Sunday, February 1, 2015

Discounted Cashflow Valuations (DCF): Academic Exercise, Sales Pitch or Investor Tool?

In my last post, I noted that I will be teaching my valuation class, starting tomorrow (February 2, 2015). While the class looks at the whole range of valuation approaches, it is built around intrinsic valuation, reflecting my biases and investment philosophy. I have already received a few emails, asking me whether this is an academic or a practical valuation class, a question that leaves me befuddled, since I am not sure what an academic value is.  As some of you who have read this blog for awhile know, I do try to value companies, but I do so not because I am intellectually curious (I don't lie awake at night wondering what Twitter is worth!) but because I need investments for my portfolio. In the context of these valuations, I have been accused of being a valuation theorist, and I cringe because I know how little theory there is in valuation or at least my version of it. In fact, my entire class is built around one simple equation:


Put in non-mathematical terms, the equation posits that the value of an asset is the value of the expected cash flows over its lifetime, adjusted for risk and the time value of money. If that sounds familiar, it should, because it is the starting point for every Finance 101 class, a rite of passage that in conjunction with buying a financial calculator sets you on the pathway to being a Financial Yoda! 

That is the only theory that you need for valuation! The rest of the class is about the practice of valuation: defining and estimating expected cash flows for different types of assets and businesses at different stages in the life cycle and estimating and adjusting the discount rate for risk and time value. Note that there is nothing in this fundamental equation that has not been known to investors and business people through the ages, i.e., the value of a business has always been a function of its cash flows, growth potential and risk and that you certainly don’t need to be mathematically inclined to be able to do valuation. So, if you don’t remember how to take first differentials or solve algebraic equations, never fear. You can still value companies.

DCF : Neither Magic Bullet nor Bogeyman
If DCF valuation is simple as its core, why does it intimidate so many? The fault lies both with its proponents and its critics. The proponents, and I would include myself on the list, have undercut the approach's usage and acceptance by:
  1. Over complicating DCF: It is undeniable that most discounted cash flow models suffer from bloat, with layers of detail that we not only don't need, but also make no difference to the ultimate value. These details and complexities are sometimes added with the best of intentions (to get better estimates of cash flows and risk) and sometimes with the worst (to intimidate and to hide the big assumptions). No matter what the intentions are, they make people on the receiving end suspicious.
  2. Over selling DCF: In the hands of bankers, analysts, consultants and managers, DCF models are less analytical devices and more sales tools, backing up a recommendation to buy, sell or change the way we do things.  While that is neither surprising nor newsworthy, it does make those who are the targets of these sales pitches cynical about the process, and who can blame them?
  3. Over sanitizing DCF: I don't know whether DCF's proponents feel that it cannot be defended on its merits or that it is too weak to stand up to scrutiny, but they seem to want to cover up the uncertainties that are embedded into every valuation and play down any hint of story telling that may underlie the numbers or uncertainty in their estimates.
Like anyone who has ever used a DCF, I have been guilty of these practices and therefore understand the motivation. At the core, it is because we are insecure both about our understanding of DCF and our capacity to explain in intuitive terms why we do what we do. If paid to do valuation, we over compensate and believe that we will be more credible if we churn out overcomplicated, number-driven models and that our clients would not pay us, if they realized how simple the process actually was.

Those who critique discounted cash flow models (and I certainly agree that there is often to disagree with), are driven by their own share of sins, where they conflate disagreements that they have with input estimation techniques, the model-builder and model output with disagreements with the DCF process itself.
  1. The Baby/Bathwater syndrome: While it is an analogy that makes me cringe each time I use it, with visions of babies flying out of bathroom windows, it is apt in its description of those who take issue with how an input is estimated in a DCF and then extrapolate to conclude that the entire process is flawed. The input that creates the most angst, of course, is the risk measure used in the valuation, with even a mention of beta generating the gag reflex among old-time value investors. 
  2. Dislike you, dislike your model: The line between a DCF model and its builder must be a gray one, since many critics seem to have trouble finding it. Not surprisingly, dislike of a user because of his or her investment philosophy, personality or style of presentation can very quickly translate into disdain about the process by which he or she values companies.
  3. Don’t like your answer: It is human nature but investors tend to like DCF models that deliver answers that they like and dislike models that do not. Even in my limited blog posting experiences, I have been lauded for using sound intrinsic value models, by Apple Bulls, when my valuations have suggested that Apple is cheap. I have also been blasted by often the same investors for using a flawed DCF model, when my valuations suggest otherwise.
As with the proponents, I think I understand where critics are coming from. After all, if you were constantly the target for sales pitches by analysts who use complicated DCF models to sell snake oil, you would be suspicious too.

