DCF – Valuation in the land of fairies & pixie dust

DCF is the valuation method most frequently taught in the classroom and it is considered the “most academic” of the valuation tools. DCF also happens to be the most misleading and falsely accurate methodology when it comes to company valuation.

BUT, before you go guns blazing into your interview castigating DCFs, a) make sure it’s not your interviewer’s favorite valuation tool, and b) make sure to understand fully the merits and demerits of a DCF. To be clear, you definitely need to know how to calculate a DCF. You should know the CAPM, BETA, and WACC framework inside out. And you should be prepared to answer questions in interviews. There is plenty of freely available academic material on this method, e.g. from NYU’s Stern School of Business http://bit.ly/XJdxII. So I won’t go into definitions here.

Instead, let me highlight the pitfalls of DCFs in company valuation.

  1. DCF valuation calls for multi-year “projection of future free-cash-flows.” When valuing a company, this usually means building a projection model and estimating cash flows out 5-10 years. Great in theory, but anyone who’s tried to project a company’s cash flows, even with access to internal company data knows that looking out beyond 12-18 months is pure fantasy. As such, the first premise of a DCF – using out-year projections – infuses significant model error into the equation. Of course, for less complex assets like bonds it is quite possible to predict cash flows many years into the future.
  2. A “perpetuity model” is often used to calculate a company’s “terminal value” – i.e. the company value at the end of the projection period. In the perpetuity model, the analyst takes the last FCF in the projection period, then guesses (sorry “estimates”) growth rates and required returns for the company into perpetuity. A final-year value is thereby derived, valuing the company as if it were a perpetual bond. If it’s tough to look out further than 12-18 months, even for a company management, how much weight would you ascribe to a first-year analyst’s guess at the terminal cash flow 10 years in the future? And how accurate do you think an estimate of “perpetuity growth”  starting in 10 years might be? By the way, these terminal values often make up 50-80% of the total DCF value. So the lion’s share of a valuation is based on nothing more than wild guesses.
  3. An “exit-multiple” can alternatively be used to calculate terminal value. The idea is that the company is sold in the last period. This sale is usually assumed at some EV / EBITDA multiple (often held constant at the company’s current trading multiple). So year-10 projected EBITDA is multiplied by the exit multiple. The hypothetical cash received from that sale is then equal to the terminal value. So this method essentially bases 50-80% of the company value on a guesstimate of year-10 EBITDA. And more importantly, if we’re going to value up to 80% of the company on a multiple basis, why not forget the DCF & the 10-year gaze into the crystal ball to determine terminal value? Why don’t we just use a one-year forward multiple to value the entire company – on EBITDA that we have a fighting chance of estimating within an acceptable band of accuracy?
  4. The weighted average cost of capital (“WACC”) is used to discount each of the company cash-flows back to the present day. The idea is that some rate of return exists at which an “informed investor” would be just as happy to invest in the company as he would be to invest in some alternative investment – and the rates for such alternative investments are calculable. Determining that required return then involves: 
    • Assuming an expected market return. Easy right? Just use the S&P 500…but returns have differed greatly over time. Which time period is relevant, the last 5-years, 10-years, or 100-years? Because that will be the basis for the discount rate you use to value your company.
    • Determining the risk free rate. No problem, government bond rates. But hold on one second. If the US was risk-free at $900 billion in government debt before the financial crisis, is it now less risky at over $3 trillion? Because 30-year bond rates – now about 40% lower than pre-crisis – suggest it is. And the market’s always right. Right?
    • Choosing a BETA to determine the company-specific risk (or required return) of the company you’re valuing. OK, so we can just look at relative historical performance of similar companies and apply that BETA to the one we’re valuing. Really? What kind of historical performance measures were Fannie Mae and Freddie Mac using to value US housing in 2005/2006? How were tech analysts calculating BETAs pre-internet bubble, say in MAR 2000. Or what cost-of-equity assumptions was Wall Street using to value  banking stocks pre-crash in mid-2007? Sadly, risk isn’t measurable in this way…and the carnage of recent crises underscores the point.

So does DCF valuation ever work?

Indeed. It works like a charm when valuing bonds (only those with limited default risk). When an asset’s cash-flows, its repayment period, the repayment amount and its relative riskiness versus other assets are all well known (say for government bonds vs. AAA corporate bonds), a DCF actually does a fine job of estimating a fair NPV. Certain projects with “bond-like” cash-flows may also be candidates for DCF, but the further an asset’s FCF profile diverges from a bond’s, the less useful (and falsely accurate) a DCF is going to be.

Leave A Reply (2 comments so far)

The comments are closed.

  1. Bob Wilkerson
    4 years ago

    Yes, the world has changed. Twenty years ago, you could still predict fairly well the future cash flows from a Blue Chip company and have a pretty good idea that their performance would be steady into the future. They had long steady histories of performance from which to set a valid Beta. Barriers to entry into their industries were formidable, allowing them to keep their market share and security. Anything unusual causing a blip in their valuation was well documented in notes. And it took generations to build a company of that size. Today you have Billion dollar companies appearing and disappearing in a decade. Technology and new ideas for its use can create a Billion dollar competitor which can undermine a Blue Chip and take its market quickly. Social media spreads new ideas and products quickly, allowing good products to survive the battles from the established until the next new Billion dollar competitor replaces it. Analysts, to use any valuation tool, need to pay serious attention to the industry being looked at, the profitability that might drive competitors to join it, and the current research efforts which could change it.

  2. Jesse Freitag-Akselrod
    4 years ago

    Thanks very much for your comment Bob. I’ll agree that the world has become more complex…and consequently even harder to predict.

    I’m not so sure that using past data to extrapolate future risk or returns was ever a good idea though. Neither now nor back in the day. Market-covariance analyses like those used to “measure” risk in the CAPM have the nasty habit of SEEMING to be good predictors…until they aren’t. Reality masquerades as smooth and predictable, but only for so long before you have a blow-up or “extraordinary” event. The truth is that crises and blow-ups are anything but extraordinary. They are commonplace and have been throughout history.

    Unfortunately, using past data usually ignores (or massively understates) the impact of crises, overlooks hidden factors and feedback loops and understates the “cost of equity.” In turn, assets are often wildly overpriced (from Tulips to Internet companies to real estate) and blow-ups are exacerbated by our use of falsely optimistic valuations. And, the longer we keep the party going, the worse the blow-ups end up being.