- Selling or avoiding overvalued stocks necessarily means we assume a stock ought to sell for what it’s really worth; i.e., that P (price) = V (value).
- A better view of market equilibrium, inspired by a 1984 Robert Shiller paper, maintains that P is really equal to V plus something else, which we’ll call N (noise).
- We can easily quantify the impact of value and noise in any stock price and take action based on an analytic framework most appropriate for that particular stock.
Fill in the blank: ___ is way overvalued; Sell!
Choices: (NYSE:A) Amazon.com (NASDAQ:AMZN); (NYSE:B) Tesla Motors (NASDAQ:TSLA); (NYSE:C) LinkedIn (NYSE:LNKD); (NYSE:D) Facebook (NASDAQ:FB); (NYSE:E) Plug Power (NASDAQ:PLUG); (NYSE:F) all of the above; (NYSE:G) none of the above.
Answer: g, none of the above! (Feel free to light your torches for flaming!)
If you made any choice other than g, then you are presuming stocks should trade at fair value and that if this isn't happening, Mr. Market is doing something bad and will sooner or later get his comeuppance giving you an opportunity to profit as the forces of light seize the day; whether that takes place next week, next month, next quarter, next year, or over the proverbial "long term."
This is not the case.
To argue that price equals value (that P = V), you would have to, in effect, suggest that all traders are rational and equally willing and able to calculate the "correct" value for each stock. If that were the case, stock market volume would plummet since there would never be a reason to trade other than personal financial planning (adding new money into your equities portfolio, or withdrawing). Selling A to buy B would make no sense, because if stocks always trade at their correct values, nobody has any incentive to make an incorrect decision to exit one properly valued position or jump into another correctly valued stock, unless for some reason they have a burning desire to incur needless transaction costs. Aside from the broker, there would be no profit opportunity beyond that offered by the market as a whole. (Sound familiar? Think efficient markets.) But there is trading; lots and lots of it, far more than is dictated by personal finance. Are there really that many crazy people out there? Or might the efficient market theory be, to put it as politely as possible, not quite correct?
A Challenge to the Efficient Markets Crowd
Way back in 1984, Robert Shiller (who recently won a Nobel prize for his work in asset valuation, together, ironically, with intellectual rival Eugene Fama) issued a bold but under-recognized challenge to efficient markets adherents (the folks who contribute so much to the misconception that P = V) and argued that the market actually consists of two types of investors: smart money (those who "respond to rationally expected returns" subject to wealth constraints; or in plain English, those who really know how to calculate value and actually do it), and ordinary investors (those who "have no model or at best a very incomplete model of the behavior of prices, dividends, or earnings of speculative assets").
It's tempting to think of ordinary investors as being lazy, ignorant, unsophisticated, etc., and undoubtedly, there are some (perhaps many) that fit this mold. But what about the stocks mentioned above; AMZN, TSLS, PLUG, LNKD, and FB. Can you, or anybody, rationally compute expected returns? Seriously! Can you really and reliably compute the present value of expected future dividends? Can you really and reliably compute the present value expected future cash flows? Even if you plug away for as hard as you must in trying to do it, could you be confident, so confident as to bet your retirement funds on it, you're your answer will match those derived by other rational diehards trying to do the same thing? And if your number is different, can you really be so sure you'd be right and they'd be wrong? Can you even rationally look at today's financial statements for any of those firms and confidently infer that these are truly representative of the company's likely future earnings power, to the point where you could multiply them by valuation ratios you are sure are correct and say "Aha, something is fishy, those stock prices are unjustified." (Regarding the latter, people do it all the time, but their ability to profit from doing this tends to be more anecdotal and erratic than consistent and systemic.)
Price Is Not Just a Matter of Value
Not only is P = V practically unrealistic, it's not even sound theory. It's better to say P = V + Something else. To get at this other thing, Shiller offers a theoretical model that in typical academic fashion uses some Greek letters which I'd rather not try to reproduce. (What's up with that? Did Aristotle anticipate the development of Anglican letters and use them when he wanted to write fancy equations?) In simple - very simple, stupendously simple - terms, P (price) is a function of the present value of demand for shares by smart money, (rational investors who calculate value based on the present value of expected future dividends, or whatever) plus the present value of demand for shares by ordinary investors multiplied by friction (the latter is my word and it represents such things as transaction costs, holding costs and, importantly, information costs).
