiency. In this paper, we study long-run performance using a 'model-free' stochastic dominance approach. An advantage of using a stochastic dominance criterion is that we can compare the entire distribution of portfolio payoffs while incorporating specific assumptions about investors' preferences. We use data similar to Ritter (1991) to revisit long-run performance after IPOs. Our main results are as follows. First, we find that investors, who prefer more wealth to less, are indifferent between an IPO firm portfolio and commonly used benchmark portfolios. Second, we find little evidence that risk-averse investors prefer small size and low book-to-market benchmark portfolios to an IPO portfolio. Our results draw attention to an important issue in long-run event studies; even abstracting from a possibly misspecified asset pricing model and difficulties with statistical inference, theory does not provide any guidance about how to choose a specific benchmark portfolio. In the case of IPOs, at least, the issue of underperformance may remain unresolved empirically, even with better statistical methods and a well-specified asset pricing model, due to the latitude available in the choice ex ante of a benchmark for comparing performance. In general it may not be feasible empirically to distinguish, in long-horizon event studies, whether the observed under- or over-performance is due to mispricing or an inefficient market or simply due to an incorrect choice of a benchmark portfolio