The other day, my friend Eddy Elfenbein wrote a post entitled “The Limits of Fundamental Analysis,” which made some good points (as he most always does), but in so doing, made far too broad and thus inaccurate a conclusion. The premise is (and I welcome Eddy to correct me if I’m off-base) that in some circumstances, fundamental analysis is inappropriate, such as when dealing with transformational businesses like Amazon was (is), cyclical firms, leveraged firms, firms with varying earnings quality, etc. The thing is, we can – and do (try to) – adjust for all of these things – and more – when performing a proper fundamental analysis! I’m going to attempt to show how, by making relatively small changes to a few key assumptions (sometimes even just one number will do!) in a simple DCF framework, we can grossly change not just how much we think a stock is worth, but why it’s worth that price, as well.
Fundamental analysis, by definition, involves examining the industry in which a firm competes, the regulatory/legal environment, the market for a firm’s goods/services, the goods/services themselves, the firm’s financial condition/performance, strategy, capital structure, reliability of financial statements/accounting controls, and several other factors, not only currently, but in the past and, more importantly, the future as well. Fundamental analysis isn’t just looking at a few ratios on Yahoo Finance or Finviz or whatever and concluding a company is a good (bad) investment based on valuation, liquidity, solvency, and/or other metrics. That can be the starting point for narrowing down firms which are more (less) likely to be worth investigating further, given a finite amount of time to allocate to identifying and researching ideas which we hope will help us invest wisely.
A quick tangent: If you’re not in the markets to invest, you are in the markets to gamble. This is not up for debate, it just is; if you find or fancy yourself a gambler, save the trouble and head to your closest casino where they’ll be more than happy to separate you from your presumably hard-earned money quite expeditiously (you may even get some “free” food and drink out of it). If you’re not sure whether you’re trying to invest or gamble, just put your money under your bed until you figure it out, you’ll be doing yourself a favor. If you’re interested in making good investment decisions, this is where the fun starts…
Since 2011, I’ve written at fair length about YOKU, and, since it is (or was at the time) a rather unconventional firm, I’ll be using it as an example to prove that, contrary to apparently popular belief, you can, in fact use a conventional approach to value unconventional companies.
Online video company YouKu (YOKU) is a good example of an unconventional firm/in an unconventional situation that Eddy’s post alludes to being inappropriate to analyze using a fundamental approach: The firm is relatively young, does business in China – which brings with it a mountain of other concerns for another time – is in an ultra-competitive, high capex, high growth industry, with limited negotiating power relative to its suppliers and customers, subject to the wavering tastes and unpredictable swings of fickle viewers and hesitant, inexperienced internet advertisers alike. YOKU still has negative earnings and cash flow despite enormous revenue growth, due in no small part to high(er) than (many) predicted operating expenses, content acquisition/production costs, etc.
Back in early-mid 2011, after the company went public, received enormous support from the analyst community, did a secondary offering, and still received oodles of praise from the aforementioned analyst community, I started learning about YOKU; the company, the industry, the target customers/users, the regulatory/legal landscape and the like. From this, I developed several scenarios, each with its own set of assumptions for revenue growth rates, margins, capex, interest rates, etc. (NOTE: The below image shows just part of the assumptions section of the model)
These assumptions – here using the second most optimistic scenario I’d developed – then feed into the model from which I derived a valuation for the stock of around $21-22/share.
As you might be able to tell, instead of using a common multi-stage growth model for revenues (e.g. x% for the 1st 3yrs, y% for the second 3, z% for the third 3, and ?% thereafter). Instead, I decided to model revenues starting from a base growth rate drawn from historical & competitor growth rates for the current year (next FY), growing in each subsequent year but at a gradually lower rate. In simple terms, revenue growth decays at a constant % year over year, in this case, -18.5% from the prior year. Look what happens if I assume the company can maintain a higher level of revenue growth each year, i.e. the growth rate decays at a slower pace, let’s say -17.5%/yr instead of the -18.5% I used initially. Boom! The projected stock value jumps +~15% from the mid $21s/share to the mid $24s! All I did was change my assumption for how well the company will be able to maintain even higher than silly-high revenue growth year to year; I changed one number in my model, by only 1% (100bps), et voila! the company is worth 15% more than it was 5 seconds ago.
If I start playing around with sets of assumptions (scenarios) instead of just altering one at a time, then things start to change a LOT more drastically than that. Let’s say I now think the market for the firm’s products/services is smaller and will not grow as fast as it and other analysts expect, the company will grab less share of that smaller and slower-growing market, and expenses will be higher over time due to wage inflation, sub-optimal expense management as the firm grows rapidly, and increasingly high content acquisition and production costs. Using this scenario (and the assumptions underlying it), the $21 initial valuation we established using a rosier set of real-world assumptions about the company/market/industry/etc drops all the way down to a measly $4.84! For those playing the home game, by simply using a scenario that reflects our new-found relative skepticism, and changing literally one number in my model, my estimated value of the stock drops by almost 80%!!!
