Less Theory, More Practice: How to Value a Stock Using a “Sustainable Growth” Model Coupled with Return on Capital

by Geoff Gannon

I get a lot of questions about valuing a company based on estimates of its sustainable growth rate. So, someone will say it seems this company can grow 8% to 10% a year and it has a return on equity of 15% or 20% a year – and they’ll ask me: “how do I determine what this stock is worth?”


My answer is usually pretty much the same. Unless you know the business very well, that kind of estimate is starting way too far down the income statement.


So what’s a better way of tackling the same sort of valuation problem using simpler math, more explicit assumptions, and better tying your model of this specific company into the way the wider world is likely to shake out over 5, 10, or 20 years?


Let’s start with a huge bit of simplifying math – something I encourage every investor to make rule #1 of any growth model.


It's much easier to pick a certain point in time: 5 years from now, 10 years from now, 15 years from now, etc. and do a point-to-point calculation instead of a sustainable "CAGR" calculation. Why? Well, you would have to decide on whether things like margins stay the same. So, do gross margins stay the same, expand, or contract over time? Do operating expenses decrease or increase over time? So, does the same gross margin percentage convert into more or less operating margin over time? With a point-to-point calculation you can assume gross profit, EBITDA, net income, etc. will grow faster or slower than sales. You can’t really do that with any kind of permanent projection. I mean, you can – and I’ve seen people do it. But, it makes no sense to say I think this company will “sustainably” grow earnings faster than sales.


A surprising number of people ask whether the calculation should be in real or nominal terms. Here’s the thing – for most businesses, inflation makes quite a difference. And investors often aren’t that sensitive to differences between units sold, nominal price per unit, and real price per unit. When I look at the long-term history of a company, I always do some calculations in real terms and some in nominal terms.


As far as whether the growth rate is sustainable or real, for many of the companies I like to look at - you could do the calculation in real terms. However, that doesn't work for most companies. Most companies would get worse real returns if inflation was higher. Not all. There are exceptions. It's a timing issue. So, asset light businesses (that don't carry much inventory, that collect bills they are owed before paying the bills they owe, companies that raise prices on customers ahead of price increases from suppliers, etc.) would be more able to grow similar in real terms. Like, NACCO and BWX Technologies and Omnicom and Keweenaw Land Association (a timber company) should all be able to have a sustainable "real" rate of growth. That's not true for most companies. For example, you can see in the period from about 1965-1982 when stock multiples were contracting in the U.S. that more and more large U.S. businesses were producing very bad real returns on equity. Inflation was a problem for them. 

How I model out a company's future sustainable growth is:


1. Do a point-to-point calculation. This is a reality check. So, if someone sends me something saying Google can grow earnings per share by 15% a year indefinitely or something - I'd say, let's pick a specific year and model that out. Either earnings have to grow much faster than sales, or sales growth has to come from things that aren't advertising, or the ad industry has to grow much faster than it did in the past, or you have to pick an end point for the year you are measuring to that is very near today - or, you'd end up with Google having a huge share of global advertising spending. This is because your model is basically going to require Google to grow revenue from ads faster than ad spending, so market share will grow year after year. This is the reason you do point-to-point calculations using specific years. If you don't, there are people who don't realize how much of the ad industry Google would have in 2038 if you're really doing a 20 year projection, Booking would have of the travel industry, etc. 


2. You want to focus on the number you care about (total return in the stock) and break it down from there. So, for example, it doesn't really matter much if Omnicom gets you a return by paying you a dividend, buying back stock, or acquiring more sales. Organic growth is - because the company is asset light - very cheap (pretty much free). So, that's different. But, you want to model it out sort of like:


Ad Spending Growth +/- Market Shares Gains and Losses +/- Shares Outstanding Increase or Decrease +/- Dividend Yield


Also, for an investor you always have to pick the ending multiple on the stock. You can pick a multiple that's EV/EBITDA, P/E, P/B, P/FCF, EV/Sales, etc. Whatever you want. But, this is an important part of the calculation because a fast growing business will eventually be a slow grower. So, if Booking or Google grows quickly for 10-20 years, that's certainly possible. But, it can't really have a much above market P/E ratio at the end of 20 years because if it does grow fast for 20 years, there just won't be growth opportunities left. The longer and faster a company grows, the harder it’ll eventually be to have an above market P/E ratio (because the sooner it’ll be 100% mature).


