Return on Capital vs Cost of capital

Hi all,

I am trying to generate a screen which has a few filters including Return on Capital (ROC) > Weighted Average Cost of capital (WACC)

There seem to be many ways to approximate each side of this inequality an it is driving me mad (I am not an accountant!)
As usual there is no single truth but your inputs are welcomed to help me get to a point where at least both sides of the equation are consistent with each other.

  1. For the WACC, I can use the 2 approaches proposed by Marc G. which I reattach here for convenience
    1.a) CAPM approach
    1.b) GLS approach

First, the GLS approach is sensitive to the universe used in the ranking needed for the cost of equity. So we get different results for the same stock depending on whether one uses say the Nasdaq100 or the R1000. Since I am only looking at large caps stocks, I decided to use the R1000. Large enough universe but not distorted by smaller caps. Any other opinion?

Second, the results of 1a and 1b are -as one could expect- different. I am thinking of using the average.

What do you think?

  1. For the denominator of the ROC (i.e. the invested capital), there seem to be many definitions around including the one kindly proposed by Yuval in another thread:

Looking at wikipedia, investopedia and others, it looks like everyone agrees with the starting point:

Invested Capital = Total assets (or total liabilities) - current liabilities.

Then comes the variations around what else to subtract:

  • Cash and Equivalent (broad consensus or at least subtract 80% of cash and equivalent)
  • Intangibles (Yuval or Marc G’s interpretation of Greenblatt but not wikipedia)
  • Other current Assets (Yuval or Marc G’s interpretation of Greenblatt but not wikipedia)
  • Inventory
  • Receivables

Any view?

I would tend to go for: Invested Capital = Total assets (or total liabilities) - current liabilities - Cash & Eq - Intangibles - Other current assets.
(Capital has been used to “produce” Inventory and Receivables so it might be a better approx to keep them in)
(I remove other current assets as nobody knows what has been put in it…)

Thank you

Jerome


Supplement2 Cost of Capital.pdf (121 KB)


Supplement4 Cost of Capital.pdf (109 KB)

For what it’s worth, here’s my formula for the weighted average cost of capital:

(0.08sharesfdqprice+(0.65*(close(0,##corpbbb)+close(0,##mort30y))/200)dbttotq+isna(pfddiva,0))/max(5,pricesharesfdq+dbttotq+isna(noncontrolintq,0)+isna(pfdequityq,0))

I think the easiest way to deal with the cost of equity is simply to guess what it is and apply it across the board to all stocks. 8% seems about right these days. You just have to make sure it exceeds the cost of debt. For that, I average the rate of BBB bonds and the thirty-year mortgage. The 0.65 multiplier is because I’m estimating tax savings of 35% on interest paid, but that might be totally off.

The challenge of cost of capital is that the number is equal to whatever you want it to be subject to

(1) the common sense proviso that none of the capital costs can be negative (that wipes out empirical attempts to determine cost of equity because while ex ante cost of equity must be positive, ex post reality gives us occasions when it didn’t turn out that way), and based on risk profile . . .

(2) cost corp. debt > risk free rate and cost of preferred > cost of corp. debt and cost of common > cost of preferred

If you try to use a fancier formula, be careful about interest exp/debt. You can get some very odd numbers popping up as debt fluctuates throughout the year (and ttm debt may not capture it, and as their are kinds of interest expense that don’t show up cleanly inthe main part of the financials. So code in some limits and over-rides.

My suggestion: Don’t overthink it. If your model is highly sensitive to the cost of capital assumptions, chances are you have other problems that need to be solved.

Flat across the board assumptions are fine but be careful about backtesting; tate environments change so a flat assumption that looks good today may be out of line in past test periods and/or may prove troublesome later on. Try to peg you across the board assumptions to something that can vary programmatically over time, such as SP 500 cap weighted div yld or cap an aggregate of interest or a Fed rate you pull in via GetSeries. Again, no need to overthink; just have ananchor that can more up or down with market conditions.

The GLS is an example of how you can creatively push the envelope if you’re into that sort of thing.

