Geoff Gannon May 16, 2022

Don’t Just “Over-Maximize” One Variable – Find Stocks That Tick a Lot of Boxes at Once

On the day of this year’s Berkshire Hathaway annual shareholder meeting, Andrew and I were in Omaha at a Willow Oak event. Willow Oak is the company that provides a lot of the administrative support functions for our fund. The event was a panel which Andrew moderated and which featured four managers of funds associated with Willow Oak. I was one of those managers. And one of the questions we were asked had to do with what we’ve learned, what we’d have done differently, etc. as investors.

My answer had to do with not “over-maximizing” a single variable when it came to stock selection. There are, of course, other things I may have learned or might wish I’d done differently. Maybe this will become a series of articles where I talk through a few of those. But, on that night, I had only the one answer: not being too focused on “over-maximizing” a single variable. So, that’s the topic we’ll tackle today.

First, what do I mean by “over-maximizing”? A stock may be clearly expensive at 24 times EBITDA, ambiguously priced at 12 times EBITDA, and clearly cheap at 6 times EBITDA. Does that mean you should like it even more at 3 times EBITDA than at 6 times EBITDA? If it’s the same business at the lower price – yes, of course. Other things equal, cheaper is better. But, other things are rarely equal. And the argument that the same business priced at 3 times EBITDA instead of 6 times EBITDA is a better buy doesn’t translate into an argument for starting with the stocks priced at 3 times EBITDA instead of looking at those priced at 6 times EBITDA. If the business is good enough, the management honest and hardworking enough, your knowledge of the industry deep enough, etc. and it’s priced at 6 times EBITDA – that’s probably enough. Often, you may be compromising more than you think on the other softer variables to maximize the hard variable of price.

This doesn’t mean I’d totally avoid stocks whose primary attraction is their price. Of stocks I talk about regularly on the podcast, two stand out as being “maximally” cheap (or, at least they were when I bought them): Vertu Motors (VTU in London), and NACCO (NC). These stocks were not just “cheap enough” on traditional value measures of price-to-book and EV/EBITDA. They were extremely cheap. Basically, they could have doubled in price and still been considered quite a bit cheaper than the average stock. Of course, that’s because they were in industries (U.K. car dealers and U.S. coal miners) where multiples were much lower than other industries.

So, my point isn’t to avoid the very low price-to-book and very low EV//EBITDA stocks. That’d be silly. There’s no reason to eliminate super cheap stocks for being too cheap and preferring only the somewhat cheap stocks. But, there is a good reason to focus on analyzing the other aspects of these companies and their industries. In the two cases I mentioned, the really big other variables to consider would be: management (especially management’s attitude toward capital allocation), likely future returns on equity (especially considering leverage used was likely to be super low at these two firms), and durability.

I think the most analytical effort would’ve had to have been put into management and likely future returns on equity rather than durability. That’s simply because durability is a very speculative thing to worry about (future predictions for electric vehicles, sources of electricity generation, etc. don’t have a good record for accuracy) and because the durability concerns were very obvious. What I mean by that is that most of what gets written about these kinds of stocks is the idea they have poor future prospects the way society is headed. So, it is easy to find negative viewpoints, people shorting stocks in these groups, etc. focused on the idea of durability. Therefore, it’s not a variable you have to work very hard to focus on when doing your research. In fact, it’s almost unavoidable that you will spend time thinking about durability just because that’s what everyone else is talking about with stocks like these. What a lot of people might overlook is future capital allocations plans, differences in management quality, and the always critical question of likely future returns on equity.

Obviously, likely future returns on equity (and the durability of those returns) should be the top area of investigation when looking at low price-to-book stocks. A stock is very cheap if it has a price-to-book less than one and a decent shot at earning ten percent plus returns on equity as far as the eye can see. It’s not particularly cheap if it’s going to average more like a five percent return on equity.