In his article, Eddy wonders how we could use fundamental analysis to value a stock like Eastman Kodak 20 years ago, Amazon during the height of the tech bubble, or Tesla today (Read NYU Stern Professor Damodaran’s work on TSLA if you want another company-specific example). The answer is simple, at least conceptually:
1. Do the research on the industry/market/company/products & services/etc. Lots of research. (I’d elaborate, but there’s entire textbooks on the subject)
2. Use that research to develop a few scenarios (at the very least three: bearish, base, and bullish) with what you think are appropriate growth rates, expenditures, etc for each.
3. Once you’ve done that, run it all through a DCF (or similar) model using your base case to get your base valuation.
4. Repeat for the other cases.
5. Do a sensitivity analysis on your valuation with each scenario’s valuation output and another factor, say, the discount rate (WACC) you used to discount the projected future cash flows.
6. (Optional): Assign probabilities to each scenario and weight each one accordingly (although this should already be partially captured in the assumptions used in each), to come up with a probability-weighted average valuation.
I’m grossly oversimplifying the fundamental analysis process, of course, but the point is that you can do this for ANY company at ANY time, in ANY situation (with minor tweaks). Whether the assumptions you’ve used in fundamentally analyzing a stock turn out to be accurate is still a question mark, but what you’ve (hopefully) done is taken a disciplined, systematic approach towards establishing the value of the company based upon real world facts facing the business/industry/market/global landscape (etc), and how you think those facts will change in the future by developing (and developing, and questioning, and refining…) your assumptions and scenarios. Could you simply look at a bunch of charts of comparable firms and extrapolate the stock price based upon those, or just apply industry ratios from Yahoo Finance or Finviz to the firm you’re analyzing? Sure, but you could throw a dart at a wall covered with numbers after drinking a bottle of scotch, too.
Ultimately, which approach you use is up to you, and the effectiveness of one over another (e.g, fundamental v. technical analysis) is dependent upon too many variables to conclude that one is necessarily worse than the others. For example, Eddy wrote:
There’s always some innovation going on somewhere that threatens to upend the entire game, and fundamental analysis won’t see it coming.
While this may (or may not) be true, I am not sure how any other method of analysis overcomes and accounts for this any better than a fundamental one does. In fact, I think fundamental analysis gives one a better, if not the best, chance relative to other methods of predicting the future (especially longer-term), which, let’s be honest, is what we’re all trying to do. Why? Because the advantage of proper (Key word) fundamental analysis is that it’s based upon a wide, stable foundation in factual research, utilizes a repeatable, robust process which allows us to account for a variety of future outcomes and events, (with which others may or may not agree) and quantify those things systematically to arrive at a valuation (range) based upon any and all knowledge which we care to obtain. Who has better odds of nailing a valuation? Someone who doesn’t understand – even at a high level – a firm’s technology, industry position, financial situation, strategy, etc, or someone who does, and incorporates that knowledge into their analysis? I think the answer there, especially over longer time periods, is pretty obvious, no? Other forms of analysis may work, too, but chances are fundamental works better, even when the full-blown work-up isn’t completed and/or is impossible to perform due to lack of historical data, lack of comps, or whatever.
To put it simply: Every method of analysis that attempts to predict the future (which is all of them) has it’s plusses and minuses, but to say we can’t use fundamental analysis in certain situations due to inherent limitations of the methodology is silly, at best, as I hope I’ve shown here. Fundamental analysis gives us the leeway to account for, develop, and quantify how we reasonably expect a business will fare under any number of circumstances, and what the value of that business is under each one of them. The only limit to fundamental analysis is the ability of the analyst to use the information available to him/her to develop reasonable assumptions and scenarios. If an analyst 20 years ago didn’t see digital revolutionizing the photography business (despite Eastman Kodak being a source of early digital photography testing in the early 1970′s), and Kodak’s strategy of sticking to film as an impediment down the road, the analyst failed, not the analytical methodology or model. We can say the same thing about tech-bubble Amazon; while it appeared massively overvalued at the time, that was because analysts had yet to realize the enormous disruptive potential and thus used growth rates that were far below those appropriate for such a firm, which had already demonstrated even then its disruptive nature. Saying fundamental analysis fails as a methodology to account for these things is a straw man argument, at best; the methodology didn’t fail, our ability to use available information to predict the future did. Is there any analytical methodology in use today that yields accurate predictions even when the user invariably fails to be clairvoyant?
None that I know of.
Rather than blaming any particular analytical approach, it’s far more effective to take a long, hard look in the mirror, because the fault most often lies not with the practice, itself, but rather with the practitioner.