Here is some Cost of Capital data by industry by Aswath Damodaran.
At Shaun’s request, I removed the spreadsheets that he was attaching. You can find them at Prof. Damodaran’s site at NYU.
Shaun, thanks for the suggestion. I hadn’t seen these spreadsheets before–they’re terrific!
Here are my revised equations now, after doing a lot more thinking and research. And thanks to Marc for showing me that I was thinking about cost of debt all wrong.
$costofdebt: 0.65*Avg(Close(0,##CORPBBB), Close(0,##MORT30Y))/100
$wacc: MktCap/(EV+CashEquivQ)0.08 + DbtTotQ/(EV+CashEquivQ)$costofdebt + PfdDivA/(EV+CashEquivQ)
(I’m using EV + CashEquivQ for the sum of the market values of equity, debt, and preferred stocks.)
$nopat: OpIncTTM*0.65
(Ideally I’d be doing things like amortizing R&D expenses in calculating NOPAT, but heck.)
$eva: $nopat - $wacc*(AstTotQ - IsNA(PayablesQ,0) - IsNA(LiabCurOtherQ,0))
The most valuable thing I’ve learned from all of this is that the cost of equity is invariably greater than the cost of debt. That had never occurred to me before, and it has fundamentally changed my way of thinking about cost of capital. We focus so much here on how bad debt is, but we rarely consider how much cheaper it is than equity!
I’ve been reading some of what Joel Stern wrote about EVA. The man basically invented free cash flow in 1972, and then EVA a few years later. His brilliance is stunning.
As for what I can do with all of this, $eva/(Price*SharesFDQ) seems promising, as does $eva/AstTotQ. This is basically substituting EVA for Net Income in earnings yield and ROA. It seems to conform to the goal that Stern had when he came up with EVA as an alternative to EPS and called it “economic profit.”
What worries me, though, is that I haven’t found anyone else using these ratios . . .
I would multiple the PfdDivA/(EV + CashEquivQ) by say costofdebt + .01 or average of cost of debt and cost of equity
Yuval, I also thought these metrics would be more widely used, but as you have seen, they can be tricky to calculate accurately. I use them a quality measure due to the inherent inaccuracies. The idea is that a company that is adding economic value is well run all other things being equal.
The actual cost of preferred equity is PfdDivA/PfdEquityA, and the portion of capital that is preferred equity is PfdEquityA/(EV + CashEquivQ). So in multiplying the two, you get rid of PfdEquityA.
And that’s a good thing because the numbers in the database are often completely wild. For example, P123 gives PG preferred equity of NEGATIVE 243 million (which is more or less the amount it paid in preferred dividends last year), and GE preferred equity of only 6 million (while it paid out 656 million in dividends).
/quote]What worries me, though, is that I haven’t found anyone else using these ratios . . .
[/quote]
That’s because cost of capital is horrifically hard if not impossible to precisely calculate (as folks here have been seeing) and because so many are so fearful of the crucial art of approximation.
But the formulas I’ve seen in the last few posts look pretty good and quite usable. There is always room to come up with more variations, but I think this thread has now produced some approaches that are good to go. And that means interesting approaches like EVA, or my noise/vale thing, and others that require a cost of capital assumption are open to you.
By the way, once you get going and have some models that make use of this, try fiddling around with cost of capital and re-testing. When I’ve done this, I noticed that the performance of the models was not all that sensitive to the cost of capital formulations – they were more sensitive to getting the balance sheet allocations. In other words, it’s more important to recognize the differences between a 25/75 deb/equity split versus a 65/35, then it is for the the cost of equity for the 75 or 35. I’d be curious to find out if others see the same with your models
Yuval, has the move to factset altered your use of these any?
After 3 years, do you still feel these are as useful in the mico universe?
Thanks
Tony
I still use EVA, but rarely. I’ve changed my definition of NOPAT since then:
$nopat = opincttm-inctaxexpttm-0.35*intexpttm
And I would guess the cost of equity is closer to 0.098 than 0.08 . . .
Otherwise I haven’t messed with the above formulae. Perhaps I should.