One of the biggest changes in my investing process over the past year has been my increasing usage of discounted cash flow (DCF) models. I used to rarely use them and now I build DCF models for every company I’m serious about. I want to start by reviewing the major things that a lot of investors (including myself) don’t like about DCFs and how I’ve gotten over these concerns.
Too many assumptions
The number of assumptions one has to make to build a DCF model is very large. It’s impossible for me to accurately make fifty different assumptions that determine a company’s future. Doesn’t this mean the final output of the model will be inaccurate and useless? The thing that took me awhile to realize is that slapping a multiple on earnings or cash flow involves the exact same assumptions. The only difference is valuing companies based on multiples makes those assumptions implicit versus a DCF model that forces explicit assumptions. You may not realize it, but slapping a 15x multiple on a stock is absolutely making assumptions about the company’s future growth rate, margin compression or expansion, return on invested capital, debt-to-equity ratios and all the other explicit assumptions in a DCF.
Beta and cost of equity
The traditional way to calculate cost of equity under the capital asset pricing model (CAPM) uses beta, which is probably the dumbest financial concept to ever reach as much mindshare as it has. I used to disregard DCFs because in my eyes they were based on this flawed logic. But cost of equity and weighted average cost of capital (WACC) are extremely important to valuation. A company that earns returns above its cost of capital creates value; a company that earns returns below its cost of capital destroys value. The key for me was to come up with a cost of equity calculation that made sense to me. All cost of equity is trying to calculate is the minimum return an equity investor should demand, which includes beating inflation and getting a premium for being the most junior piece in the capital structure. Dale Wettlaufer at Charlotte Lane Capital had a great write-up last November on how he calculates cost of equity and WACC. My calculation differs only slightly from his. I have a lot more thoughts of WACC, but I’m going to save that for a future post.
Terminal growth rates
Most DCFs have two major components: a period of years (usually 5-15) that are forecasted explicitly and then a terminal value that represents the value of the company thereafter (in perpetuity). Estimating what a company is worth in perpetuity ten years from now is an impossible task and I honestly don’t love any of the popular methods. However, the terminal value often accounts for over 50% of a company’s valuation so it’s very important. I think the best compromise is to use several different methods (with different growth rates) and average them together. Growth rates should approximate expected future inflation and/or expected GDP growth—something in the 3-4% range seems reasonable to me. In the long-term it’s impossible for a company to grow faster than the overall economy.
Buffett doesn’t use them
This was almost certainly part of my aversion to DCFs early in my investing career. While I learned a ton from Buffett when I was starting out, I also came to realize that a lot of what he talks about does not apply to what I’m doing. I also think Buffett contradicts himself a bit when commenting on discounted cash flows:
“To value something, you simply have to take its free cash flows from now until kingdom come and then discount them back to the present using an appropriate discount rate.”
No one is going to disagree with that and that’s precisely what a DCF does, yet he’s against trying to actually calculate those future cash flows:
“If you need to use a computer or calculator to figure it out, you shouldn’t [buy the investment].”
He doesn’t need a computer or calculator because he’s a genius:
“Every time we start talking about this, Charlie reminds me that I’ve never prepared a spreadsheet, but I do in my mind.”
Well, I can’t prepare spreadsheets in my mind, so I’ll stick to Excel.
What changed my mind
If you ask value investors what a company’s fair value is, the vast majority will say something similar to the first Buffett quote above: a company is worth its future free cash flow discounted back to today. Then those same investors (for a long time I was included in this) hate on the one valuation method that explicitly measures future free cash flows. So everyone agrees what a company’s value is based on, but no one values companies based on that. When I first realized that disconnect, it was a bit of an “a-ha” moment for me. This report from Credit Suisse, What Does a Price-Earnings Multiple Mean?, is a great read that compares DCFs to multiples when valuing companies.
I was reading a rather convincing write-up on Value Investors Club recently that used a DCF to value the company. The DCF spit out a very nice upside potential at the end, but I thought the WACC he used seemed low. I plugged in what I thought was a more reasonable WACC and the new per share value was lower than the current share price! I’m not saying I’m right and he’s wrong, my point is that you can make these models output anything you want with (seemingly) small changes. This is one of the major problems with DCFs.