The topic of ROE is a nice transition to the opposite side of the value/quality spectrum which illustrates this same idea of the pointlessness of truly “over-maximizing” a single variable. A lot of the more Buffett/Munger type value investors of modern times who would say the Graham approach is outdated can easily see the problem with paying 0.5 times book value for a really bad business instead of 2 times book value for a really good business. So, they get the pitfalls of “over-maximizing” a single variable on the price side – but, they may miss the same point when it comes to the quality side.

A lot of emails I get from people stress the really high returns on equity the business they are writing to me about has. They’ll say they want to buy this stock because it has 50% returns on equity (or whatever it might be), but the P/E is 35 (or whatever it might be), so they are really torn. It’s a great business. But, maybe it’s too high a price to pay for a great business. But, Buffett says to pay up for the really great ones. So…that’s where they want my advice. Now, I like high returns on equity. And, on the podcast, I’ve used OTCMarkets (OTCM) as the poster child for high returns on equity many times. Using as a guide (as we often do on the podcast), OTCM has had a return on equity greater than 50% in each of the last 7 years. That’s great. But, it’s also overkill. There just isn’t much reason to pay 2 times more for each dollar of earnings for a business that has twice the return on equity of another business if both businesses have suitably high returns on equity. A good cut-off for “suitably high” is something like 20-30%. Why 20-30%? Well, it’s very hard to compound value at a business by more than 20% a year for the truly long-term. Remember, since Buffett took over Berkshire – the stock has only done about 20% a year. And, even if we were to take Buffett’s peak CAGR period – we still aren’t over 30%. This works for most amazing public companies you can think of. Take their IPO price. Then, carry it through to today. The founder’s fortune has probably not compounded at better than 30% a year unless they’ve been given plenty of options along the way.

So, if Company A has earned 20-30% returns on equity over the last 10-15 years and Company B has earned 40-60% returns on equity over the last 10-15 years – I’d seriously consider the company that is cheaper, seems more durable, has the management you like better, is in the industry you understand better, etc. I wouldn’t obsess about the one with the higher returns on equity.

There’s an obvious exception to this rule. If the company with the higher return on equity can actually re-invest its entire earnings and translate that into higher growth – then, that’s the better stock. In the long-run, this is almost never the case. For example, OTCMarkets has only grown sales by 13% a year and earnings by 19% a year over the last 10 years. Most of the 50% return on equity gets paid out to you in dividends rather than plowed back into growing the franchise. The dividends are appreciated. But, if you pay too high a multiple for the stock – those dividends don’t add up to a heck of a lot.

If we compare Copart (CPRT) to OTCMarkets – we see Copart has averaged more like a 20-30% ROE in the years where OTCMarkets earned a 50-100% ROE. Copart and OTCMarkets trade at fairly similar EV/EBITDA ratios. I don’t think that’s crazy. OTCMarkets certainly doesn’t deserve a much higher ratio just because the ROE is higher. It’s very unlikely that OTCMarkets will have a much higher growth rate than Copart over the next 10-15 years. So, why pay a much higher multiple just because the ROE is higher?

Berkshire is another good example of this. At times, Berkshire has had a few phenomenally high ROE businesses inside it (See’s Candies, Scott Fetzer, etc.). But, it wouldn’t make sense for Buffett to pay much higher multiples for those businesses than he would’ve for something like GEICO. GEICO always had zero chance of matching the returns on equity of those companies. But, GEICO could retain a lot of earnings for a very long time while still earning above average returns on equity.

It’s really the long period of using lots of earnings at some positive spread over your opportunity cost as an investor that makes a high ROE stock the right buy for you. If a company can re-invest a lot of earnings for a long time at a 20% ROE and you only have 10% type return ideas on your stock watchlist – then, that stock can grow your future wealth to be bigger than it otherwise would’ve been. The stock doesn’t need to have a 40% ROE to do that. Again, the higher ROE is mathematically a plus. When it comes to returns on equity, the higher the better – all other things equal. Just as when it comes to price, the lower the better – all other things equal.

That brings me to the final variable where we can argue about “over-maximization”.

This variable happens to be the one I’m least qualified to talk about: growth.