A Return to Basics
The first step in spanning the divide is to strip away the layers of complexity that we have built into valuation over the decades and return to the equation that I started this post. At the risk of stating the obvious, I would like to draw on four simple and self-evident propositions that get overlooked or ignored frequently in the discussion of discounted cashflow valuation (DCF).
  1. The Duh Proposition: For an asset to have value, its expected cash flows have to be positive at some point in time, but that does not imply that the cash flow has to be positive every single year and it is quite clear that you can have a valuable business (asset) with negative cash flows in the first year, the first three years or even the first seven or eight, if it can deliver disproportionately large positive cash flows later in their lives. It is true that those whose DCF toolbox has only one model in it, usually the Gordon Growth Model (a stable growth dividend discount model), have trouble with such companies, but using the Gordon Growth Model to value most equities is the equivalent of doing surgery with a  hammer: painful, ineffective and designed to come to a bloody end.
  2. You can hate beta (or modern portfolio theory or all of academic finance), but still love DCF: This may come as news to its worst critics but the DCF model does not come prepackaged with modern portfolio theory and its most famous handmaiden, the beta. In fact, while the discount rate in the discounted cash flow model is usually risk-adjusted and reflects the time value of money, the model itself is completely agnostic about how you adjust for risk (you can come up with your own creative ways of making the adjustment) or even whether you adjust for risk. The DCF model is a descriptive equation of a cash-flow generating asset or business, not a theory or a hypothesis.
  3. It is the asset's life, not your time horizon: A DCF model is designed to value an asset over it's life, and is really not malleable to what you (as the investor looking at the asset) believe your time horizon to be. If the value of an asset is the present value of cash flows over its life, what is that life? It clearly depends on the asset. If you are valuing a machine whose functioning life is only one year, all you need is one year's cash flows, but if estimating a value for a rental building with a 20-year life, it would be twenty years. With public companies that at least in theory can last forever, we do stop estimating cash flows at a point in time and assume that cash flows beyond that point continue in perpetuity, but this is an assumption of convenience, not necessity. In fact, there is nothing that stops you from replacing that perpetuity assumption with one that assumes that cash flows will continue for only 20 or 30 years after your closure year.
  4. You will be wrong, and it is not your fault: If you take expected cash flows (where the expectations are across a wide spectrum of outcomes) and discount those expected cash flows at a risk-adjusted discount rate, it should go without saying (but I am going to say it anyway) that the present value that you get is an estimate of value. Thus, you are almost guaranteed to be wrong when valuing assets with any uncertainty about the future, and more wrong when there is more uncertainty. So what? The market price is just as affected by uncertainty, and you are judged not by how wrong you are in absolute terms but how wrong you are, relative to other people valuing the stock.
Ten Myths about the DCF Model
While the architecture of the DCF model is simple and the truths that emerge from it are universal, there is a great deal of mythology around DCF valuation, some of it promoted by model-users and some by model-haters.
  • Myth 1: If you have a D(discount rate) and a CF (cash flow), you have a DCF. As a DCF-observer, I see a lot of pseudo DCF, DCFs in drag and other fake DCFs being pushed as discounted cash flow valuations. 
  • Myth 2: A DCF is an exercise in modeling & number crunching. There is no room for creativity or qualitative factors.
  • Myth 3: You cannot do a DCF when there is too much uncertainty, thus making it useless as a tool in valuing start-ups, companies in emerging markets or during macroeconomic crises.
  • Myth 4: The most critical input in a DCF is the discount rate and if you don’t believe in modern portfolio theory (or beta), you cannot use a DCF.
  • Myth 5: If most of your value in a DCF comes from the terminal value, there is something wrong with your DCF, since the value rests almost entirely on what you assume in that terminal value.
  • Myth 6: A DCF requires too many assumptions and can be manipulated to yield any value you want.
  • Myth 7: A DCF cannot value brand name or other intangibles. 
  • Myth 8: A DCF yields a conservative estimate of value. It is better to under estimate value than over estimate it.
  • Myth 9: If your DCF value changes significantly over time, there is either something wrong with your valuation (since intrinsic value should not change over time) or it is pointless (since you cannot make money on a shifting value)
  • Myth 10: A DCF is an academic exercise, making it useless for investors, managers or others who inhabit the real world.
Each of these myths deserves its own post and I plan to cover all of them in the next year (one myth a month). Stay tuned!

