Before you turn off this podcast for fear of further math, let me assure you this episode is not about math. It’s about consistency. The math just helps us put that consistency into numbers. My inspiration for using the variation coefficient comes from 1934, when a guy named Ben Graham wrote a textbook called “Security Analysis”.

Ben Graham was Warren Buffett’s teacher. Graham taught Buffett using the 1940 edition of Security Analysis. And on page 506 of that book you’ll find a chapter called “Significance of the Earnings Record”.

Now 70 years later, in 2010, I’m focusing on the free cash flow record instead of the earnings record. I think Graham would approve. Back in 1940, he didn’t have the same financial statements we do today. Public companies weren’t forced to report their cash flows until 1987.

So today’s investors get to see a lot of numbers Graham never did. And that means we should tweak Graham’s approach in a couple places. One tweak I like is focusing on free cash flow instead of earnings.

So I’m rewriting Graham a bit here. Instead of talking about the significance of the earnings record, we’re talking about the significance of the free cash flow record.

Ben Graham thought about stocks in terms of earning power. He defined earning power as a combination of “a statement of actual earnings, shown over a period of years, with a reasonable expectation that these will be approximated in the future, unless extraordinary conditions supervene.”

The important idea here is that Graham wanted a demonstrated record of earnings. He didn’t use estimates. He didn’t predict the future. He just studied the past.

Graham used averages taken over many years. He hated the P/E ratio. Whenever one of his analysts would come up to him and say a stock’s price-to-earnings ratio was 7, he’s asked if that was on last year’s earnings or on the last 10 years’ earnings. He preferred to use the last 10 years.

Here’s what Graham said about average earnings: “The record must cover a number of years, first because a continued or repeated performance is always more impressive than a single occurrence and secondly because the average of a fairly long period will tend to absorb and equalize the distorting influences of the business cycle.”

What Graham is really talking about here is statistics. Ben Graham was a math whiz. You have a computer. So you don’t have to be a math whiz. But you do have to understand the principles behind probability and statistics the way Graham did. You don’t have to do the calculations yourself. But you do have to make the interpretations.

So, yes, this episode is about statistics. It is about math. But it’s not about crunching numbers. It’s about the big ideas.

The question here is how to find a stock’s earning power. Its true earnings. Its natural tendency.

To show you how simple this idea is, I’m going to talk baseball for a minute. My apologies to any foreign listeners. But baseball is the one sport where even the most mathphobic fan sees stats every night and actually knows how to use them.

Even a lousy batter will have a game where he goes 3 for 5. But not even the dumbest of fans will think that means he’s a .600 hitter. One: because there are no .600 hitters. And two: because it’s only one night.

Every baseball fan gets that. But not every investor does. A lot of investors think that last year’s earnings are a better gauge of a company’s future performance than the last 10 years.

Why? Because investors focus on trends. But, again, let’s go back to baseball. We all know hitters can have slumps. There are good players who go 3 for 25. But fans don’t say “Oh well, I guess he’s a .120 hitter now.”

No. They say he’s in a slump. They say he’ll snap out of it someday. And he always does.

Companies, and even economies, have slumps too. Trends matter. Sometimes trends even tell you things. But eventually trends reverse. And most numbers revert to the mean. Hitters have slumps where they go 3 for 25 and hot streaks where they go 11 for 25. We don’t say that makes them a .120 hitter one week and a .440 hitter the next week.

From week to week we’re more likely to say a hitter’s luck has changed than his skill has changed.

But do we say that about companies? Do we allow for a little meaningless variation in their earnings? A little wiggle in their numbers? Or do we try to explain every tiny change in their earnings with some growing trend?

The truth is that there’s no perfect number. Companies do get better and worse over time. So do baseball players. But companies, like baseball players, also bounce around a bit. They aren’t perfectly consistent. And part of the change in earnings from year to year, is what we would call “noise” in any other set of numbers.

Graham wrote about this problem. He said: “A distinction must be drawn, however, between an average that is the mere arithmetical resultant of an assortment of disconnected figures and an average that is ‘normal’ or ‘modal’, in the sense that the annual results show a definite tendency to approximate the average.”

In other fields, it’s easier to separate the signal from the noise, because you have a clear idea of what the signal is.

In investing, we’re not always clear on what the signal is. What exactly are we trying to measure?

Think about this for a second. When you look at a stock’s earnings, what do you want to know? Do you want to know what it earned last year? Do you want to know what it will earn next year? Or do you want to know roughly what it will earn for the next 10 or 15 years?

Personally, I only really care about the last question. I want to know what the company will earn, on average, in the long-run.

The truth is that stock prices don’t follow earnings in the short-run. A 20% pop in earnings does not mean a 20% pop in stock prices. Investors try to anticipate changes in earnings. They worry about declining industries like newspapers and punish those stocks with low P/E ratios. They get excited about growing industries and give those stocks high P/E ratios.