Here's the payoff: As we learned in elementary-school math, anything multiplied by zero becomes zero. Therefore, as friction approaches zero, then so, too, does the impact of ordinary-investor demand leaving the field clearer for domination by smart money. In fact, if F actually reaches zero, then so, too, does ordinary-investor demand thus giving rise to a state of affairs wherein P really would equal V). On the other hand, as friction rises and approaches infinity, the impact of ordinary investors grows and eventually dwarfs the impact of smart money. For the sake of clarity, and thanks to Stanford's Dr. Charles M.C. Lee, let's change our terminology a bit. We'll refer to smart money demand as "value" and ordinary investor demand as "noise." So now, P = V + (N * F).
Let's go back to our earlier question: What might you make of computation of fair, reliable and rational valuations for AMZN, TSLA, PLUG, FB and LKND? What might it take for you to accomplish the task, assuming you can do it at all? More to the present point, what might you make of F, friction, particularly, the information-cost component? I'd say F for these stocks is high, stupendously high and perhaps even within hailing distance of infinity. So what role would you expect noise to play relative to value? I'd say pretty darn close to 100%. Given that, how should we react if we encounter commentary asserting that any of those stocks are overvalued? I'd say the best possible answer is "Well duh!" I might make similar observations about developing nation equities (many of us have had some interesting experiences with China for example). And we could probably make similar observations about the smaller end of the U.S. market (nano-caps as I call them).
What about Wal-Mart (NYSE:WM) or Coca Cola (NYSE:KO)? These are different. While it's always a challenge to compute theoretical valuations, some situations are harder than others and as this sort of thing goes, you probably don't need to be a rocket scientist to come up with rationally defensible sets of numbers for these stocks. WMT and KO are established companies about which we know quite a lot. So I'd say that here, F is a heck of a lot lower than it is for the other stocks I mentioned, and that V is likely to play a much larger role in driving the stock prices.
Let's Make This Actionable
Theory is great. The ability to make better real-world investment decisions is better. So let's take the next leap. To do that, we'll have to agree that reasonable approximation trumps false precision. That's OK considering that doing so would put us in pretty good company: Note the adage widely attributed to John Maynard Keynes but based on a statement by British logician Carveth Read, "it's better to be roughly right than exactly wrong," or an observation by Fischer Black, a quasi-Nobel laureate (his work on the options pricing model was recognized by the Nobel committee in 1997 as they awarded the prize was awarded to his collaborators, but Black was excluded because these prizes are not given posthumously), after moving from Massachusetts Institute of Technology to Goldman Sachs to the effect that "markets look a lot more efficient from the banks of the Charles than from the banks of the Hudson." So for starters, I'm not going to get neurotic trying to quantify F. I'll assume that F and N are subsumed into one grand item, which I'll continue to call N (noise). That allows us to present a simple and now quantifiable equation: P = V + N.
We always know what P is. So once we quantify value, we'll know the impact of noise (it's just P - V, price minus value). Easier said than done, you might think given the way people argue so much about value. But think carefully about those arguments. Is value really so complicated? Or does it just look that way because many don't recognized the impact of noise and are getting it mixed up with bona fide value? If we understand the distinction, we can actually handle V with a quick easy-to-implement rule of thumb. Here's an idea for how to do that (and remember, I'm big on approximation right now): If you have a bond that is paying you interest of $50 per year, and you know that the cost of credit is 5%, what's the value of the bond? Obviously, its $1,000; computed as $50 divided by 5%. In other words, value is income divided by cost of capital. We can also do that for companies.
If Company A's income is $375 million a year and if cost of capital is 8%, what's the value? Again, its income divided by cost of capital, which in this case, would be $375 million divided by 8%, or $4.69 billion. That's V. Let's now assume market capitalization is $6.5 billion. If that's the case, then we know that N (noise) accounts for $1.81 billion of market cap or 28% of the total. (I know I haven't addressed growth. I'll get to it soon.)