A Trial Run
I know that some of you are skeptical about my pitch but if you are, at least give the process a try. If you feel a little rusty on the basics or have questions about details, you are welcome to take my class in real time or the online version of it (which is less trying and has shorter webcasts).

21 comments:

Anonymous said...

at what time today?

Anonymous said...

Do you monitor if DCF values predict changes in market values over time and hence are a good valuation guide?

Typically we see DCF values from banks/analysts always have a higher value than public equity market values as the banks use optimistic assumptions. Together with the fact a DCF is a long term valuation and most investors think short to medium term explains why they are not used much.

Anonymous said...

Hi
I tried to enter the webcast and I was asked for username

Ben said...

Dear Aswath,

Thanks for this nice article. I would have two practical questions, where I would love to hear your opinion:

One topic that always arises is the limited scope of comparable companies, both for a multiple-based valuation as well as for the beta derivation. How do you handle this subject, especially when looking at venture / growth companies? Most venture and even growth companies operate in a very specific field (e.g. online retail), whereas the companies publicly trading would typically be active in more than this field (e.g. Amazon - online retail, cloud solutions, logistics etc.). What is your advice on picking a relevant set for the beta? Do you use overall industry groups, for example?

The second topic where I would love to hear an opinion you is "rolling" WACCs. I saw in some rare occasions people using a WACC that starts at a higher rate such as 30% in t=1 and interpolates down to 15% in t=3, with explanations such as "the company is in an unusual state" or "the market is only currently in a downturn and will flip in the next 2 years". Is this complete nonsense, or would you give it a try to overcome temporary obstacles?

Many thanks and as always, thanks a lot for your interesting blog.

Best
Ben

Unknown said...

Prof Domadaran,

I am studying for the CFA and look forward to watching your webcasts as we get into the financial analysis sections of the course.

Thanks so much for posting these webcast!

Anonymous said...

"If most of your value in a DCF comes from the terminal value......"

I guess you would then disregard any DCF model attempting to value Tesla.

Aswath Damodaran said...

Anonymous,
You do know what a myth is, right? The notion that a DCF value is useless if the bulk of the value comes from the terminal value is a myth.

Anonymous said...

Hi professor,

I've seen you use residual income for two purposes: to value financial companies and to evaluate management (your Jan 9th post). Is there a reason you don't use residual income models to value non-financial businesses?

I find myself preferring these types of models over DCF. If you think they're flawed, it would be great to learn why.

Thanks!

Anonymous said...

Dr. Damodoran,

If you are valuing a domestic firm's nominal cash flows that operates in different areas of the world growing at different rates, should we still use the domestic 10 yr as a proxy for terminal growth? I would think so using interest rate parity arguments but I wanted to check.

Anonymous said...

Professor,

A variant of the DCF model that I find to be more insightful is the Discounted EVA model. EVA being a form of economic profit. While mathematically equivalent, investors can learn more about a company or asset's prospects by the trend and level of its EVA which cannot be said about cash flow.

Aswath Damodaran said...

I am sorry but EVA and excess return models are over hyped and over sold. They are not alternatives to DCF but a different way of presenting DCFs. You may gain a little bit of insight into where value is coming from, using these models, when you value mature companies, but if you use them right, they will yield exactly the same values as DCFs. They are not easier to use and in fact are much harder to put into practice, if your company is growing or transitioning.

Walter said...

Mr. Damodaran,

All of this is ok for kids, playing around et et, why not come out and swim with the big boys.

How about this - I dare you to start your own fund so we can see how you perform for real. Sounds fair?

