Richard Beddard not only endorses a condemnation of CAPM, but he wants to know if I am reading. I say it is the worst valuation model, apart from all the others.

CAPM is not perfect. Deriving the formula needs one dubious assumption (that investors have logarithmic utility curves). The assumption is probably roughly right, and, therefore, so is CAPM. This suggests that we probably need a more complex adjustment for risk than the simple linear one in CAPM.

However, the CAPM formula is roughly right. It is possible to derive a very similar (but less usable) formula without any problematic assumptions (using a discreet time approach). In addition, the major alternative, arbitrage pricing theory, is also very similar in some ways (only with multiple betas to estimate).

The as yet to be discovered “correct” model probably:

1. is significantly more complex than CAPM
2. has more parameters that change over time (appetites for risk change in more complex ways than simply increasing and decreasing),
3. has a non-linear relationship between risk and return
4. is still some sort of DCF.

Until then what better alternative is there to CAPM? Where do we get our discount rates from?

There is certainly a positive correlation relationship between risk and return, so CAPM at least gives us an approximation for that, until something better comes along.

In practice investors may rely on ratios and screens, and may be under the impression that they have avoided the issues with the CAPM. The problem is that all these models have even less accurate methods for adjusting for risk. This is very often on the lines of, assume the same risk as other companies in the sector, and add a fudge factor for what I think about it (with a sector relative PE, for example).

Finally, Richard mentions Jame’s Montier’s blog and book (from which he got the comments on CAPM). The blog excellent and I have added it to my feed reader.