Even if we put aside the evidence of what actually happens when earnings go up and down, reason still tells us the long-run is what matters. A stock’s price is normally many, many times higher than its earnings. Why? Because investors are betting on future earnings.

Even if you’re only planning to own a stock for a year or two, you’re planning to sell it to another guy who’s planning to sell it to another guy who’s planning to sell it to yet another guy. Each of you is betting on getting a good price. That means each of you is betting on the future looking bright when you sell.

And that means it’s not just next year’s earnings that matter. A stock’s price is based on investors’ view of its future earnings as far as the eye can see.

That means we, like Graham, want to focus on earning power instead of earnings. We don’t want to limit ourselves to talking about next year. And we aren’t so arrogant that we think we can predict earnings ten years out. But maybe, with a little help from statistics, we can find that “definite tendency” Graham talked about.

On page 507 of the 1940 edition of Security Analysis, Graham presents a chart of two companies and their earnings for each year from 1923 through 1932.

The two companies are S.H. Kress and Hudson Motors. Kress had average earnings of $4.36 a share. Hudson had average earnings of $4.75. But Graham didn’t use the chart to talk about averages. He used it to talk about variation.

Just listen to these numbers. I know it sounds weird having someone read numbers to you. But I’m trying to prove a point. You don’t need to be a math whiz to get statistics. It’s not about number crunching. It’s about common sense.

Here are the earnings per share for Kress: 3.39, 3.06, 4.12, 4.65, 5.26, 5.76, 5.92, 4.49, 4.19, and 2.80.

Now I know you’re not a computer. And I’m going to guess you’re not a math whiz like Ben Graham. But I’m also going to guess your mind did 3 things automatically. It looked for a trend, it looked for a tendency, and it looked for wiggle.

In other words: you tried to see the growth rate, mean, and variation in that set of numbers. Since you aren’t a computer, you probably couldn’t do it in your head. You can’t put what you learned in numbers. But that doesn’t mean you didn’t learn anything.

Listen to this second set of numbers, for Hudson Motors, and I think you’ll find you learned a lot more than you can put into numbers.

Hudson Motors earnings per share were: 5.56, 5.06, 13.39, 3.37, 9.04, 8.43, 7.26, 0.20, (1.25), and (3.54).

Which companies earnings varied more? Was it the first set or the second set?

Obviously it was the second set. That means you do know something about statistics. Even without putting anything into numbers, you could tell me that Hudson Motors’s earnings bounced around a lot more than Kress’s earnings.

I’m sure you also noticed that the trend for Hudson was worse. Kress’s earnings only dropped around half from peak to valley. Hudson’s earnings dropped 1 and a half times in a straight slide.

You can eyeball all of this. We’re talking statistics but we aren’t talking numbers. This isn’t math. But it is investing.

So why do I want to bring in stuff like standard deviation and the coefficient of variation?

It all goes back to what I’ve said in other episodes. Numbers don’t lie. Humans lie. And avoiding numbers won’t help us avoid lies. We lie to ourselves all the time. And we’ll do it with or without numbers.

It isn’t really about numbers making us smarter. As we saw with the chart in Graham’s book, the human eye can see obvious statistical relationships. You don’t need a computer. And you don’t even need math.

But the human eye won’t put what it sees into writing. Instead it’ll try to make sense of what it sees. And if you’ve told yourself a story in words that doesn’t jive with the numbers you’re seeing, you’ll probably keep the story and forget the numbers.

But when we actually write the statistical relationship down using a number, we force ourselves to face the facts. We make it harder to stay in denial. Oh, you can do it. But it’s harder. When you’re confronted with numbers you don’t like, you can stay in denial, but it’s going to hurt a lot more. Every time you look at that number your brain is going to feel like it’s being pulled in two directions at once. It’s going to hurt. And that pain might just stop you from doing something stupid.

Warren Buffett likes to say he doesn’t have more good ideas than other investors. He just has fewer bad ideas.

He’s not kidding. Most investors don’t suffer from a lack of good ideas. The truth is they have tons of ideas; they just can’t sort the good from the bad.

The key to Buffett’s success, and in a way even Ben Graham’s success, comes down to one word: selectivity.

Ben Graham diversified. Warren Buffett doesn’t. But by today’s standards, Graham put a lot of money into his top few picks. And he didn’t turn his portfolio over much. Even more importantly, he looked under every rock. Graham wasn’t afraid to buy tiny companies. But he’d buy big companies too. He was willing to go into any industry. And Graham never did any top down stock picking. He didn’t make big picture calls. He just focused on the stock in front of him.

That’s what you should do. And that’s why I’m talking about free cash flow margin variation. I want you to drill down and really get to know the stock you’re looking at.

But more than anything, I want to keep you from making big mistakes.

I said the free cash flow margin variation was one of 4 numbers you should check for each stock the same way doctors check a patient’s vital signs.

The other vital signs I want you to check are: the Z-Score, the F-Score, and the real free cash flow yield.

