Bruce Greenwald on Valuing a Franchise: Pt. 4

So we’ve made it this far:


The organic growth part is actually quite easy. Again this is increase in sales that requires very little new investment by the company – basically just increased demand. If we were investigating Home Depot, we might just look at same store sales growth. But in his examples, Greenwald often takes a macro approach, saying that this term is going to be GDP growth +/- some percent.

GDP growth might be ~4%. To this we might add (subtract) 1-2% if the company sells to wealthier (poorer) customers. Similarly we might add (subtract) 0.5-1% if the company sells services (goods).

OrganicGrowth = GDP +/- Rich vs. Poor +/- Services vs. Goods

I’m going to give IBM +0.5% for selling to wealthier clients (businesses), and an additional +0.5% for directing their business more and more to selling services. So…

OrganicGrowth = 4% + 0.5% + 0.5% = 5%

Sidebar: do all companies you evaluate get the tailwind of ~4% for organic growth? No – only companies with a real competitive advantage that creates barriers to entry. Otherwise, entry by new competitors will appear to meet increases in demand. For now, I’m assuming IBM has such an economic moat – but will need to verify that.

So, we know IBM is at a P/E of 10, or E/P of 10% and we’ve now found Organic Growth. Let’s fill in more of our expected return.


And we’re storyboarding, remember? So let’s round to the nearest integer to make sure the brain stays with the big picture.


I’m going to do something really lazy here for brevity. For marginal ROC I’m going to go to Morningstar and just grab IBM’s trailing twelve months accounting return on invested capital. It’s ~30%. I’m also going to just assume their cost of capital is ~10%. When Greenwald works an example, it’s clear he’s already read the 10-K and knows roughly what these numbers are. I do it backwards – making rough guesses first, then reading the 10-K to verify later.


We finally have an estimate. ~17%. If we now crack the 10-K and find that all of our guesses at organic growth, return on capital, cash retention policy, etc. appear to be correct, should we go with it?

Hint: There’s an important element that’s always a part of NPV value investing that should also play a part in IRR value investing. Can you guess what it is?

12 thoughts on “Bruce Greenwald on Valuing a Franchise: Pt. 4”

  1. This process of calculating your return on growth all seems logical exempt for the last part where you simply add the 5% organic growth. Irrigardless of what the perfect number is (GDP or otherwise) its the only number in the equation which has NO relation to the price paid. It implies that if I pay 100x earnings or 1x earnings I still get the same % return on my investment from the organic growth. This seems wrong Love to hear your comments thanks

  2. thanks for the comment joe. my understanding is that Greenwald is working from the Gordon growth model:

    P = E / (r – g)

    But then rearranges the equation to solve for return instead:

    r = E/P + g

    so yes, the g is just additive in the equation and you can try it in a spreadsheet using the IRR function to verify. if you buy at a 100x P/E and the yearly growth in earnings is 5%, you do eventually get the 1% + 5% = 6% return.

    now one of bruce’s pet peeves seems to be that even veteran investors will just grow out earnings in valuation models without factoring in the cost required to grow the earnings. you can only use the above added “g” if growth comes for free – which it does in the organic case (but not for active).

  3. thanks for your quick comment
    I was just looking at VISA which is a company that pays out almost 100% of their earnings in buybacks and dividends
    Its screwing with my head because its one of the few companies I have ever encountered that may be able to do 10% organic growth for almost free

  4. But I now see why this makes sense The organic growth has to be “long term” growth – if it was only 3 years at that growth rate then the yield equation no longer makes sense
    Then my question is how long is long term?

  5. hi joe – well “long term” is essentially “forever” since the r = E/P + g equation is for a perpetuity. that being said, i think i remember one student asking bruce a question about this and he had a few interesting points to make:

    (1) the long-term growth rate of any company will probably track that of its industry +/- a percent or two, and the long-term growth rate of an industry will track GDP +/- a percent or two.

    (2) don’t get too hung up on figuring out the exact growth rate. you’re not doing the E[R] equation to figure out what EXACT return you expect to make – rather you’re doing it to make a BINARY decision – does it look obviously cheap or not?

    (3) the people who do well investing in franchises don’t spend time figuring out the exact growth rate (quantitative). rather, they spend time assessing the strength of the franchise (qualitative).

    I’d add maybe 2 more points:

    (A) All E[R]’s are not created equal. One company can have an E[R] = 15% due to a 2% dividend & buyback yield and 13% expected growth, while another’s is comprised of a 10% dividend & buyback yield and 5% expected growth. Obviously the latter seems like a safer bet (a bird in the hand…)

    (B) I don’t have data to prove it, but you can probably do quite well with a lot less work just buying presumed franchises when your starting forward E/P would be greater than 8% or so. Who knows what growth will turn out to be, but as long as it’s positive you should do fine.

  6. What confuses me in this equation is that he is using E/P and ROC together. Shouldn’t it be NOPAT/EV instead of E/P, so that you take into account the whole of the capitalisation which is precisely what ROC tries to do?

    And then, what happens when capex + WK + dividends + buybacks is bigger than NOPAT for an extended period?


