Before investors sue companies for economic losses, they may need to re-evaluate the evidence, says Penn Law Professor Jonathan Klick. In Broudo v. Dura Pharmaceuticals, the Supreme Court ruled that investors can only recover losses if they prove that a company’s fraudulent action directly caused a decline in stock prices.
To establish this link litigators rely on event studies, a statistical method to assess whether an event, such as an announcement about earnings or a merger, impacted a company’s stock value and to what degree.
Klick, an empirical law expert, is challenging the standard approach to event studies, which he says is analytically unsound. The research has “the potential to completely change the standard practice in securities fraud litigation,” said Klick, who is collaborating with Jonah Gelbach of the University of Arizona. Other co-authors include Eric Helland of Claremont McKenna College and Robert Sitkoff of Harvard Law School.
An event study measures the difference between the actual stock return and the predicted stock return on the day of the event. If the difference is significant, then shareholders can allege that the event did in fact impact stock prices.
The standard approach, which compares the individual return against the variability of returns implied by a normal distribution, is useful when the sample sizes are large and the normal distribution provides a valid approximation of distribution of the stock’s returns, said Klick. An analyst, for example, might conclude that the normal return for a stock based on market behavior should be 5 percent, but the return is actually 20 percent. At first glance, this might seem like a significant deviation. However, individual stock return distributions might deviate significantly from the normal distribution. For some firms, this 15 percent deviation between the predicted return and the actual return might be quite common even when nothing out of the ordinary is happening.
For single firm event studies, Klick and his colleagues are proposing a non-parametric approach in which analysts compare the stock’s performance against the company’s observed empirical distribution, not the normal distribution.
The problem, said Klick, is that securities litigation relies on faulty event studies, which might either create a case where there isn’t one or dismiss a case where there actually is abnormal behavior.
Using data on the performance of 3000 securities over a 7-year period, Klick and his colleagues have shown that the standard approach often yields inaccurate results when doing single firm event studies. The empirical approach solves this problem. It is also easy to implement, and can be extended to event studies involving more than one company.
The empirical approach “will generally outperform standard method and will never do worse,” said Klick.
Klick and his colleagues plan to produce a series of papers for law, finance and economic journals and to present the findings at scholarly and practice-oriented conferences. They also plan to write a non-technical guide for judges, clerks and practicing attorneys.