I haven’t talked about the last one yet. But all 4 are meant to protect you from making a big mistake. The Z-Score warns you about bankruptcy. The F-score warns you about a business taking a turn for the worse. And free cash flow margin variation warns you about the future being different from the past.

It’s not a perfect number. It won’t warn you about technology. It won’t tell you newspapers are an endangered species. It can only tell you about things that have already made their mark on the past record. It’s backward looking.

But it’s one of my all time favorite numbers. It’s been a huge help in steering me away from some big mistakes.

The free cash flow margin variation coefficient can also help you see old stocks in a new way. I’ll give you a current example.

But first, I should tell you how I calculate the number. If you use Microsoft Excel, it’s really easy. Just type in the past 10 years of free cash flow and sales. You should get these numbers from EDGAR. If you’re lazy, you can use GuruFocus. But it’s always better to go to the original source.

Once you have free cash flow and sales for each of the last 10 years, add a column where you divide free cash flow by sales. That’s the free cash flow margin. Now use the “AVERAGE” function on those 10 free cash flow margins. Then use the “STDEVP” function on those same 10 free cash flow margins. And, finally, divide the standard deviation by the average.

That’s how I like to measure free cash flow variation. There are other variation measures you can use. But this one works fine. Since I’m always encouraging you to focus on stocks with 10 straight years of positive free cash flow, you’ll normally be working with all positive numbers. This measure works better with positive numbers. You can also run into trouble if the average free cash flow margin is close to zero. But, let’s face it, if a stock’s average free cash flow margin is close to zero, why are you thinking about buying it?

Okay. Enough math. Let’s talk about a real world example.

Last week, I was looking at two different wide-moat companies. The two companies were

**Pepsi (NYSE:PEP)**and

**United Technologies (NYSE:UTX)**.

These are two of the world’s bluest blue chips. So you’d expect free cash flow margin variation to be low at both companies.

It is. But I think you’d also expect Pepsi, which makes snack foods, to have a steadier free cash flow margin than United Technologies, which makes things like elevators, helicopters, and jet engines.

That’s what I expected. And I was wrong. Pepsi has a variation coefficient of 0.09. That’s unbelievably low. Microsoft’s is probably around 0.25. Anyway, Pepsi has a variation coefficient of 0.09, but United Technologies is lower at 0.07.

I can’t remember seeing a stock with steadier free cash flow margins than UTX. I’m sure there’s a stock out there somewhere. But I can’t remember coming across one.

To give you some idea of what a variation coefficient of 0.07 means, let’s start with United Technologies’s 10-year average free cash flow margin. It’s 7.76%.

If UTX’s free cash flow margin followed a bell curve, 7.76% would be in the middle of the curve the way 100 is in the middle of an IQ test curve.

On IQ tests, 2 out of 3 people score between 85 and 115. 19 out of 20 score between 70 and 130. And 997 out of 1,000 score between 55 and 145. That’s because an IQ score has a mean of 100 and a standard deviation of 15.

Do the same rules apply to a company’s free cash flow margin?

No. You can’t assume that. But you can assume that a steady average is more likely to hold in the future than a shaky average.

And that’s all we’re talking about here. The number itself doesn’t matter.

I never want you to think about United Technologies’ 10-year free cash flow margin variation as being exactly 0.07. It is. But that’s not what matters. What matters is that it’s less than Pepsi’s.

In other words, don’t use the coefficient as a number. Use it to rank stocks. Use it to compare stocks. Use it to say UTX has steadier margins than Pepsi. Use it to say Pepsi has steadier margins than Microsoft.

Don’t get caught up in the math. Don’t let the numbers take over your analysis. Numbers, like words, are tools. You can use numbers and you can misuse numbers.

I love the 10-year free cash flow margin coefficient of variation. But I never use it to predict how likely a specific free cash flow margin is in the future. I never use it to draw a bell curve.

I use it to force myself to pay attention to how steady or wobbly a stock’s free cash flow margin is. And I use it to challenge my assumptions. I assumed Pepsi was a steadier business than United Technologies.

After I saw United Technologies’ super steady free cash flow margins, I started looking more closely at the stock. That’s when I realized it wasn’t just some weird number I missed. The more I looked at UTX, the more I saw how biased my first impression was. Maybe elevators and engines can be steady money makers. Maybe they can be as steady as snacks and soda.

Maybe. I don’t know. But I do know UTX is a much steadier business than I thought. I didn’t take that steadiness seriously until I was forced to see it boiled down into one number. I thought I knew the business when all I really knew was my own prejudice against products like elevators, engines, and helicopters.

That’s what the 10-year free cash flow coefficient of variation is for. It’s meant to keep you from making big mistakes. And it’s meant to open your eyes a little wider with just one number.

Well that’s all for today’s show. If you have an investing question you want answered call

**1-800-604-1929**and leave me a voicemail. That’s

**1-800-604-1929**.

Thanks for listening.

### About the author:

*Geoff Gannon
*

**Geoff Gannon**

Salsero- 5 years ago