In theory, the maximization of growth – when coupled with self-funding (that is, an ROE greater than the growth rate) is where my argument falls apart. There is no doubt that the power of sustained compounding will lead to much, much greater net present value calculations for stocks that grow earnings per share even just moderately faster than other similarly priced stocks. In fact, the influence of growth mathematically seems so strong that we should just prefer maximizing this variable to the point of ignoring almost all other variables.

Except, of course, it hasn’t worked that way in practice.

Value stocks have actually outperformed growth stocks over time.

And, while a higher starting price is one part of that – the unexpectedly poor future earnings growth of growth stocks versus the unexpectedly good future earnings growth of value stocks is actually a really big driver of that underperformance too. I mean, if growth stocks were just pricier than value stocks to start with but always actually delivered the way analysts expected them to – growth stocks would earn their keep. They don’t. Not, as a group. In fact, very high growth rates for industries and companies in one measurement period (3 years, 5 years, etc.) are often followed by really weak periods right after.

Having said that, I think the issue of selecting stocks on the basis of past growth isn’t as bad as statisticians, quants, etc. would say. As a group, stocks with really strong past growth and really rosy future earnings projections don’t actually end up growing all that fast. But, is it really true that we couldn’t foresee strong growth at a stock like Copart? About a decade ago, I picked that stock as part of a group of stocks which – if price wasn’t a concern – ought to be bought and held. Is it just hindsight bias that Copart turned out that way?

There are a few points to consider here about why it might’ve been easier to foresee strong earnings per share growth at Copart and why it wouldn’t be so easy to foresee such growth at companies growing even faster than Copart had been. This is the “over-maximizing” part of the stock selection process. I’m not saying don’t look for growth stocks. I’m not saying avoid fast growers. I’m just saying something that has grown super fast isn’t necessarily the stock to focus on when there’s another stock that has still grown very fast and yet scores better on a lot of other variables.

So, what could’ve helped predict Copart’s strong growth these last ten years?

One, Copart was a wide-moat business. This wasn’t just me saying that. It was well-recognized at the time. I’m sure Morningstar scored it as such, for example. It was a leader in an industry and the industry was somewhere between an oligopoly and a duopoly even at that point. It was a mature enough business. And the industry was already settled enough (even ten years ago we were very far into the internet phase of this business). Those are the “soft” signs.

Now, on to the harder signs: the actual growth numbers. What we need to predict with a growth stock is EPS. Earnings per share growth depends on the number of shares outstanding (will it shrink or grow?) and net income. Net income depends on the relationship of profits to sales (will margins expand or contract?). And then sales depend on prices and units (will price per unit, revenue per customer, etc. shrink or grow?). Some of these things would be hard to predict for Copart. Others wouldn’t. Shares were going to fall (looking at the management, the compensation, and capital allocation options told you that). Margins were a more complicated analysis. Increased profitability per customer seemed likely. However, actual profits per car sold and stuff like that would be trickier to estimate. Still, the business does have real fixed costs. So, it did seem likely operating profits could grow faster than sales due to economies of scale. At this point, we have EPS growing faster than earnings (because shares are falling) and profits growing faster than sales. Now, there’s a couple big things I couldn’t predict: corporate tax rates fall and inflation (in cars and other stuff) rises. Without those things happening, growth would’ve been lower. So, no matter what – there was no way to foresee the 20% or so EPS growth that happened at Copart over the last decade. However, taking just the top line – that’d be 10% growth.

Could we have predicted better than 10% EPS growth?

I think so.

Putting aside the corporate tax cut, results right up till the year before the pandemic (2019) look like things you could’ve guessed even ten years before. You might not have been able to quantify exactly what EPS growth would be. But, you probably could’ve predicted that Copart’s chances of above average EPS growth being sustained for a long time were much, much higher than that of the average stock.

Note, however, the really important part here. Copart actually had reasonable sales growth both back at the beginning of this period and even through to today. We’re talking 10% growth in an average year and 20% growth in an amazing year. The founder even talked about this in the book “Junk to Gold”.  A slower, less lumpy (but still faster than average) growth rate was something Wall Street favored.

