This thing traded at 15 times in 1988. It trades at the multiple it does b/c of the Buffett premium.
Every company he buys investors drive up the price and keep it there.
Back in 1988, the bearish crew was predicting a great depression. It's been a dry spell for these guys for a number of years, now they have to make up for lost time. look at the record of these losers.
You have to be a rational optimist to make money over the long term.
I also enjoyed T. Vic and R. Hagstrom's books a lot. Both have been a tremendous help to me and were fun to read. My only disagreement is over the way they apply DCF. Of course, if their valuation technique is wrong, it makes it tough for either man to apply it in the real world.
The idea that there is a difference between discounted cash flow models and dividend discount models is wrong. I think this came about partly because of serious errors in the DCF examples in Tim Vic's and Robert Hagstrom's books. Only distributions to investors and the terminal value should be discounted and added to come up with intrinsic value. Reinvested cash flow is just a source of growth. If you treat is as distributed, you have double counted and miscalculated the intrinsic value. You can prove this to yourself using a bond example. If you assume a 5% coupon, and a 5% discount rate, the bond should be worth 100 cents on the dollar. But if you use DCF the way Hagstrom and Vic do, their model will produce an intrinsic value of $152 for a $100 bond -- clearly wrong. These guys would have you believe a 5% bond in a 5% interest rate environment should be priced to yeild 3.3%! Their misuse of DCF may explain why both men fall short of Warren Buffet in portfolio management.
+++John Burr Williams argument that the value of a company is the cash that will be taken out of it over it's lifetime (dividends and breakup value). Now my thinking is that this is ultimately true.+++
it is true. recently i posted here a valuation of MO a.k.a. altria exactly along those lines.
however, i think you'll find that for many situations, such a "rigorous" approach requires so many dubious assumptions that it renders the entire exercise meaningless. for example, i have never heard a convincing valuation of BRK predicated on a dividend discount model.
other discounted valuation models take shortcuts and make simplifying assumptions. for example, valuing "free cash" (however defined), even if not distributed to the owners. buffett thinks this way, cf. "look-through earnings".
again, though, with a discounted dividend model you certainly do not have to use the company's reinvestment rate as your discount rate in the valuation. these two things have no necessary connection.
the discount rate you use is your answer to the question, "what's a buck worth if i get it tomorrow instead of today?" it is a mathematical expression of the time value of money. it has nothing to do with the internal processes used to generate the buck that appears tomorrow.
however, models that value "free cash" never actually received need to incorporate certain assumptions about how that "free cash" is reinvested by the company. these assumptions are often made implicitly, and are sometimes highly inappropriate.
perhaps that's what you're remembering?
but i dunno; never read john burr williams's book. maybe i'm missing the whole point...
Ever since I heard that "discounted cash flow models implicitly assume reinvestment at the discount rate" I have tried to make sense of it. I'm still not 100% convinced that it is true, but to be safe I use that and ramp up to the historical discount rate, and just increase my MOS.
However, there is one way that it kind of fits for me. And that came about after reading John Burr Williams argument that the value of a company is the cash that will be taken out of it over it's lifetime (dividends and breakup value). Now my thinking is that this is ultimately true. Also, I believe it is true that for discounted dividend model, you have to use the existing re-investment rate as your discount rate, because that is the rate at which you will re-invest those dividends.
However, there is a difficulty in determining how much dividends a company can pay out over it's lifetime, especially if a company can reinvest their free cashflow profitably instead of paying out dividends. I believe that if they can invest retained earning profitably (ie Berkshire), then someday down the road company will be payout those successfully retained earnings as dividends. So a discounted cash flow model is acceptable to use in place of a discounted dividend model because it accounts for this.
And taking all of this into account, I think that it is appropriate to use a discounted cash flow model with the existing rate of return (and then ramp up or down) to the historical rate of return in determining the intrinsic value of a company.
All comments on my thinking are greatly welcomed.
+++I have heard that discounted cash flow models implicitly assume reinvestment at the discount rate.+++
can you elaborate?
this sounds wrong to me.
