This is somewhat of a "dog bites man" story. Feng Li, an assistant professor of accounting at the University of Michigan did a comprehensive study (55,000 firm-year observations) on the relationship between the readability of annual reports and the level and volatility of earnings.
He found that the annual reports of companies with poorer performance and/or more volatile earnings were much less readable. He also found that good performance was more persistent for firms with more readable reports.
One obvious interpretation is that firms try to bury bad performance with abtruse writing. Another is that poor writing and poor performance are both caused by a lack of ability. A third is that poor performance is more difficult to explain than good performance.
Either way, it's an interesting study. There's a PDF version on SSRN here.
I wonder if there's a similar pattern in the annual reports of firms whose top executives are eventualy prosecuted for fraud?
HT: Marginal Revolution