OK jbe, I'll repost here what I submitted to the Unitrode thread. For anyone else reading this, upon seeing that I am also a free cash flow freak, jbe asked me to critique his method of using price-to-free-cash-flow ratios as a valuation tool. Here is my long-winded answer:
**********************************************
jbe, nice to hear from you!
IMHO there are two major steps to using free cash flow (FCF) as a valuation tool:
1. Recognizing that FCF, not net earnings, represents the true economic return from a business.
2. Conservatively estimating the stream of future free cash flows you expect the business to generate, and bringing those totals back into today's dollars to see whether what you're paying is less than what that future stream of cash is worth to you today.
(Anyone who thinks that the above is a bunch of nonsense will probably not want to read on...)
Sounds like you've got the first part right on the money (so to speak ), but if you ask me (I guess you did) I think you're missing the payoff. Please accept my apologies in advance if I sound like a know-it-all. The following is all merely one person's opinion!
"In any event, a question for you: does using the free cashflow per share and price/cashflow ratio give me a reasonably accurate picture of the free cashflow picture, or not?"
I think you have the right bullets, but the wrong gun! I'll explain why I think that.
Ok, look at free cash flow per share first. Assuming that you have enough information to properly calculate it (as you alluded to in your post), this tells you the fraction of extra cash "coming to you" if you own one share.Maybe the company could pay this out as a dividend, but more likely it will reinvest it back into the business. And if this is a company generating a high return on equity, I think this is the best use for that cash - it's like a tax-free dividend reinvestment plan. Another option is share buybacks - which is great if the company can't find internal uses for the cash that generate a high return on investment. So let's assume the company does something wise with your chunk of the cash it's throwing off - so it's "coming to you" one way or another.
So what's this year's cash flow worth? I guess pretty close to it's nominal value. But that's obviously not all you're paying for when you buy shares! You're paying for a percentage of all future free cash flows as well.
So how do you add 'em all up and decide how much each share is intrinsically worth? Well, I treat stocks as if they're bonds. I'll get back to that.
The price-to-earnings ratio is pretty much the most common valuation tool out there. Probably because it's the simplest, easiest, and quickest to use. But I think you get what you pay for (effort-wise). And I think you run into the same problem using the price-to-FCF (let's call it P/F). Try to answer this question: "What does a P/F of 13 mean? Is it high or is it low?" I think the answer is: "There's not enough information." People seem to have a vague notion that P/E's are "allowed" to be higher when interest rates are low like they are now. But why? Sez who? How high is Ok? Will they tell us if "they" change the rules? What I think is silly is this "rule of thumb" that says a company is undervalued if its P/E is less than it's percentage growth rate. Again I ask why! Who calculated this ancient wisdom, what assumptions were they making, and do those assumptions apply at all times and for all companies? I have no idea, and no one I've asked has ever given a logical answer to the question.
The P/E is data without context. It's like experimental results where you don't know how the test was done. I believe you just can't draw meaningful conclusions from it.
There is an alternative, which is based on the value of money over time. Everybody knows that in an inflationary economy, money loses it's value over time. So you would not pay one dollar today for a contract which promises to deliver you one dollar ten years from now - you'd lose buying power. Instead, you'd be willing to pay somewhat less than a dollar. In that case, the contract would basically be called a loan, and the difference in absolute dollars would be called the interest paid to you to compensate for devaluation of that money (plus a bit extra so you can profit from your generosity) That's what a bond is. You pay a sum of money now, for a larger sum later, and hope that the difference more than makes up for inflation.
