With the threat of a US recession quite high, this superb new paper (pdf) by John Campbell and colleagues deserves close attention.
The big message is that stocks that are in distress risk - close to bankruptcy - actually under-perform the market. This flatly contradicts economic theory and common sense, which says greater risk* should carry higher expected returns.
But there's an important subsidiary point - a huge proportion of the risk of failure is in fact unpredictable.
Campbell shows that factors such as profitability, share price returns and volatility, and liquid assets can predict some of the variation in bankruptcy risk across firms. That's what you'd expect.
What you wouldn't expect is that these account for only a small fraction of the variation in distress risk; for 12 month ahead forecasts, the pseudo-R-squared is a mere 11.4% (Table 4). Apparently, this is better than other measures, such as Altman's Z-score, whose predictive power has been criticized.
Many distressed companies survive. And some apparently sound ones fail.
I suspect this corroborates one of Paul Ormerod's findings. He's shown (pdf) that the pattern of corporate failures looks very like the pattern for the extinction of species; it's a power law. This in turn implies that there could be a big element of randomness and unpredictability in corporate failure. Maybe investors can no more foresee the demise of firms than dinosaurs could foresee their own extinction.
* You might think it's only non-diversifiable risk that should carry higher returns. That's what they teach at business school. But it ain't necessarily (pdf) so.
The z-score isn't a predictive model; it's a classification model. It classifies companies into high and low risk of financial distress. There is a big difference between this and prediction of financial distress and it does not surprise me that a measure which is developed for one purpose doesn't perform well at another. I think the author of the paper on z-scores is quite confused and hasn't really thought at all about the meaning of the statistical tools he's using.
btw, I can't quite agree that "common sense" says that stocks with poor profitability and shonky balance sheets would be expected to outperform.
Posted by: dsquared | August 04, 2006 at 08:45 AM