For the income figure, I'm not going to use net income or EPS. I'm going to instead use Net Operating Profit after Tax (or NOPAT) because I want to be capital-structure neutral and because I don't want to bother with non-operating items (the Graham & Dodd classic makes a big deal of cleansing the analysis of this stuff; we'll let it find its way back into the price via the noise component). NOPAT simply is operating profit after depreciation (or EBIT) multiplied by one minus the tax rate.
Cost of capital is easy to define (which is why just about every author, guru, teacher, etc. does so) but it's beyond brutal to apply in the real world for reasons too comprehensive to be addressed here. But I've found that you can get by with a lot of approximation. So I'll just assume cost of debt is three percentage points above the ten-year treasury, cost of preferred is a percentage point above the cost of debt and cost of equity is five percentage points above the cost of debt at which point I can take a weighted average based on a company's capital structure. Please don't bother me about the details; if you want different spreads, that's fine. Do your thing. I'm just interested in something that relates to balance-sheet structure and changes automatically over time as market conditions change to facilitate testing, which I do a lot. And speaking of testing, I've examined a lot of approaches to cost of capital and you'd be amazed at how little we accomplish when we try to max out on precision (mainly because the more precise we try to get, the more we stumble over unknowable things and erroneous assumptions).
So here's a simple example:
Operating profit (EBIT): $575 million
Cost of equity: 8%
Tax rate: 35%
Market Cap: $6.5 billion
NOPAT = $575 million * (1 - .35) = $575 million * .65 = $374 million
V = $374 million / .08 = $4.675 billion
N = $6.5 billion - $4.675 million = $1.825 billion
Impact of N = $1.825 billion / $6.5 billion = 28%, therefore . . . .
Impact of V = 100% - 28% = 72%
It's just that simple. In the Appendix below, I'll show you exactly how you I do this with a stock screener. You don't have to go that route. If working with an individual company, you can use a single cost-of-equity assumption across the board, look up the items you need on any one of a bunch of advertising supported (i.e. free) web sites, and then do the rest with a hand calculator, or rather, a smart-phone calculator app, even a simple free one. It's even better if you have one of those cheap spreadsheet apps. But if you want to use these ideas to build actionable investing strategies, as I tend to do, stock screening is the way to go.
Before looking at how the stocks mentioned above fare, I need to address growth. Notice it's not part of my computation of V. That might strike many as odd considering that even the Ivory Tower bedrock theory, the Gordon Dividend Discount model (price equals dividend divided by the difference between required return and growth), includes growth. And growth is, after all, an important difference between stocks and bonds and the reason why equity investors so often tolerate low- or zero-yield. Here are three responses:
1. Growth estimates tend to be unreliable so it's best to consider their role to be part of noise.
2. Many like to use a margin of safety; assuming zero growth satisfies that.
3. Be patient; in a future article I'll replace NOPAT divided by cost of equity with another more involved (but still very manageable, even in my screener) valuation model that includes a non-noisy growth expectation.
Whichever answer(s) you pick, understand that for now the estimates I'll show for the value component of price or market cap are conservative, making the estimate of the noise component a bit high. (But as you'll see, even these numbers will appear reasonable.)
Breaking down some stocks
First, let's set some context by considering the broad market.
I'm going to eliminate financial stocks and companies that report deficits in items necessary to do the computation. So, for example, if a company's EBIT is negative, assume no role for value and that the entire stock price is based on noise. Among companies for which the calculation could be done as of 3/17/14, noise contributed, on average, 30% of market cap for the large-cap stocks in the Russell 1000 and 40% for the smaller-skewed Russell 2000 universe. Given the role of friction and the likelihood that the information-cost component of friction would be higher for smaller issues, we shouldn't be surprised to see a larger degree of noise among these stocks.
Let's flip it around. Based on those numbers, we recognize that value accounts, on average, for 70% of market capitalization among Russell 1000 stocks that can feasibly be valued, and that value accounts for 60% of market cap among Russell 2000 stocks with usable numbers. Bear in mind, too, that my value estimates assume zero growth so are, therefore, probably too conservative.
