Nobel prizes notwithstanding, the transition from theoretical to real world markets makes the CAPM a little less obviously rational as an investment tool.
The question of whether a given investment strategy is rational is not nearly as banal as it looks on the surface. Superficially, it might be as ‘simple’ as answering the question of whether that investment strategy maximizes expected utility given unbiased subjective probabilities. However, deeper epistemological issues infest this seemingly tidy and clean-cut notion of rationality when it comes to assumptions buried within the strategy which are of unknown accuracy or which are known to be false. A second class of issues centers on the notion of ‘expected’ utility and its relationship to actual, real world, unpredictable, waiting-to-be-discovered utility. The first class of issues can be summed up with the question “is it rational if it gives approximately right answers but for incomplete or wrong reasons?”, while the second class of issues can be summed up with the question “is it rational if its outcomes usually work out well in the real world but sometimes fail miserably?”.
Case in point: the capital asset pricing model (CAPM).
Although it works nicely in an idealized, single-shot, non-dynamical model, in reality we know that markets change over time, we know that beta is not a good proxy for returns, we know that past performance does not guarantee (or even place statistically reliable bounds on) future performance, we know that risk is not symmetrical (i.e., risk is not mere variance, it is the probability of losing money), and we know that not only does correlation not remain constant over time but it also increases strongly during significant market downturns.
Therefore, it is an empirical question — not a matter of a priori reasoning — as to whether CAPM actually does maximize expected utility over any relevant (non-instantaneous) timeframe in any relevant real world (non-static) market, or whether the mathematical notion of expected utility at a single snapshot in time has any useful bearing on real world actual utility as it develops over time. Indeed, historical evidence from sharp market declines suggests that the CAPM’s reliance on historical measures of correlation and volatility as (faulty) predictors of future risk may do a very poor job of maximizing actual utility.
This alone would not be a deal-killer for the rationality of employing CAPM in making investment decisions. After all, we human agents routinely employ quick heuristics which yield approximately correct answers in place of more time-consuming but analytically robust methods which give correct answers. It makes sense to do this in situations where the cost of small inaccuracies is demonstrably lower than the cost of eliminating those accuracies. So we might ask instead: is it rational to employ an investment strategy which we know to be based on false premises? Provided that the outcome of doing so is good enough, does it matter that it’s not as good as it could be? And is it rational to employ an investment strategy which protects us against theoretical risk but which we know does not protect us against some kinds of real risk and can in fact open us to severe risk in the event of dynamics taking the market in directions with radically different parameters (such as correlation) than those specified in the original temporal snapshot’s boundary conditions?
Once we move out here to the real world, could it be that the burden of proof actually lies with those who view the CAPM as paradigmatically rational?
All material on this site is carefully reviewed, but its accuracy cannot be guaranteed; please do your own checking and verifying. This specific article was last reviewed or updated by Dr Greg Mulhauser on .