Value and Momentum
If you know academic finance (especially asset pricing), you've heard of the value effect and the momentum effect. If you haven't, the "value effect" is the fact that, over time, "value" stocks outperform "growth" stocks. The distinction between the two is most often determined* by sorting on the ratio between the accounting value of a company (as indicated on its balance sheet) and its value in the stock market. A firm with a stock market value that is many times the accounting value of its equity is a "growth" firm, while a firm with a market value close to or below its accounting value is a "value" firm.
The "momentum effect" is the finding that stocks that have performed well over the past year (especially - oddly - the early part of the last year), which are stylized "winners" generally perform better over the next month than stocks that performed poorly ("losers").
I just found - I'm likely not the first, though I haven't seen this elsewhere - that if you sort stocks simultaneously on value and momentum, the value effect comes from the "loser" stocks and the momentum effect from the "growth" stocks. In other words, loser-growth stocks subsequently underperform loser-value stocks, winner-growth stocks, and loser-growths stocks, which are all fairly comparable (though the momentum effect does seem to exist in all five of my value/growth portfolios - it's just much weaker toward the value end of the spectrum).
I haven't figured out what this means, but this seemed like a good place to share it.
* Other stock price ratios, like price-to-earnings or price-to-sales, give similar results.
inefficient markets
I'd just like to point out that systematic optimism isn't necessary to cause this effect. If people misjudge their abilities generally, even if these errors weren't tilted in one direction, people who underestimate their abilities to succeed as lawyers would avoid law school even more aggressively than they should, but some people would still be too optimistic. Models that assume rational agents are useful in environments in which agents aren't rational, but they require not only that errors not be strongly biased in one direction, but also that the unsystematic errors be able to offset. Assuming rational agents in a model that explains stock prices can work tolerably well if the optimists can be offset by pessimists who can sell short. A model of undifferentiated rational agents is unlikely to explain trading volume, in which traders who are too optimistic and traders who are too pessimistic can both drive volume up together.Whittier College of Law has a 40% unemployment rate for the class of 2011. The bar passage rate is 66%, and the tuition is $38,000. In contrast, Columbia 2011 grads have an unemployment rate of less than 1%, with a tuition of $51,000. Obviously the inputs matter here. Columbia professors aren’t that much better than Whittier professors. Rather, Whittier is probably taking $38,000 a year from individuals who are marginal lawyer material. They’re selling people a dream.
This where cognitive biases come in. A rational person will ask if an individual with a unimpressive LSAT and low G.P.A. and lack of genuine passion for law can be a great lawyer. Most people are quite rational…about other people. When it comes to oneself there’s a strong bias, perhaps for evolutionary reasons, to delude oneself about one’s intelligence, attractiveness, conscientiousness, etc.