I’ve come to really like DCFs because it forces me to think about every aspect of a business explicitly. When I’m calculating cost of goods ten years into the future, I’m forced to think about the company’s historical cost of goods and how and why it might change in the future. Of course a DCF isn’t required to go through this thought process, but it does ensure I think about every single line item in the financial statements. Even when using DCFs, it’s crucial to remember the limitations and that the basics are still what matter most—competitive advantages, growth opportunity, shareholder friendly management—and these things can’t be described with numbers. And when in doubt, be conservative by using a high discount rate and a low terminal growth rate.
5 thoughts on “My Love-Hate Relationship with DCFs”
Have been following your blog for awhile, and the Xpel stuff was great. My two cents on DCF:
They’re useful if the assumptions put in are conservative. If you start using assumptions that chances are won’t happen or are a coin flip, well obviously you might be in trouble. People go for 20, 30 years etc. Unless we’re talking about a DCF on KO in ’88, that was assured many, many more years of nice profits I tend to stay away from long term forecasting and look for a bulk of the value up front.
As far as using WACC, I’ve always had a much simpler way of looking at it. The DCF tells you how much you should pay for something to achieve the return your looking for. That return is represented by the discount rate. There is no “absolute value”. If someone is looking for a 6% return, he obviously would be able to pay a much higher price than someone trying to hit 12%. So I’m kind of at a loss why someone either wouldn’t use their opportunity cost rate or their minimum return hurdle rate.
As far as Buffet not using it, well, he kind of does use it. He just doesn’t need to sit down and write it out. I’m not as smart, so I do. I’m not saying he actually does the whole calculation in his head, but I’m pretty sure he’s never bought anything that wouldn’t look very good on a conservative DCF.
I agree the terminal portion throws me off. I like the idea of using an average that you mentioned.
Appreciate the blog and wishing you much continued success.
Thanks for the comment. I don’t think the WACC should be used as the discount rate either. The discount rate, in my opinion, should be each investor’s personal IRR hurdle rate. This is exactly what I was going to discuss in a future WACC blog post. I agree with your Buffett comment.
Interesting read here. I like to avoid using spreadsheets and the like because it forces me to just memorize the applicable figures of the company. Memorizing the figures allows you to be swifter when market developments occur.
For most of my investments I try to avoid asking the hard questions. I instead project out earnings and see how many years it takes to add up to 5x the current market price (20% of the current market price). So if a company is growing at 30% per year and the stock is trading for 45x earnings. The payback period is approximately 4.5 years.
I usually look at an EV/FCF forward payback to incorporate cash and excess assets i.e. if a company owns 300k acres of timberland that can/will be sold for $2000 an acre with no debt and no excess cash. The EV of the company is $1,5b and the FCF is $100m and the fcf will grow at 10% per year: then the payback period is less than two years.
I have heard that Selim Bassoul uses a similar methodology to analyze deals in that he makes sure he only buys companies that after 3-5 years are the equivalent to being purchased for 5x earnings or less.
You have a better memory than I do 🙂 That’s an interesting valuation method, hadn’t seen that before. One thing I love about investing is how many different ways there are to be successful. Just have to figure out what works for you.
Interesting discussion. Buffett is usually pretty vague whenever he’s asked a question about valuation… I think part of the reason is that there is no one correct way to value a business and it is always dependent on the circumstances at the time one is making the investment. I do think that one can get pretty accurate with regards to making future cash flow projections (whether one simply decides to project earnings out a few years or do a full DCF). What matters ultimately is how one incorporates all the available information to make the forecast.. similar to making a probabilistic estimate… a good forecaster can get pretty precise and come up with a 63% chance of an event occurring in the future… and be right. On the other hand, someone that is not as granular might come up with a 70% chance of something occurring and also be right… ultimately in investing what matters is what factors you focus on and how much weight do you give each factor … I don’t think there is necessarily one better way to do it
Comments are closed.