I don't do any calculations that involve the "reinvestment" rate or returns on capital.


Why not?


Isn't that important?


Well, it's very important. But, it's possible to work out a simple formula that incorporates reinvestment without doing any sort of actual ROC calculation.


Project a certain growth rate. Then, determine how much you think the company will have to retain, spend in the business, etc. to achieve it. The past record is very helpful in this calculation. It’s especially helpful over longer periods of time – like 5 years, or an entire business cycle or something.


So, with Omnicom the organic growth rate is free because they have "float". The faster they grow organically this year, the more cash flow they get now. So, if you calculate Omnicom will grow sales by 0%, 1%, 2%, 4%, or anything else - it doesn't really matter. You can simplify your assumptions by just assuming Omnicom always has all of FCF available to issue stock options, make acquisitions, buyback stock, and pay dividends regardless of how fast they grow.


Similarly, with banks we said something like Bank of Hawaii can grow 3% a year forever while paying out 100% of earnings (in dividends, buybacks, etc.) If BOH grows much faster than 3%, it then needs to retain earnings to keep its capital levels in line with the historical norm. But, we don't really think BOH will grow fast. So, we can just assume BOH will - at a constant P/E multiple - return earnings yield + 3% as a stock. All of the earnings can be paid in dividends or buybacks.


For companies with significant amounts of working capital, PP&E, etc. this changes.


But, again, it's possible to incorporate the "incremental return on investment" purely in terms of the "cost of growth". Which is how I always do it.


This confuses people. But, there's an excellent reason for why I do it.


I'm not purely a value investor or purely a quality investor. I will consider companies with 30%+ after-tax returns on equity and also companies with 5% returns on equity if they are incredibly cheap. Some value investors would never consider a company with a 5% return on equity, but that's not the right decision. Actually, as long as a company with a 5% return on equity isn't going to plow any of its cash flow back into the business - it could be a good investment at the right price.


For example, let's say I find a net-net trading at about 2/3 of its NCAV. Historically, the company has returned something like 5% to 10% on NCAV. But, say it has returned 10% on NCAV if you exclude net cash. In other words, it might be earning 10% on just receivables plus inventory less accounts payable, accrued expenses, etc. Assume here that PP&E is unimportant. This isn't a weird example. I won't say the company's name. But, this is basically a "live" example of a net-net that exists today. So, this is the kind of problem an investor would actually face: Should I buy this net-net or not?


Is it a good investment?


The quality oriented investors would say "no". In the long-run it will return between 5% and 10% a year depending on capital allocation at the company. It's a cigar butt.


Actually - it depends. The company is more attractive the more it does two things:


1) Returns cash in dividends, buybacks, borrows money, etc. (that is, the more it uses financial engineering)

2) The less it grows


The reason here is that a low ROE is really just a high cost of growth. It only becomes a problem if the company tries to grow. The fact past owners of this stock funded a lot of slow, low-return growth doesn’t matter to you – the investor who buys in today. If the company has mostly decided it doesn't want to grow and you can buy it at a nice price relative to cash on hand, free cash flow yield in an average year, etc. - the ROE doesn't really matter.


Likewise, how much does the ROE of an amazing company matter?


It matters a lot if the company grows. But, less so if it doesn't.


My own view is that thinking about ROE in terms of a "return" on your money is the wrong way of doing it. What you get is growth. What it costs to fund growth - from a shareholder's perspective - is an incremental addition to equity. So, I don't really think in terms of ROE or incremental ROE. I think in terms of:


1) How much can this company grow?

2) How much owner money has to be put up to fund this growth?