Consider what this means. Notwithstanding the rallies we've had of late that pushed stocks upward despite a still ho-hum economy, value remains very important in the investment community. (That may not be readily apparent from what people say, but it's much more noticeable considering what they're doing, as evidenced by the numbers.) The stocks that are noisiest are, for the most part, those that had to be excluded from my screen because of inadequate data. In other words, the market is most inclined to bypass value where it has no choice! On average, if a stock can be feasibly valued, then estimates of value are, indeed, be the main drivers of the stock price.
That's huge. So much commentary implies or out-and-out accuses those who own overvalued stocks of being stupid, lazy, deluded, etc. It's the inability to reliably estimate value, rather than the forces of darkness, that lead investors to downplay or ignore value for various stocks.
Let's look at some specific stocks starting with Figure 1, which shows the value/noise breakdown for a selection most of us would probably deem pretty solid (whether we'd buy, sell or hold at this moment isn't the point; the idea is that we understand the companies and can make reasonable estimates of value).
This looks pretty much on script. Because Mr. Market can value these stocks, he does value them: He does a reasonable job of setting his prices, not based on the proverbial manic-depressive mood swings popularly attributed to him but based on reasonable use of the numbers.
But even here, N is not equal to zero. Given that we can't see into the future, even the best and most reliable estimates of value involve at least some uncertainty. If you really need proof of that, I suggest picking up any of the valuation books by Aswath Damodaran and applying the techniques to the stocks listed in Figure 1. You'll be able to come up with reasonable answers, but it won't be easy nor will your conclusions be iron clad, since you'll encounter many instances in which you could justifiably change some assumptions and come up with different numbers.
Now let's look at some other stocks. I assume we all know what these companies are about (if you don't, do a ticker search on Seeking Alpha; you'll find plenty of content that will get you up to speed quickly).
All of these situations are united by extreme difficulty in calculating reliable valuations, typically because their businesses are all, to one extent or another still "emerging" thus making it pretty much impossible to calculate the sort of operating profit figure that would satisfy Graham & Dodd regarding its representativeness. Note, though, that this isn't black or white. Some situations are more extreme than others.
LULU, for instance, is the least noisy among these stocks and that makes sense. Its basic business model, specialty apparel retailing, is pretty well established. What's less readily quantifiable, however, is the niche the company trying to create with high-end yoga/fitness apparel. Questions of size and profit potential are unanswerable at this point in time and open the door to noise, a lot less noise than we see with PLUG, but a lot more noise than we see with WMT.
GOOG and GMCR are similar in that they both fairly new in the grand scheme of things, but have been around long enough for us to have a solid sense of what they do and the sort of profits they might earn. More debatable and less quantifiable, however, are ongoing evolutions in their respective areas (GMCR has a higher-end coffee maker and has been broadening distribution, and GOOG is still carving out its place in the cloud) that generate noise, again, not as much as at PLUG, but a lot more than at WMT. AMZN, too, is very much in a state of flux regarding its place in the cloud, the logistics of its core merchandising activities, and even content creation.
Meanwhile growth expectations, a legitimate and important aspect of value, are often likely to be seat-of-the-pants or may require serious input from high-priced consultants or specialists who may or may not, at the end of the day, improve all that much on seat-of-the pants guesswork. Damodaran's books do, actually, suggest how you might approach this (he even has one book called The Dark Side of Valuation: Valuing Young, Distressed and Complex Businesses). But even these processes involve considerable subjective judgment and Damodaran readily concedes you're not really aiming to get this right; you'd just be hoping to be less wrong than others (I can't recall if this point was in any of his books, but he clearly drove it home at a presentation I attended in which he demonstrated an approach to valuing FB which, as I recall, was still insufficient to explain all the price action we've seen since that time). Ultimately, we have to recognize that for stocks like these, the information-cost component of F, friction, is very high, thus elevating the role of noise.
This alone is important. If someone suggests that you should short or buy puts in AMZN or TSLA because the issues are overvalued, listen further at your own risk. Value isn't influencing either of those stocks, not due to the forces of darkness (as many suggest) but because value can't be reliably computed. So it's pretty much all noise.
