Bruce Greenwald on The Folly of Discounted Cash Flow

The way you’re normally taught to value a company in modern finance is to project out some cash flows a few years, come up with a terminal value for the remaining cash flows on out to infinity, and then discount them all back to the present day. It’s discounted cash flow, and it takes the following form:


But to paraphrase my root guru…

If you're using DCF for equity valuation, you're wasting everyone's time.
If you’re using DCF for equity valuation, you’re wasting everyone’s time!

There are at least 4 reasons you don’t want to do this.

(1) DCF is a summation of cash flows in the near term (you might guess these correctly) with others far out in the future (blind guesses). When you mix good information with bad information, you get bad information.

(2) You might think you’re predicting a single number when predicting a cash flow, but it is a margin on sales, and margins are a function of the general level of economic activity, labor costs, raw inputs costs, competitive forces, etc. You have to correctly predict the interplay of multiple variables!

(3) DCF ignores the balance sheet.

(4) In the summation, most of the value is in the terminal, which is a 1 / (r – g), which is enormously sensitive to small changes in assumptions.


That last item is really what makes DCF unusable. Let’s look at an example. You’re valuing a company with $1 in FCF. And you think it’s going to grow by 10% over the next 5 years, then drop to 5% growth forever. A middle of the road cost of capital is 10%, so you use that as the discount rate.

Fair Value = $26 when r=10% g=5%
Fair Value = $26 when r=10% g=5%

Fair value came out to be $26. But if we change the discount and growth rates assumptions by just 1%, fair value drops 25% to become $19!

Fair Value = $19 when r=11% and g=4%
Fair Value = $19 when r=11% and g=4%

Or, if we do the tiny 1% moves in the opposite directions, fair value increases 60% to become $42!

Fair Value = $42 when r=9% and g=6%
Fair Value = $42 when r=9% and g=6%

Would you trust a valuation model that can only get you to within an X to 2X price range? Are you confident you can nail r and g to within 1% so that the range isn’t even wider?!

Luckily there is another way… (to be continued)