If not you can ignore my call and go back to teaching kids

Aswath Damodaran said...

Walter,
Since I have no need to heed your call, I am going to ignore it anyway and continue to teach kids (though many of my kid used to big boys before they came back into my classes). Since you big boys are doing such a good job delivering returns to your stockholders, you obviously don't need me in your midst.

P.S: Just because I don't charge people to manage money does not mean that I don't invest money.

Anonymous said...

Dr. Damodoran,

Sorry to keep pushing but I left a comment earlier with regards to terminal growth rates. Again, when working with US domiciled companies with international operations in other faster growing nations, should one still stick to the US government bond rate for terminal growth? My original thought was yes, because of interest rate parity relationships but I'd love to hear your thoughts.

Chris said...

Your article never articulated the reasons why I don't use DCF.

(1) It takes too much time. I reject about 50 stocks for one I buy. There is no way I am going to create a DCF model of each. And if I did so, all that effort would invariably 'cause' me to buy the stock because I would than have become so emotionally invested in it.

(2) Valuations either model cash or accounting earnings. Both of those have a zillion possible modifications. Before I throw out the accountant's depreciation cost, to replace it with cash flows, I would have to know -for each piece of equipment- how old the equipment is, how much technological improvements have happened since it was bought, how much longer it will last, whether the replacement will be leased or purchased, what the replacement cost will be, etc etc.

GMAB. I just shake my head in disbelief when analysts tell me that they can model the cash flows for equipment replacements better than the accountant's rough and ready depreciation percentage.

Aswath Damodaran said...

Chris,
Why would I articulate the reasons you don't do DCF? That is for you to do. I use DCF. So, I will articulate the reasons I do DCF. You don't. So, you can articulate the reasons you don't. And if your post conveys the reason, you seem to be saying that it is because you think accounting book value is a more reliable starting point. Well, good luck with that!

Dan S said...

Sir,

Great post.

Given you are producing an estimate, which is likely to be incorrect, how do you feel about scenario analysis, or probability weight DCF?

Thank you

forex tips said...

This is a nice article which will clear doubts from the minds of the trader and will be useful too.

Intrinsic said...

"It is undeniable that most discounted cash flow models suffer from bloat, with layers of detail that we not only don't need, but also make no difference to the ultimate value."

I think "bloat" is necessary to deal with the uncertainties and to understand a business/asset/project/investment. Modelling the underlying business by identifying and testing the variables driving the business is crucial. From the perspective of a business manager, without the "bloat" and simplifying the business to just key ratios is like driving blindfolded. A buy or sell side analyst can afford to simplify. She can ditch a stock, but a company cannot reverse an investment so easily.

If you get the ratios right, I agree that the complexity makes no difference to value. Surely, how excess returns happen is all about all the things you are trying to do right/well way behind the ratios and you need to start decomposing them to make decisions.

The excel tools you provide are very useful. I still need to do the bulk of the work elsewhere to provide the inputs.

Most of the time, the "bloated" models are not seen by all the managers let alone the outside world as selling tools, but as tools for decision making. For the outside world, lenders/pms/analysts get a hugely simplified CF model. It is up to them to come up with the DCF (frustratingly the D part is frequently misused with just appearances of methods).

Anonymous said...

Dan S "probability weight DCF"

This is a dangerous concept. A weighted single coin flip for $1 head $0 tail is 50 cents which cannot happen in real life. If you do 100k flips, a weighted outcome of $50k may be a reasonable estimate. Probability weight values belong with large number of samples only.

Chris: "I just shake my head in disbelief when analysts tell me that they can model the cash flows for equipment replacements better than the accountant's rough and ready depreciation percentage"
Depreciation is driven by rigid accounting rules for book values and tax rules for tax, both according to classes of assets. They rarely coincide with real asset lives. That is not to say that business managers do not game these rules and replace assets before they run out of live span.

psuedonym said...

I agree with the Professor that EVA is much harder and much more fiddly to use. There are far too many subtle ways to get it wrong than to get it right. The number of times I had gone public with numbers after many corrections only to discover in the middle of a presentation that they were still wrong ... scary. Is it not trademarked to CS/CSFB?