CAPM Fail?

I came across something funny in my stock screening this morning when I noticed something was odd with the discount rate that my (automated) valuation model was using for one particular stock, APOL. It was incredibly low, implying that APOL was a very safe investment.

APOL may very well be a good buy. I don’t currently own it, but it certainly screens well, with historically predictable yearly increases in sales, net income, and free cash flow. High ROE, current ratio, interest coverage ratio, etc. Basically it passes the test of almost everything I normally screen for.

But considering that the SEC has launched an informal inquiry into its accounting practices and that there has been the stray negative article regarding the usefullness of for-profit education in general, it seemed odd to me that APOL could be considered (by my computer) as an ultra-safe investment.

The problem, it turned out, was that during some experimentation I’d switched out my model’s standard discount rate of 14-16%, with the Capital Asset Pricing Model instead. And I’d forgotten to switch it back.

I’ve grumbled before about not liking CAPM, but there’s one thing it maybe captures well. If you’re trying to value many stocks quickly on the fly, the stocks with a history of large gaps up and down in price during quarterly earnings will naturally have a larger β, and therefore your valuation model will use a larger required rate of return (Ri) as it probably should.

Ri = Rf + β x (RM – Rf)

The problem was, my computer was basically using:

Ri= Rf

Or in other words, the future cash flows of APOL were not at all tied to what the stock market (RM) may do, but were as predictable and certain as the future cash flows of 3-month US treasury bills! (Rf)

A quick visit to Yahoo Finance’s Key Statistics Page for APOL showed why:

APOL's Beta

Which can sort of be verified by inspection:


And so I’m supposed to discount APOL’s future cash flows by Rf, which is currently about 0.13%?! I don’t think so!

In the end, I decided to sort of fuse CAPM with my own usually fixed 15% discount rate (a little higher for small caps, a little lower for large caps). What I came up with was:

Ri = 14% + β

Not perfect but should keep me closer to the conservative 14-16% range for all the quick daily valuation screens. Then when I spreadsheet all the financial data for the rare stock that passes all my initial quick tests, I can tweak the model’s knobs to fine-tune the valuation.