The Capital Asset Pricing Model Is a Joke if You Ask Me: (Used to Calculate the Cost of Equity). Here is the formula and I will cover every variable Ke = Rf + B(Rp). I will start will B or Beta because the risk free variable is not necessary
Beta - Is a measure of Volatility measured against a benchmark, or more specifically a linear regression of the daily returns of an equity measured against the daily returns of one of the major Indices. But why is the assumption made that volatility is a bad thing? Small and mid-caps necessarily have a higher beta due to a smaller share count (generally speaking). Especially in the case of small caps who may be thinly traded, thus causing small purchases or sales to dramatically influence the price. I find it odd that especially now that the lower Beta stocks i.e large money center banks, GE, etc turned out to be the most risky. They weren't even solvent! Briefly getting into specifics, how can you run a regression of an equity against and index when that equity is part of that index?? Even is Beta accurately measured risk, it would still be distorted by that fact because part of that regression is necessarily 1:1
Rp - Risk Premium - The mainstream thought on this to input what the usual market average return will be or the expected market return for a given year. But how does that have any correlation to the "risk" of equity? It's utter nonsense.
So what are some alternatives? Well basic short term liquidity and Solvency ratios, which the banks would have failed. Two other great ratios to use are the current ratio (current assets to current liabilities) and TIE ( times interest earned). Although TIE has to do with debt, you want to know if a firm has viable capital structure. So the higher the TIE, the less risk of default. Some others are operating cash flow i.e free cash flow to liabilities, free cash flow as a % of net income.
ALTMAN-Z (which has a great track record for detecting companies that would go bankrupt) and other alternatives to this model. Another model is the Beneish Model with calculates the probability of accounting manipulation (and like the Z-score, has a rather good track record).
If anyone cares to know the formuals of these models just make a comment. BTW, the Beneish model had a 78% accuracy rate over a ten year period (I believe 1992-2003). It is calculated Enron has a 100% chance of accounting manipulation