Let me try to explain valuation models this in simplified terms:
A P/E analysis first assumes something about what kind of company you have, typically by comparison with others. Then it assumes that for this kind of company, each $ of earnings corresponds to a certain company value. Upthread, for instance, Sleepyhead used Toyota as the basis of comparison, and then "added some" ("if each $ earnings in Toyota is worth $25 market price, then it should be at least $30 for Tesla). Used this way, the method does not really look at fundamentals - if Toyota is overpriced, then the value you find for Tesla will be overpriced as well. However, instead of using comparison you can build a P/E up from fundamentals - for instance you can assume a typical discount rate to find that a company with flat earnings for the foreseeable future should have a P/E around 7.
A DCF analysis looks at the present value of a company for an investor who buys and holds it forever. That investor cares about how much dividends the company will be able to pay (this is the fundamental basis for all share prices), which also is largely reflected in earnings. So you project the earnings for the foreseeable future, and then you figure out how much that is worth in today's money. So this metric looks directly at the fundamental value of the company, based on assumptions.
(A P/S analysis, by the way, assumes that each $ of
sales should correspond to a certain stock price. While it can be used for an occasional reality check, it disregards even the profitability of the company (in addition to all the other simplifications of the P/E). It was used during the dotcom era to defend sky-high valuations of unprofitable growth companies. Most of them never reached profitability. However, comparing to companies of equal profitability with P/S gives exactly the same result as a P/E.)
The DCF approach is even better with stable companies than with growth companies (or depending on your POV, is less unreliable for mature companies). DCF is premised on the efficient market hypothesis that a company's value today is equal to the expected value of future cash flows.
No, DCF is not premised on efficient markets. It provides an estimate of the fundamental value of the company, regardless of the efficiency of markets.
Efficient markets only come into the picture if assume that long term fundamentals are reflected in the share price (which I think is a basic premise of this thread, regardless of which method you use to assess those fundamentals).
With a stable company, the assumptions needed to estimate future cash flows are less heroic.
Yes, it is easier to value a stable company, but DCF does not make it simpler or harder. With DCF will scratch your head over what to assume about future cash flows, and correspondignly with P/E you will scratch your head over which P/E number to chose. Each method has advantages and disadvantages in this respect:
- With DCF, it is easy to get an answer that is "exactly wrong". You fill in a spreadsheet with assumptions of what the capital requirements will be in 2022 and stuff like that, basically pulling the figures out of your ass. Then you show the spreadsheet to someone, and they think "wow, how detailed he has worked that out" and believe they have The Right Answer. On the other hand, the advantage is the understanding you get by
checking and varying the assumptions. By trying different scenarios, you can learn a lot about which assumptions are the key ones, and then you can make some scenarios (and ideally, as I advocate for high-uncertainty situations, you do a probability-weighted average of those). For instance, Prof. Damodaran's model showed that his assumptions for capital requirements and employee options were quite important, while some others were not. To a certain extent you can say that the model is nothing, the modeling is everything.
- A P/E hides the assumptions in a black box. If you for instance use a comparison with Toyota, you are basically assuming that the value drivers of Tesla and Toyota are similar. You disregard the possibility that the future earnings trajectory may be quite different, that there may be fundamental differences in capital structure, capital intensity, hidden liabilities, brand value, tax rates, ZEV credits, new business opportunities etc. (or at least you assume that all these factors even out and on average don't matter). As used by Sleepyhead, it is even worse, because he is using a ratio comparing the Toyota of today to the Tesla of 2018. A skilled P/E user such as Sleepyhead will try to take the potential differences into account and guesstimate a correction. Just as with DCF, an unskilled user will mindlessly apply a ratio without knowing what he is really assuming. This is the garbage in/garbage out factor playing out alike for both methods. But I would say that the disadvantage even for Sleepyhead is that he can easily overlook some factors that could be important in the analysis. Also with P/E it is important to make scenarios for a company like Tesla, because your One Main Predicition will most likely not hit the mark. And finally, P/Es are very cyclical.
Today's levels have only been surpassed twice since 1880, so may not be sustainable. Using comparative ratios in this market may give a result that overshoots fundamentals.
Also, because one applies a discount factor to future earnings, the flat profile of a mature company's earnings doesn't depend as much on the discount rate you apply (which, properly considered, should include risk factors, which are also hard to estimate).
Actually this is wrong. Risk factors (as in "uncertainties about future cash flows") should go into the cash flow model - either by making the projection a "middle-of-the-road" scenario or by doing multiple scenarios and then do a weighted average based on probabilites. The only risk that should go into the discount factor is the cyclical risk (so-called
beta). The discount rate will always be contentious, but for a stock where the debate is whether the right value is $50 or $1000, other assumptions will be much more important than whether to use 7% or 10%. In any case, this assumption is hidden any P/E ratio you might use, so even if you don't see it you are still assuming a certain time value of money (you just don't know which).
Long and the short of this: I think a DCF model for valuing Audi stock is reasonable, but not so for TSLA.
I think the exact opposite: Audi is a know quantity, and you won't go very wrong with a P/E. With Tesla, there is more value in digging into the underlying assumptions, which is what you do with a DCF.
It seems to me, trying to understand this, that in addition, if you compare the DCF value to an ordinary stock price, you would be comparing apples to oranges. Under the assumption that Audi's revenue remains constant in the future, the DCF value of its stock price would probably be much lower than the actual stock price. Meaning, if I understand this correctly, that the DCF value is much lower than the prices one might compare it to (if one doesn't know it is a DCF value). One might discount the value twice, if you follow my thought. A percentage of 10% per year was mentioned, which to me suggests that DCF "prices in" a certain growth factor (and one might not be aware of that). If that is the case, and one hears the DCF value of a theoretical company with constant revenue, one would think its price will go down, whereas in reality, it might remain constant.
You are right that using DCF today's value will be lower than tomorrow's, but that is the expected behaviour of stocks. If you predict that the stock should be at $500 in 2030, you most probably should not buy it at $300 today (because it would only give you 3.1% annual return). DCF factors this in, so that when it outputs a value for today, it takes into account that you want the stock to rise at a certain minimum rate to want to hold it.
So when you look at the per share result from a DCF, you compare it to today's stock price - if the DCF is higher, that indicates that you might earn more than the time value of money by investing.
I hope this clarifies a bit. I am a bit surprised that there is an almost ideological debate about different valuation methods. They are all tools in a toolbox - use them correctly and when appropriate. Or feel free not to.