CAPM and Cost of Capital

Although CAPM is imperfect, it is still the best model out there when you consider accuracy in conjunction with practicality.

And CAPM is only used to estimate cost of equity capital. This is combined with the cost of debt capital to find an overall weighted average cost of capital. That's how the effects of leverage are taken into account.

Other factors that you mentioned would probably fall under company-specific risk, which can be adjusted through (more or less arbitrary) risk premiums tacked onto the cost of equity.

qoute "Where does CAPM take into account liquidity, leverage, threat of substitutes etc? I guess my question is, is there a different way to estimate the cost of capital?"

Those are unsystematic risks. The CAPM measures only systematic risk (as measured by beta) because that is supposed to be the only risk that the market rewards. The other risks are (supposedly) diversifiable and are thus not rewarded. Right or wrong, that's the basic rationale.

Also, leverage impacts beta.

Other alternative models may be better explaining the past, but its predictive power is less than CAPM. In, 1991 Famma and French found out that only 8% of the returns are explained by beta, which gives us an idea to dump the model. Nevertheless, the reason why it still persists is because of its simplicity and nobody found out a more better model than CAPM. When you’re predicting the future, by definition you will be wrong 100%. I’d rather screw up using only one variable, which is beta, rather than bringing on other more variables and screw up.

I recently read this article:

http://harvardbusinessonline.hbsp.harvard.edu/b01/en/common/item_detail…

Found it to be interesting.

Do a google search and you can locate a free pdf of it fairly easily. I'd post the link but as a result of copyright issues do not want to get WSO in trouble.

I should have been more specific. I know that CAPM only estimates the cost of equity and the WACC estimates the cost of capital. I didn't think about cost of debt, that of course includes leverage etc, which makes much more sense.

Two questions:

1) Doesn't Beta include leverage as well beacuase we use leveraged Beta.

2) The reason I was asking this originally was because we just covered Famma and French and the predictive power. Since CAPM does a poor job, could one theoretically use the Famma and French model, which includes size, momentum (I think), etc to calculate cost of equity.

Beta includes leverage because more leverage affects the cost of equity also, not only cost of debt.

The problem of using the proxy model stems from the basic shortcoming of regression we’ve learned at stat101. For example, let’s say a study found at that children who had a bigger feet turned out to know a lot more vocabularies than who had smaller feet. Can we say that having bigger feet is the reason for children to know a lot of vocabularies? Absolutely the idea is false. A lot of empirical studies have risks of extrapolating causation out of regression results, but we tend to overlook the confounding factors. There is exactly the same problem with the proxy model. Extracting risk from the studies contains tremendous leap of faith.

The Fama-French Model (APT) adds size and value effects, explaining an additional 8 or so percent of returns. Another guy added a liquidity factor, which purportedly accounts for another 4 percent. There is also the build-up approach, which is quick and dirty. Add the YTM on the company's debt plus a small premium. Problem is you have to estimate the premium. I think throwing in more variables just gives you a false sense of precision, when, as someone pointed out, none of the models are exact because they assume a fixed perpetual covariance with the market.