This brings me back to the high ROE problem from a different angle. Yes, there are companies that have had and may continue to have very, very high EPS growth rates. But, these usually aren’t the companies where it is easiest to predict continued high EPS growth for the next 10-15 years (which is really what matters to someone buying the stock today).

On top of this, the stocks with the absolute highest sales growth rates usually aren’t (yet) the stocks with particularly high returns on equity. Tesla (TSLA) averaged sales growth of 50-100% in a lot of years before becoming profitable (so, ROE was negative). It’s actually grown a bit slower since first turning a profit. Presumably, Tesla’s highest ROE years will be in future periods where sales growth (while high) will be nowhere near as fast it was back when the company was unprofitable. There are a lot of companies like this.

But, not every company is like this. In fact, there are a few famous counterexamples. For our purposes, one giant one will suffice.

Something like Meta (FB) is a clear exception to my “too high an ROE, too high a growth rate, etc. is often overkill – so, don’t pay an absurd looking multiple for it” argument.

Facebook (now Meta) has had high returns on equity, high top line growth, and been classified as a “growth stock” for its entire history as a public company. It has deserved that designation and – with hindsight – it’s a high return on equity, high top line growth, strong future prospects stock that delivered exactly as the past record would suggest it should have. With a stock like this, “overdoing” the focus on growth and ROE to the exception of worry about almost any other variable would actually have worked. So, buying Meta (then Facebook) because it had 2x the growth rate of an equally good looking stock – wouldn’t have been a mistake. This is the pure Phil Fisher type growth approach.

If there are rare growth maximization examples like Meta – can’t there be rare examples for things like a seeming “over-maximization” of ROE too?


It’s probably even rarer…

But, yes, a total focus on ROE could work too.

And it could work even at a business (and in an industry) where future growth prospects appeared to be nil.

Now, I’d strongly caution against betting on any stock where the only attraction was a high ROE combined with no growth prospects.

However, I will admit this is actually an investing playbook that’s been successfully run many times.

In fact, it’s done all the time. Just not in public markets. Maximizing ROE at the expense of everything else can work if the capital allocation and management’s approach to the business are designed around maximizing economic profit for the shareholders at the expense of everything else. Companies like Teledyne under Singleton and Berkshire under Buffett come close to this ideal. And some private equity controlled LBOs actually achieve this ideal. In theory, the same things could be done without taking the company private. Debt could be increased, shares bought back, special dividends paid out, entire divisions spun-out, sold off, etc.

These seem like arguments for the idea of just buying a high ROE stock without worrying about other variables. But, I think the truth is that these are actually stocks that can only be identified if you look for a combination of really unusual capital allocation and really strong underlying ROE possibilities. Just knowing the ROE was high won’t identify these companies. In fact, some wouldn’t even show up (till the shift in capital allocation was well underway) on high ROE screens.

Many of these compounders – Berkshire, for example – were more about the capital allocation being so extreme it squeezed ROE from something that wouldn’t have shown up as having any ROE at a site like

A highest ROE screen alone is not going to find you the best stocks out there. But, a high enough ROE combined with the right management implementing the right capital allocation is the “screen” that will get you a lot further.

So, my advice is to worry less about finding a stock that perfectly maximizes one variable (the fastest growing, the highest quality, the cheapest) and instead look for stocks with the right combination of a few variables working together in a self-reinforcing sort of way to drive especially strong returns.

A really cheap stock that looks like it is going to keep buying back its stock as long as it stays this cheap.

A high return on equity stock with a long runway to reinvest that free cash flow for years to come.

Sure. It’s a familiar list. We talk about these kinds of stocks all the time. But, we’re often tempted to pass over these strong combinations when we find something just a little bit cheaper, just a little bit faster growing, etc.

Often, I think we do this because we put more trust in a single series of numbers than in our gut interpretation of a constellation of figures working together over time. However, it’s going to be that way that constellation works together that will drive the really good compound returns in whichever stocks we pick. So, why not start by looking for the stocks where a couple things are going right in ways that fit together instead of looking at only the very top stocks on whatever overly simplistic screens we run.