Imagine a timber company. Well, if the amount of timber per acre can grow 4% a year without owners putting in more money – then, that’s an important fact to know regardless of what the company reports in earnings. The economic value of the business grew 4% without you doing anything. But, consider if you grow the amount of timber the company owns by buying 4% more acres this year. Well, you are paying for that. So, the price paid per acre of timberland becomes incredibly important to the analysis if the company is buying more timberland. If the company isn’t buying more timberland, the growth rate of the trees matters – not how much the company once paid for the land. In both cases, it’s “cost of growth” that matters. In one case, cost of growth will look a lot like ROE – if the company is growing through buying more and more timberland each year – but, in the other case the “cost of growth” will have no relation to ROE.


Lately, Omnicom has grown at about 2% a year. That's all you get regardless of how high the return on capital is. You could try to calculate it as a re-investment rate (though, in Omnicom's case - the reinvestment rate calculation result would be nonsensical, because incremental capital would sometimes be nil or less than nil so the ROIIC you'd get would be "NMF" or negative). I think it's better to think of Omnicom as: well, does it shrink 1% a year, grow 1% a year, grow 3% a year etc. and then is that growth free.


In the case of a low ROE type business - yes, you could do a reinvestment rate calculation too. But, again, I think it's not the best way of thinking about it.


Why not?


I've said before that I think the things you need to focus on when analyzing a stock are those things that are:


- Constant

- Consequential

- Calculable


Return on invested capital does fit that bill at some companies at certain phases in their history. So, you could do the calculation for Cheesecake Factory today or Howden Joinery. Basically, it's a calculation on new store openings. It used to matter more at those companies (when they were growing store count faster as a percent of their existing store base). But, it's still a "consequential" number and because it's a repeated store model it's definitely a long-term "constant" number. It's also easily calculable. In fact, I'm sure management at these companies has some sort of payback period or ROI targets (probably in cash terms) for new store openings.


But, at a lot of companies the return on incremental invested capital and the reinvestment rate isn't really going to fit the "constant, consequential, and calculable" test. At cyclical companies, manufacturers, companies carrying a lot of working capital, etc. the number will - at least in cash terms - fly all over the place. In fact, capital will sometimes flow out of the business. In fact, in the real world, that's often how a company gets its ROE up. It gets very miserly about using capital in that business, it runs down inventory, it improves receivables collection, etc. 


If I created an Excel that used the theoretically correct idea of reinvestment rate and return on incremental capital - it'd be all over the place.


But, if I pick a point in time that's 5 years, 10 years, or 15 years in the future and I assume a growth rate in sales, gross profit, EBITDA, etc. of 3% or 6% or 9%, I can often come up with a reasonably good approximation of how much owner money I think would need to be retained to hit that number.


Remember, it won't be that useful to know the ROIC if you get the growth number completely wrong. So, if ROIC is 50% and growth comes in at 4% a year over the next 10 years instead of 8% a year as you expected - your ending valuation will be off by quite a bit.


I like looking at sales and gross profit most because this can be more easily tied to real world things like population growth, inflation, nominal GDP growth, industry growth relative to the economy, market share, increases in spending per capita on something, etc. Like, you can see if the predictions people are making for this specific company could fit with a likely future reality.


Store growth is a good example of that. You can say new stores will grow by 8% a year for the next 15 years. That sounds reasonable. But, for some U.S. restaurants, that would mean more locations than any full service restaurant has now. That seems unlikely. It might not be impossible. But, that's a red flag. So, you can look at what kind of saturation we're talking about.


It's also a good idea to focus on numbers that don't move around a lot which are things like store level economics, "turns" of inventory and receivables, and income statement lines that are near the top. So, sales and gross profit. Of course, investors care most about the bottom line. But, for most companies, the bottom line is much harder to predict because of economies and diseconomies of scale, business cycles, etc. 


The more constant and predictable numbers to base your decisions on will be:


* Typical store level economics

* Number of stores

* Number of customers

* Number of units

* Sales relative to those things

* Gross profits relative to those things


And then: cash conversion cycle, asset intensity, etc.


Things like operating margins are long-term very difficult to model out without knowing what growth rates will be, because these are the things that can widen out a great deal with scale. It would be difficult to know what the eventual EBITDA margin could be at Facebook or National Cinemedia or something without knowing if the business is already at about 100% of the potential in terms of audience and ads. The numbers lower down the income statement would look very different in the future if Facebook was on a planet with 80 billion people instead of 8 billion people or if National Cinemedia wasn't already in like 50%+ of all movie screens. 