This doesn't mean you can't make money on the short side. But to take that position, forget value. Analyze the noise on its own terms. Dig into the prevailing market narrative and look for reasons to believe it may improve or deteriorate near term. For example, is the GMCR narrative too optimistic? Has the company come closer to maturity than many suggest? Is the market too optimistic about where AMZN and NFLX can eventually take their businesses? Maybe that's the case. Either way, don't focus on EPS, margin, P/E or even debt (when the capital markets believe in a business, debt can be refinanced as often as it needs to be) or anything like that. That's not what drives the stock. It's about products, markets, competition, and other qualitative narratives. When you need to quantify, as many (including myself) often do, look for numbers that can give clues to trends in investment community attitudes toward such topics. Think about short interest, insider activity estimate revision, momentum, technical analysis, etc.
By the way, what did you think of AAPL and MSFT as presented in Figure 1? Were you surprised to see noise being so low? It seems like those two generate more than enough noise to shatter anyone's metaphorical eardrum. Do I really need to count how many pixels have been generated on Seeking Alpha alone on this karmic (in my opinion - I believe Steve Jobs and Bill Gates had many past lives together and will likely have many more pairings in the future) duo? What I'm talking about here is investment noise, stories about which the investment community actually cares. Just because somebody says something about a stock, pro or con, that doesn't mean investors (the ones who move money rather than words) are interested in what they have to say.
Use the value/noise breakdown as a starting point for your analysis of a stock. You can't get the right answers unless you start with the right questions. Value/noise can point you toward the right questions.
For example, if you're interested in AAPL or MSFT, you really need to understand the numbers before you think about anything else. If you're interested in AMZN, NFLX, or TSLA, don't obsess over the digits. Tune into the narrative around the stock and where you think it's heading, whether it's likely to get more or less favorable.
Are you allowed to disagree with Mr. Market? Could you rightfully believe the market is underappreciating Apple's potential to evolve even now, given how entrenched it's become? Yes, absolutely. That's where the fun and profitable action can be.
Let's take rumors about Apple making an iWatch. Mr. Market can read. He knows about it and so far is choosing to dismiss it. So I would not sell AAPL merely if I think iWatch is fiction. The market is probably not really factoring this sort of noise into the stock price, so there's probably no money to be made if the rumor does, ultimately prove false. On the other hand, if I believe in it and I believe it can be large enough soon enough to compel Mr. Market to reassess his views on Apple, that could justify buying.
Speaking for myself, I'm more a numbers guy and wouldn't likely move either way with regard to AAPL and iWatch rumors. But this example should give you an idea of how one might quantify a value/noise investing strategy. One such model I have screens the Russell 2000 (a group likely to have more noise in general) looking for stocks for which noise is conspicuously low, 25% or less of market cap, and where sentiment data suggests noise may be starting to increase and in a bullish direction. But I'm getting ahead of myself.
So far, I've been neutral about value and noise. I haven't discussed whether value or noise is likely to be bullish or bearish for particular stocks. That will be the focus of my next article. After covering that, I'll move on to specific stock-selection strategies.
In Portfolio123, this model can be created using the free-form screening interface. It makes liberal use of the SetVar function, which allows us to define an item in the context of the screen by naming it @item and then specifying its formula. Once that's done, we can then use @item as we would any other screening factor. This model has been saved here on Portfolio123 with visibility available to all Portfolio123 users.
Here are the rules (note that rules on Portfolio123 are not case sensitive; capitalization here is presented just for visual convenience):
NOTE: The above items establish the cost of each kind of capital
NOTE: These items define capital and the percentage of capital represented by each category
NOTE: This is the weighted average cost of capital
IncTaxExpTTM>0 and OpIncAftDeprTTM>0 and PfdEquity(0,qtr)>=0 and ComEq(0,qtr)>0
NOTE: This rule eliminate companies lacking one or more important data items
NOTE: These three rules actually calculate the role of noise
NOTE: This last rule identifies the ticker(s) of the stock(s) that interest you
To see the results, run the screen using the pre-defined report entitled "Screen Factors." The column in which you are most interested is the one labeled @NoisePct
If you want to simplify the cost of capital computation to select a specific across-the-board number, you can eliminate the first eight rules and substitute revise the next to the last rule as follows, assuming in this example, you want to set the cost of capital at 9%.