Here’s an excellent paper on calculating return on capital in a very sophisticated manner:

https://research-doc.credit-suisse.com/docView?language=ENG&format=PDF&source_id=csplusresearchcp&document_id=806230540&serialid=C0owv4XbV7zL%2BTQLggWqjPthH7IUpSwUZpiIwdvDgtA%3D

Doing this in P123 will be a significant challenge, but it might be worth some effort.

CAPM. More like CRAPM.

That’s just my opinion. Very smart people disagree because they look at the markets differently.

Beta makes sense if you look at markets, very high level, as a a linear combination of asset classse’ risks vs rewards. However, when you begin to account for rich and non-linear understandings, including intuitions born out behavioral economics, the sufficiency of linear normative models fall flat.

I feel that one’s effort can be better spent on refining approaches other than CAPM. GLS looks like a sensible approach. The only good use case for using CAPM is when prices are the only inputs available.

Thank you all.

Any view on Invested Capital? How come it seems to differ so much?

Jerome

My view is to match the numerators and denominators according to money flows. For example, GAAP earnings (before preferred divs and NCIs) flow to all classes of equity; the denominator should therefore represent all equity classes. EBIT flows to all owners, including contingent owners; therefore all capitalized owners should be represented in the denominator.

The same thought process can be applied to every flavor of cash flow.

The secondary decision is how to represent capitalized versus non-capitalized forms of financial interest. Sometimes it’s better to go the indirect approach (e.g., start with total assets and subtract the operating liabilities). Other times, it’s better to go the direct route(e.g., sum up debt, equity, and contingent ownership).

There’s not one right ROIC, but there are clear lines between logical and illogical attempts at it.

Yuval, did you (or anyone) ever attempt to replicate this in P123?

https://research-doc.credit-suisse.com/docView?language=ENG&format=PDF&source_id=csplusresearchcp&document_id=806230540&serialid=C0owv4XbV7zL%2BTQLggWqjPthH7IUpSwUZpiIwdvDgtA%3D

I used to use a simpler version of ROIC, which is (OpIncTTM - IncTaxExpTTM - 0.35*IntExpTTM) / (DbtTotQ + EqTotQ).

Mauboussin actually relies more on CFROI, or cash-flow return on investment, which he describes in Appendix A of this paper. My version of CFROI was basically unlevered free cash flow divided by net operating assets. So, for instance, you could use:

(OperCashFlTTM -CapExTTM + IntExpTTM * (1 - TaxRate%TTMInd / 100)) / Max (1, DbtTotQ + IsNA (NoncontrolIntQ, 0) + IsNA (PfdEquityQ, 0) + ComEqQ - IsNA (CashEquivQ, 0))

(You could argue that instead of subtracting CapExTTM for free cash flow you should add CashFrInvestTTM. I’m on the fence about that.)

This wouldn’t work for companies in the financial sector. Instead you’d want to use CFROE, which is levered free cash flow divided by book value. To get levered free cash flow, you don’t add back the untaxed interest expense.

In a different paper, I believe Mauboussin also recommends subtracting the industry’s weighted cost of capital from the CFROI and CFROE. And calculating that is a real bear.

To calculate ROIC strictly according to Mauboussin’s formulas requires access to some line items that we do not currently offer. But stay tuned. We’re planning to make this much easier . . .

Yuval, do we have any idea on when this might be “much easier”?
I’m working on something now that needs a good ROIC figure but if it looks like it may be a while, I will go to my backup source for this metric.

Thanks
Tony

If you want a really in-depth treatment on ROIC, NOPAT and Valuation, I highly recommend the book “Valuation” by David Wessels, et. al. David Wessels was one of my professors at Wharton, and is a former McKinsey consultant. There are a lot of nice tidbits in the book and you get an intuitive understanding of the pros/cons of refinements to the standard approach to calculating Invested Capital, NOPAT, etc.

The CS research piece is a great document as it contains some very familiar material. You can get an older edition of the Valuation book for a reasonable sum. I’ve designed factors based on some of the core principles in the book and they’ve served me well over time.

Is this still on the P123 to-do list?

Yes, it is.