Compare Facebook and Twitter for instance.


Facebook is a success now. As long as it stays a success, that’s easy to model even if you use the bottom line instead of the factors I suggest. But, what about Twitter? Something like that is more difficult things to model out. But, you'd still do it the same way. You'd still say how big is Twitter's audience? How many ads will that audience see? How much could those ads be sold for? If you don't do it that way - if you just try to project out based on bottom line financials as of today - you'll value a business that isn't yet having financial success like it never could have success. Return on capital comes into when asking how much would it cost to grow audience, grow ads served to that audience, or grow prices paid for those ads? Is it a big number? A very small number? That’s what’s really going to determine future returns on capital. It would be hard to see that looking at today’s results though.


That gets back to the net-net type business. I see this all the time with tiny banks selling below book value. Investors assume that because they are earning a 5% ROE today they are worth a huge discount to book value. In reality, if a tiny bank could either have a much higher ROE once they grew scale or once they were bought and plugged into another bank - then they could ALREADY be worth book value to someone who is truly future oriented. Really, what you'd care about is the quantity and quality of the deposits. A one bank branch will never earn a good return on equity as a standalone business - but, one branch is obviously worth book value or more to an acquirer. You could say it’s speculative to assume a business would be worth more to an acquirer. But, it’s also speculative to assume a profitable business will never increase scale. A static ROE assumption for a business with increasing returns to scale doesn’t make sense. Of course, for most industries scale is important to a point and much less important beyond that point. A bank with $100 million in assets is likely to be less efficient than one with $1 billion in assets or $10 billion in assets. There’s much less evidence – at least on the cost side – of $100 billion in assets or $1 trillion in assets getting you even better returns. Yes, I can name some banks with great economics at that size. But, I can also name a couple with similar economics at much smaller size. What they have in common isn’t overall size, it’s the percentages of deposits in each city, it’s the deposits per branch they have, and it’s the type of depositors they have. So, again a “point-to-point” analysis still makes sense. The difference between the economics of 1 branch and 10 branches is big in a way that the difference between 500 and 5,000 isn’t.


That's why I would focus on modeling out the sort of basic, easy to connect to the real world, and close to the "top line" numbers like deposits, store count, audience, etc. and work from there instead of trying to say I think the sustainable EBITDA growth rate here is 10% a year. An EBITDA growth rate is always making assumptions about all that stuff I listed above. So, if you are wrong in your assumptions, you will be way off on things like EBITDA growth, return on INCREMENTAL capital, etc. Whereas if you break down your assumptions about the company's future growth into a set of 3 or 5 key factors: number of branches, deposits per branch, cost of deposits, etc. - and you pick a specific point in time (like 2023 when I think the Fed Funds Rate will be "X") then you are on much firmer footing.


Often, this kind of analysis will match up nicely with real world common sense. Like, a company doubling the number of stores in the same town is going to grow earnings faster than sales while a company increasing its nationwide store count by 10% through filling in the most rural places it hasn’t yet reached is not going to drive any sort of earnings growth beyond store count growth.


Most importantly, I think that kind of point-to-point calculation based on 3-5 key factors for the business will also give you a better understanding of the business, the challenges it faces, etc.


For example, it will focus you on the "key constraint", the bottleneck that is most holding the company back. I just did a podcast on Tandy and talked about how I thought the lack of ability to add a lot of good, new store managers meant the key constraint for that company is store growth. So, I needed to see either an increase in store SIZE (which would still only require 1 store manager but would do more sales) and a decrease in SHARE COUNT (which would increase the amount of sales per share for investors without needing to increase the number of managers) to give me confidence in buying the stock.


So, when you break a business down that way you can come up with possible solutions. How can Tandy give investors a good total return?


1) They can find a way to attract more excellent store managers

2) They can grow sales faster than they grow store managers

3) They can grow sales per share faster than they grow sales


And we get to that kind of answer just by breaking the business down by key factors like number of stores, square footage per store, sales per square foot, and number of shares outstanding.


Start Getting Geoff’s Sunday Morning Memo