  7. As for your first question, I’ve thought about it a lot too. Originally, I though Bruce was just speaking generally, and when he said E/P, he really meant NOPAT/EV. I thought that was fine since there seems to be a spirit of allowing yourself to be a little sloppy here, as despite it being an equation you’re really just ball-parking the #’s to make a binary decision of “obviously cheap” or “not” (with no expectation of making exactly the return you compute).

    As I think about it more, I’m not so sure. We’re taught a variety of valuation methods based solely on dividends, on FCFE or FCFF, and they all depend on assumptions and give different answers. If the company’s capital structure stays the same, then the cash you get is of course (%Div + %BB) x E/P and the cash they retain (after paying interest) and mix with new capital for new projects is the remainder. Fine. Another twist is that cost of capital should be for new ventures, not just the whole firm’s. So if a quite safe company (say Proctor and Gamble) is now investing in Cuba, that cost of capital could be quite high – not their ordinary. It’s easy for things to start getting complicated…

    At the end of the day, maybe it’s easier to stick to low debt companies so that E/P is roughly equal to NOPAT/EV if it really bothers you. As for the latter question, I can only guess that you have to use what you think is sustainable. How’s that for passing the buck?!

    Just to give you a feel, there is one video where Bruce is talking about using this analysis to look at AXP (I think it was) during the financial crisis. I think he said he got an expected return of 40% or something crazy like that, so he knew it was OBVIOUSLY CHEAP, though he didn’t really expect to make exactly 40% per year in perpetuity by buying. Hope that helps.

  8. Bruce a Value Investing God: he can speak in general terms and be sloppy anytime he wants.

    I took the (huge) liberty of sort of correcting Greenwald’s equation, let’s call it the Another-Value-Guy-modified Greenwald equation:

    a. (((dividends + share buybacks + interest)/NOPAT)/EV) + (((capital expenditures – depreciation + investment in working capital)/NOPAT)/EV) * ROIC/WACC + (nominal GDP growth * industry elasticity) = asset return form a franchise investment “ARFI”

    Now this is the ASSET return – you are using EV and ROIC, so no equity returns here yet. You would need to do another step to get to equity returns, which is essentially plugging the result into a classic WACC equation:

    b. ARFI = market cap/EV * expected return on equity + net debt/EV * cost of debt, rearranging:

    c. Expected return on equity = ((ARFI – net debt/EV * cost of debt) * EV)/market cap

    Important disclaimer: this equations assumes that all growth capex is capex on top of depreciation.

    How does that sound?


  9. I think you’ve got a great start to your dissertation under Bruce. I look forward to hearing how well it works once he gives you the key to the Compustat lab.

  10. I believe the calculation is highly flawed, to the point of completely unreliable. I’ll have to research it more in order to determine if this is actually a Greenwald theory. Knowing what I know about him, I find it unlikely that he’d advocate using a P/E yield in any equation because he generally goes out of the way to not include information that is based on irrational assumptions.

    Many studies have been conducted that show that the PE of a stock is a terrible indicator of value. It’s fairly routine for a company to have a high PE and be undervalued. Furthermore, as Buffett explains, price is what you pay, value is what you get. Trying to determine a rationale conclusion about a stock using a PE metric is to say that the market participants consist of rationale minded individuals that comprised the PE of that stock through their daily trading habits. That completely discounts the notion, the the basic tenet of value investing, that Mr. Market is irrational.

    Example, a company that trades at a PE of 1 will produce a much different outcome in the formula than a company trading at a 10 multiple, even though the economics of the business have no changed at all.

    GLRE: .06 + .94(14.91/7.05) + 3.4%
    PE of 1: 5.45%
    PE of 10: 23.88%

    Even if you choose to apply this metric to stable companies such as JNJ, American Express, and others, you are still relying to a great deal on the market participants to produce for you a rational PE. Believing that’s plausible completely dismisses the legitimacy of any value investing principles.

  11. This document ( states:

    “The earnings return or earnings yield is simply the inverted PE (1/(p/e)). Simple enough, but the earnings number to use in the p/e-ratio has to be the adjusted sustainable earnings as used in the EPV-valuation. Say that the p/e- ratio is 14, then the earnings yield is (1/14) 7,1 percent.”

    That makes a lot more sense to me.

  12. Thanks for the comments Jim. If you do want to research this further I’d point you to some of Greenwald’s actual slides here:

    The part I’m referencing starts on the 16th slide (Procedure in Practice) and then the last 3 slides of the deck have some examples.

    I agree that in practice you’d at least want to make sure current earnings are “normal”. While Greenwald uses “P/E” on some slides and “Multiple” on the other, I don’t believe he’s so much giving a formula as expressing the SPIRIT of how you would go about this.

    What earnings yield are you getting at the current price?
    What % of that yield comes straight to you in dividends and buybacks?
    What % does the company hold onto to reinvest, and what is the present value of the future cash flows that will come from that?

    You obviously have to make your own calls on what the current “normal” owner’s earnings are, returns on capital, cost of capital, etc.

    That all sounds like passing the buck and it would be if what you needed to come up with was a PRECISE estimate of your expected return, but you don’t have to know a man’s weight to know if he’s fat.

Leave a Reply

Your email address will not be published.