I have written about this before, but it is so important that it is worth repeating: the cost of capital is forward-looking.
The cost of capital – be it weighted average, debt, preferred stock, or common stock – is the rate of return that investors require to compensate them adequately for the risk of providing that form of capital (buying securities or lending) to a company. Because it is required to induce them to invest, it is – by definition – expected. It reflects their assessments of the risks confronting a business and its ability to generate satisfactory expected returns. Because the future is not certain, there is always uncertainty about any security’s rate of return. Even “riskless” U.S. Treasury Bonds have uncertainties: the real rate of interest, the rate of inflation, maturity risk (which is why longer-term bonds usually have higher yields than shorter-term bonds), and (perhaps during last year’s paralyzing debt ceiling debate) default risk.
If we were working with publicly traded securities, the task of determining the cost of capital would be much easier than it is for private securities. For bonds with fixed coupon interest rates, we can project the amounts and timing of interest and principal payments. Using current market prices, we can then compute their yields to maturity, which are the internal rates of return. That is the cost of public bond capital! If we use the Gordon Growth model, we project dividends (estimated by Wall Street consensus, albeit with more uncertainty) and calculate the cost of capital as the IRR using the current market price as the investment. That is the cost of public preferred and common stock capital!
In the private market, we do not know the value of the stock (or debt, for that matter) so we cannot use the IRR procedure. Instead, we project future returns (dividends or interest and principal) and estimate a discount rate (the cost of capital) to arrive at a (DCF-based) price. However, the discount rate is forward-looking, and not known with certainty.
THE BIG ASSUMPTION WE MAKE, whether using a build-up or CAPM, is that we can USE HISTORICAL DATA, WHICH ARE REALIZED RETURNS, to ESTIMATE EXPECTED RETURNS. If we use a build-up, we use Ibbotson and Duff & Phelps data or perhaps Private Cost of Capital survey data for our risk premiums. If we use CAPM, we look to beta data (!) based on HISTORICAL relative volatility.
I am not going to argue that this is correct, because it is not (will the future REALLY be like the past?), but I believe that this data gets us in the ballpark of reasonable costs of capital. It is the best we can do, since we do not possess crystal balls.
Whenever I am challenged on my cost of capital conclusions, I am willing to give a little (e.g. agree with someone who is within plus or minus 2% of my estimate using the same baseline data such as Ibbotson) because of the inherent uncertainties of estimating the future cost of capital, particularly the company-specific equity risk premium. I do not worry about sophisticated tweaks (such as the mid-year convention, supply-side cost of capital, adjustments to historical data to reflect expectations, and so forth) for the simple reason that they build on fundamentally imperfect estimates of the cost of capital and do not make it any more accurate.
As Clint Eastwood said, “A man’s got to know his limitations.” With regard to estimating the cost of capital, my masculinity is indisputable.