We reported in a recent paper that during the 2008-09 financial crisis, for the average firm, idiosyncratic risk, as measured by variance, increased by five-fold. This finding is important for securities litigation because idiosyncratic risk plays a central role in event study methodology. Event studies are commonly used in securities litigation to determine materiality and loss causation.

Many bits of news affect an issuer’s share price at the time of a corporate disclosure that is the subject of litigation. Because of this, even if an issuer’s market–adjusted price changes at the time of the disclosure, one cannot determine with certainty whether the disclosure itself had any effect on price. An event study is used to make a probabilistic assessment of whether in fact it did. Use of event studies generates a certain rate of Type I errors (disclosures that had no actual effect on price being identified as having had an effect) and a certain rate of Type II errors (disclosures that had an actual effect not being identified as such).

This paper sets out a simple model of the tradeoff between these Type I and Type II errors. The model is used to establish three fundamental points. First, an economic crisis can radically worsen this tradeoff by making it much more difficult to catch a disclosure of a certain size without introducing more Type I errors. Second, during crisis periods a relaxation of this standard (and hence an increase in the acceptable rate of Type I errors) may actually decrease Type II errors by less than it would in normal times. We prove that whether the decrease is greater or smaller in crisis times depends on whether the disclosure’s actual impact on price is more or less negative than a definable crossover point. Third, whether relaxation of the standard in troubled times would increase or decrease social welfare is ambiguous. It depends on distribution of potentially actionable disclosures in terms of their actual impact on price and the social costs and social benefits of imposing liability for disclosures of each given level of actual negative impact on price.