Geoff here.

Someone who reads the blog sent me an email about how to value a stock. I answered that email with the idea of peer comparisons.

Peer comparisons are really a short-term valuation approach. If you buy **Interpublic (IPG)** at 0.7 times EV/Sales because you expect it should one day trade in line with the other 3 big advertising companies – which all trade at 1.1 to 1.2 times sales – then you are making a short-term bet. You hope the business will do just dandy. But you are betting on getting a 50% to 60% pop at some point from multiple expansion.

In a situation like that it’s not just the price you paid and the performance of the business that matters. There’s also another issue – how long will you have to hold the stock? Is that 50% to 60% multiple expansion pop going to be spread out over 1 year, 3 years, 5 years, or 10 years?

It matters. A lot.

But not if you plan – Warren Buffett style – to actually buy and hold a stock forever. That’s what I’ll be talking about today.

Most people will tell you to do a DCF. I am against that. I think that in an analysis of a stock's truly long-term return potential, there are really just 3 critical variables:

- Earnings yield (Earnings/Price)
- Sales Growth
- Return on Investment

If you want to think of the harmonic mean of these numbers as being the best estimate of your long-term returns in the stock - and by long-term I mean buy and hold forever - I think that makes sense.

So, for example, these three numbers at **Coca-Cola (KO)** are:

- Earnings Yield = 5%
- Sales Growth = 9%
- Return on Investment = 19%

The harmonic mean of 5%, 9%, and 19% is 8%. That's not a bad guess of what Coca-Cola stock can return to you if you buy it and hold it from now until the end of time. However, the key assumptions are the earnings yield (not controversial in Coke's case), sales growth (which could be lower in the future than in the past), and ROI (which I just took the most recent year of data).

**Assume Your Return Will be Close to the Lowest of the 3 Rates – Not the Highest**

Why use the harmonic mean?

When I’ve talked about forever return potential in the past, I’ve used the earnings yield times the sales growth times the return on investment and the cube root of that number to try to make the reason for this kind of calculation clear.

This is a snowball measure. It is about you paying $100 for $5 of earnings today, having those earnings – to the extent possible – put back into the business, having those earnings producing earnings upon earnings, and so on…

I think I did a bad job explaining this. I made it sound more complicated than it really is.

So let's talk about a mean. Everyone understands the idea of a mean. The idea of an average.

The reason for using the harmonic mean is that we’re really trying to look at the flow of the earnings you are buying today through the future – but with a key concern. The key concern is that there’s usually a bottleneck.

I think investors tend to overestimate the long-term return potential of a stock with especially good earnings yield, growth potential, or return on investment – but mediocre scores on the other numbers.

And I think investors tend to underestimate the long-term return potential of a stock with a moderately good earnings yield, moderately good sales growth potential, and a moderately good return on investment.

A stock with a P/E of 10, ROI of 10, and the possibility to grow sales close to 10% a year – is potentially a good investment.

In fact, as we’ll see below it can be a better investment over the long-term than a company with a P/E of 20 that has a great return on investment. The same would be true when compared to a stock with a P/E of 5 (so an earnings yield of 20) but a return on investment of just 9% and sales growth potential of just 5%.

A stock like that is fine if you know how the bottleneck will be solved. Big dividend payments combined with a consistent 5% sales growth rate may do the trick and turn that stock with a P/E of 10 into a terrific bargain. But you want to know how the bottleneck – in this case, the cash build – will be solved.

Obviously, you want to take this return number and compare it to other investment opportunities. That could mean other stocks. It could mean bonds, etc. I think Coke is clearly more attractive than bonds. Even 30-year Baa bonds only yield 5% right now. While we are saying Coke is priced to return closer to 8% a year. And how much more dangerous is Coke than a Baa bond? There are very few stocks as safe as Coca-Cola.

But is Coca-Cola the most attractive stock out there. Let's take a look at **Carbo Ceramics (CRR)**:

- Earnings Yield = 8%
- Sales Growth = 18%
- Return on Investment = 21%

In this case, the harmonic mean is 13%. Now, you may totally disbelieve the assumptions made here. I think I believe the earnings yield assumption in the sense it isn't far from average earnings unless I am totally wrong about the macro situation in the industry (hydraulic fracturing) which I certainly could be. I don't believe for a second that Carbo will achieve 18% sales growth over the long-term. And I suspect the ROI will come down. But these are guesses. I know Carbo will face more competition. The company's growth has attracted competitors.

**Use the Average Rate as a Best Case Estimate of Long-Term Returns**

So we can think about some of our current investment choices like this:

Carbo Ceramics: 13%

Coke: 8%

Baa: 5%

The question then is whether the advantage over Baa bonds is enough to interest you in Coke, whether Carbo's expected return advantage over Coke is enough to interest you in Carbo, etc.

Remember, how comfortable you feel with each asset depends on how comfortable you are with the assumptions that went into the return estimate. So, for Baa bonds it was that interest payments will be made, and that you will get back your principal.

For Carbo and Coke it was assumptions about earnings yield (that today’s earnings are “normal”), sales growth (that the last 10 years is a guide to the next 10 years), and ROI (that today’s return on investment is a guide to tomorrow’s return on investment).

These are huge assumptions. And they are hard to quantify. That’s why I encourage people to focus on marrying a return expectation they like with a level of comfort about the business and its competitive position that is very, very high.

Your purchase decision then depends on which assets you like best that offer the best long-term returns. So, in this sense, it isn't even necessary to value stocks. You just need to look at the kind of long-term returns they can achieve and your comfort level holding the stock forever.

So, the assumptions checklist for a buy and hold stock investment looks like:

- Earnings yield
- Sales growth
- Return on investment
- Your comfort level with holding the stock forever

** **

**Be Honest about Your Comfort Level – Focus on Risk More Than Return**

Where your comfort level is matched, you should pick the highest expected annual return. So, if you really would be just as comfortable owning Coke as a Baa bond - you should definitely not buy a Baa bond. You should buy Coke. The math favors Coke by a lot if you plan to hold forever. If you plan to hold for a couple years, it's much more of an open question.

Obviously, no one will feel as safe holding Carbo as holding Coke. So, then it becomes a question of whether past results alone predicting 13% returns for Carbo can more than offset your lack of comfort with that stock relative to the expectation of 8% in Coke using the same looking backward to see forward approach.

I try to keep a list of the stocks I'm most comfortable holding with a long-term return expectation purely based on earnings yield, sales growth, and ROI assumptions next to each name. That way, I can think of **Dun & Bradstreet (DB)** as a 9% bond, Carbo as a 13% bond, etc. And ask myself how safe I feel with each.

**Focus on the Bottleneck**

I also want to stress the idea of a return bottleneck here. This explains why – in some situations – I care so much about past behavior in terms of buybacks, etc.

If you have no idea what DNB is going to do with their capital – the long-term return expectation might be something like 7%. If you know they will buy back stock – the return expectation is at least several percentage points higher.

That’s because growth is the weak link for DNB. They can’t put capital back to work in the business. So you need to know where their earnings will flow. Will it be buybacks, dividends, or acquisitions?

That’s critical. For some companies the worrisome number is ROI. You need to know the company’s ROI is cyclically depressed right now – or it doesn’t make any sense to buy the stock.

For others, it might even be earnings. In that case you need to know “normal” earnings are much higher than today’s earnings.

Regardless, it tends to be the low number among these 3 that causes you the most problems as a buy and hold investor.