DCF models, generally speaking, don't assume any reinvestment. if they did, it would really wreak havoc on the "present value of an annuity" formulae used by BRK subs...
however, there are numerous ways to construct DCF models for companies that incorporate faulty assumptions concerning the precise definition of what constitutes "free cash flow" and how that relates to retained earnings or capital spending.
you may be referring to some of those, which are good things to keep in mind if you're a DCF freak.
the other good thing to keep in mind is their fantastic, peg-the-meter sensitivities to certain exponential rate inputs.
but speaking generally, from theory, the discount rate you choose says nothing about reinvestment assumptions. it says everything about what you mean by "risk free"... including the risk that rates will be raised (a second-order or third-order term, but still there)
<<it would suggest a business valuation of 53 times trailing earnings, which just doesn't make a whole lot of sense. >>
It doesn't make a lot of sense because historically you could get it a lot cheaper and the opptunity risk is far more on the downside than the upside.
If you eliminated the short term volitilty, the 53 times seems quite a bit more sensible.
When facing a very long discount rate the decision should also be based on those very long periods.
If you were a unalterable choice of allocating 10% of a 50 year time deposit to KO stock or to a 7% compounding account, with no opportunity to sell between now and 50 years, at what price would you chose to make some allocation to KO. G
Given that the current long term interst rates are 5% and trying to get that in a tax deffered compounding vehicle is even more difficult, a rational buyer with long time frames could make a KO investment and come out ahead... or with a large margin of saftey or risk reward factor.
I think a prudent thing to do with the DCF analysis is to use a long time frame, but assume a lower residual at the end....say KO at 15 times earnings at year 50....or at the very maximum the current mutiple.
I'm not that addept in doing the calculations, but lets say I assumed a 2% dividend yield, (the dividend yield has been neglected either the earnings have been over counted like they were cash to sellers or the dividend overlood in others examples) a 8% earnings growth rate and a 8% dividend growth rate.
The 2% and growing dividend stream must be valued in itself. Then the residual sale at 15 times future earnings in 2053 would need to be valued.
Using a.80 dividend and 1.76 earnings for this year I don't exactly know how to discount the 50 year dividend stream.
I'd like to know what it was worth to a 5% 7% an 10% investor if it grew at 8% then ended without residual after 50 years.
The stock earnings of 1.76 would grow 47 fold over 50 years...while that seems like an awful lot, probably half of that could be attributed to nominal price level factors, some to growth in population of its markets, leaving only a fraction to new buisness enterprises and penetration into new markets. Still that fraction would still be a challenge as large as it would be in absolute dollars and size of the starting base.
47 times 1.75=82 in earnings 2053 times 15 P/e = 1,233 a share value at 15 times earnings in 50 years. What is the present value of that at 5% 7% and 10%. I've lost my calculator that I knew how to make these calcluations on.
Add the value of the income stream and the value of the residual at 15 times earnings at various discount rates and I think you get an aray of numbers. Ideally you'd also make an aray including various earnings and divend growth rates and resulting present values at various discount rates to get to a level of comfort.
"How about using 5% for year one, 5.5% for year two, 6% for year three, etc. That way you ramp up to 10% if that's what you think a "normal" rate would be."
This is a good idea IMO, especially in the calculation of the terminal value where most of the deviations in valuation occur. We certainly forecast expected growth rates and costs, why not use a central tendency in the terminal value before discounting to present?
The advantage there would be that intuitively it would give a better notion of fair valuation.
Otherwise, you'd have to account for it with a fatter MOS.
That said, ultimately the specific valuation method isn't as important as the discipline that is used with it. There are so many places where errors in the valuation process may occur, that it becomes much more important to become aware of the limitations of any valuation technique.
I think also having multiple mental models as a check is useful. If one uses a valuation process and then finds via another that the implied annual ROR is 50% in perpetuity. Maybe then, you might want to check your figures.
Some very smart people who can technically run these models way better and faster than you and I, came up with some absurd valuations, not so long ago. It's not the model, it's how it's done.
Wouldn't it be funny if all this while that we're wringing our hands over how much cheaper KO needs to get, WEB's nibbling at KO.
On the other hand, I'm not sure he's really going to mess with that perfect, round 200,000,000 number unless KO does get cheaper.
Has anyone else ever thought that there was some signal (such as the permanence of the holding) WEB was sending with that nice round number?
I have heard that discounted cash flow models implicitly assume reinvestment at the discount rate. Therefore if one assumes anything but the risk free rate, then you make assumptions about the reinvestment abilities of the business that may not be valid.
Because I don't know where interest rates are headed, I assume that in the future they will return to their historical rate. So (like in your suggestion) I discount near term cash flows at the current 30 year treasury rate and increase the rate towards 7% (a rounded up version of the historical rate of 6.5%) over the first five years, stay at the historical for the next 5, and use 1% for infinity and beyond (thanks Buzz Lightyear).