"Words, Words, Words"
In the novel Contarini Fleming, future Prime Minister Benjamin Disraeli wrote "with words we govern men". Do words also govern whether a securities class action complaint will be successful? U.C. Berkeley School of Law Professor Adam B. Badawi attempts to answer that question in a forthcoming paper, How Informative Is the Text of Securities Complaints? After studying some five thousand private securities class action complaints that collectively contain over 90 million words, Professor Badawi concludes that text analysis and machine learning can predict whether a case will settle or get dismissed with about 70 percent accuracy in the best models. Among other things, he observes that words that allege fraudulent conduct over long periods of time are more likely to produce settlements and excessive reliance on media reports appears to be associated with the eventual dismissal of cases. These observations make intuitive sense but this conclusion is beyond my ken: that "the results suggest that there is meaningful information embedded in text and, relatedly, that excessive reduction of dimensions may produce omitted variable bias".