Sector crashes and sector booms

In this paper we introduce sector crashes (and their counterparts: sector booms) in an effort to enlarge the number of financial crashes available for study. We find that the behavior of these sector crashes is similar to the behavior of general market crashes. Using this similarity and using the larger number of crashes available, this allows a more detailed and quantitative analysis of financial crashes. Based on intuitively appealing definitions with a sound statistical and economical foundation, we obtain several interesting results. We find an average crash size of around five standard deviations and our results indicate that roughly one half of all crashes is followed by what we call `aftershocks'. While outperformance doubles the likelihood of a crash, we find that crashes occur less frequently after outperformance that is preceded by strong underperformance. Apart from the increased likelihood of a crash after outperformance we find that a crash seems to be an event on its own and that most crashes are surprisingly consistent in terms of length, size and speed. Whereas one might expect differences between crashes and booms, our results also indicate remarkably similar behavior for both.