Many people consider the long (30 year) bond a proxy for the time value of money. It's called the "risk free" rate, or the discount rate. If the "risk free" rate of return is 7%, anyone can earn 7% on long term money by buying a government bond. So it's pretty silly to put your money in something that earns less than 7%. For example if someone were to offer to pay you $1628.90 ten years from now in return for a $1000 dollar loan, you'd probably say "Are you nuts? That's 5% interest! I can get 7% in a government bond and they're a lot less risky than you!" It wouldn't be a productive use of your capital. (1000 * 1.05^10 = 1628.90)
So if you look at it backwards, you can "discount" a future lump of cash back to the present using the discount rate. If a relatively risky investment promises a payment of $2000 ten years from now, you'd better not pay any more than $1016.70 for it because you'd do better putting your money in riskless government bonds. (2000 / 1.07^10 = 1016.70)
Now assume you're thinking about buying a company with the following free cash flows per share (which are paid to you as dividends):
Year 1: $1.00
2: $1.00
3: $1.00
4: $1.00
5: $1.00
Then the company goes bankrupt!
Let's assume you're pretty darn confident about those being say the worst case cash flows (barring nuclear war or something). The maximum you should pay for one share is:
1/1.07^1 + 1/1.07^2 + 1/1.07^3 + 1/1.07^4 + 1/1.07^5
=0.93 + 0.87 + 0.82 +0.76 + 0.71 = $4.10
Three things to note. First, notice how the longer you wait for a dollar, the less it's worth to you. Second, if you pay $4.10, you don't get any bonus for taking the risk of buying into this company instead of a good old bond, unless the company does better than your conservative estimates. So you would want to pay considerably less than $4.10 per share. Third, the lower the discount rate, the more you'd be willing to pay.
Now things get more complex as you try to make the situation more realistic. KO won't go bankrupt in year 6. To avoid repetitive math, there are formulas which do all this discounting for you. You enter the company's growth rate, the discount rate, the starting FCF and the length of time to be considered. But you have to be careful - KO will probably not go bankrupt in year six, but neither is it likely to grow at 15% a year until the end of time! Believe me, you can justify some pretty insanely high stock prices by extrapolating out the growth rate unrealistically far!
So this tool has a very rational basis - it allows you to calculate what a share is worth if you want to earn about the same return as you get on a risk free long term bond. By introducing Buffett-like factors of safety, you can calculate what you'd be willing to pay in order to achieve a very high rate of return. And it's all based on real dollars, not obscure rules of thumb. The "fair value" P/F ratio is just one of many ways of expressing the results of this valuation. But alone, it does not explain what discount rate, growth rate and time duration were used to derive it!
But it's also a very dangerous tool if not used carefully. Your results really will vary wildly by small-seeming changes in variables. You can't allow these fancy formulas to give you a false sense of precision! For example, changing the discount rate from say 8% to 7% will make your stock's apparent intrinsic values leap upwards dramatically. So if you significantly misjudge any of your inputs, your output could burn you.
So here's my methodology:
1. Don't ever even bother trying to do this on any company that you haven't become incredibly confident about, based on a great deal of qualitative and quantitative analysis.
2. Try to assume a ridiculously low growth rate. If people say Cisco is going to grow at 30% a year, assume it's probably going to grow something like at least 12% a year, and work with that.
3. Always add several percentage points to the discount rate. If the long bond is at 6%, assume you want to make at least, say, 9% (to compensate you for the risk you're taking by pretending you can predict the future of this bloody company)
4. Try to be conservative with your assumption about how long that average growth rate is likely to last. Something like "Cisco's annual growth rate could realistically average around 12% for 10 years, and then say 5% thereafter." And be especially careful with the thereafter part. Even though distant earnings get less and less significant, infinity is a long time! Maybe try 10 or 20 years instead!
5. Buy the stock at at least a 30% to 50% discount to this conservative fair value.
Obviously, you have to believe that over the long term, the stock will track reasonably closely the growth in intrinsic value of the company's free cash flows. If this is not a good assumption, doing the above is a complete waste of effort. I will point out that Warren Buffett has been reasonably successful thinking this way.
jbe, I hope that the above ranting addressed your question about your technique. I shared my thoughts on the matter anyway. If you are still interested, I could email you the formulas I described, or even my template spreadsheet.
I'll look at your other questions in my next post. I swear it'll be much shorter than this one!
Andrew |
