Why the DCF method sucks

And closing my Grupo Catalana position

I don’t like the discounted cash flow (DCF) valuation method, especially when a company is growing. I don’t know many investors who actually use it and it looks like something that some ivory tower academic came up with in an afternoon. If I had to list its flaws I would start with:

• The fact that a DCF does not really count cash flows if the company is growing. Say a company with a 10% return on invested capital (ROIC) grows 10% a year reinvesting all its earnings. This means the company generates 0 free cash flow. So do I count net profit instead? Or just count fake cash flows by normalising for working capital movements and only subtract maintenance capex? Should we really call it an imaginary DCF model? Might as well call it a discounted net profit model then?

• ROIC is not included in a DCF calculation. Say you have company A that generates a 10% ROIC and company B that generates 100%. Company B should clearly be valued higher if both were growing 10%/year. As company B can pay out 90% of its earnings while company A 0%.

• It pays no attention to what happens to the FCF that is generated. A bond does pay out cash, which is very different from cash that is generated and not paid out by a company. The DCF method does not distinguish between this. I suppose you give the cash hoarder some arbitrary higher discount rate?

• The concept of “cost of capital” is ridiculous when applied to equity. Unless it is a preferred equity. Munger and Buffett trash this idea more elegantly than I could. Cost of capital in an equity DCF was the NFT of finance before NFTs were a thing. Most people don’t really understand it, so they just go along with it because they are afraid to look stupid.

• Assessing risk using a discount rate is counter intuitive and opaque. If the discount rate is increased by a small amount you get a wildly different number without really knowing what that exactly implies. Using probability trees and making things like dividend/buyback payout, holding period, expected return and ROIC explicit would mean there is more to argue with than some vague hard to interpret discount rate. Which I suppose would actually be a bad thing if you are an empire building CEO with little skin in the game.

• Terminal value multiple. I thought we were counting cash flows? And now all of a sudden we are using multiples again? What is this nonsense? Why not skip counting those imaginary cash flows and just straight up use multiples instead?

I am actually surprised that Warren Buffett has not railed against it harder given all its flaws. 

So what is a better approach? Well there are three states a company can be in: 

• Rapid growth very close or equal to ROIC.

• Steady state where growth is expected to be significantly below ROIC and ROIC possibly reverts to the mean of ~10-15% and a significant portion of returns have to come from dividends/buybacks

• Shrinking 

When a company is shrinking and is going to liquidate, a DCF is not a bad way to value it. Especially if the company returns all FCF and is essentially in liquidation mode.

But for the other two we are going to need something better than a DCF model or “anything below a 10x PE is cheap” which many value investors seem to use. I will get into a few simple formulas and then give an overview of my approach. There will not be any advanced maths, as a good valuation method has to be simple and intuitive. 

Compounder method: For when a company is growing rapidly

Valuing a growing company with a very high ROIC is difficult as the range of outcomes is very wide. Buffett has famously said his holding period is forever. And if your holding period is forever you can pay near anything for a good company, as long as you believe it has a really durable moat. 

But Berkshire is an institution that lives on when Buffett dies. I am a person and I'm going to sell my shares and spend a good portion of that money at some point. So my holding period is at best 30-40 years and probably shorter than that. And I don’t feel comfortable predicting rapid growth very far out.

If it is certain a 20% ROIC compounder will keep reinvesting all its FCF at 20%/year for the next few decades then all you really need to do is figure out: 

• ROIC (which we assume to equal eps growth)

• Holding period (how long you think this growth rate persists)

• Exit multiple at which you can reasonably expect to sell as earnings growth stalls out somewhere near GDP growth and the company is unable to reinvest all its cash flows

• Minimum expected return over the holding period

The formula for the maximum amount that you should pay to get at least your minimum expected return then looks as follows:

\(buyingMultiple = (epsGrowth^y \times exitMultiple) \div minReturn^y\)

For example when absolutely certain of at least 20% growth, to get a minimum 15% CAGR over a period of 10 years with a 15x exit earnings multiple (assuming growth levels off after 10 years) you should pay no more than 23x earnings:

\(23 = (1.2^{10} \times 15) \div 1.15^{10}\)

This simple formula is actually a great way to get an intuition of what is priced in. It can be combined with base rates and by looking at TAM to get at a quick rough idea what assumptions are baked into a growth stock. For example if a 20% ROIC compounder holds 10% market share already in a market that is growing 7% a year, then in 20 years at 20% growth (assuming stable margins) they would hold ~100% market share. If it really keeps compounding forever at that rate, its revenue will be greater than global GDP at some point.

Another example, say your compounder will grow 25% for a decade, but you are satisfied with a 15% return already, and the exit multiple is actually 20x:

\(46 = (1.25^{10 }\times 20) \div 1.15^{10}\)

Going past 50x earnings we have to start making pretty aggressive assumptions over an already fairly long period of 10 years to only get 5% above the S&P 500 long term return.

Exit multiple 

So now we are at the low growth stage and strongly considering selling. Or we need an exit multiple to plug into our above formula. Company is only going to grow a few % a year and dividends/buybacks will have to do some of the heavy lifting. Long term nominal S&P 500 eps growth is about 7-8%. On nominal GDP growth that was only 6%. To be conservative I wouldn’t go above about 5-6%. Because population growth is going to be lower, and globalisation seems to be reversing (this really boosted S&P 500 returns over the past 50-60 years). And some home grown companies might push out American companies (like Apple to Xiaomi).

The following formula is useful to determine at what earnings multiple we need to be able to reinvest dividends (or buybacks):

\(exitMultiple = 1 \div (expectedReturn - epsGrowth) \times payoutPercentage\)

Where maximum payoutPercentage is estimated by doing the following (assuming zero eps growth from pricing power):

\((ROIC - epsGrowth) \div ROIC\)

Let’s use Pepsico as an example. Let’s say earnings are expected to grow 6% in the coming foreseeable forever, payout % was approx 75% in 2023 with ROIC is north of 20%, so they are pretty shareholder friendly. Very predictable demand and margins, you could say it is very likely it will show steady growth in a mature market. Using this formula our exit multiple is:

\(18.75 = 1 \div (0.10 - 0.06) \times 0.75\)

Which is close to its current 21.3x NTM PE (analysts forecast near term growth to be a bit higher than 6%).

A 10% ROIC company that grows 5% over time and pays out the rest should be valued at only 10x earnings by someone demanding a 10% return. In fact if ROIC = expected return, growth rate doesn’t matter (if the company is not hoarding cash that is). 

This formula starts to break when growth is very close to expected return. Or when growth is significantly above nominal GDP growth. Expecting a 10% return of a company with a 100% ROIC with 9.5% growth and 0.5% coming from dividends/buybacks, the formula says you can pay up to 181x earnings. Of course if growth disappoints and only comes in at 8% then that plummets to 46x earnings. Adjusting your return expectations to 12% instead of 10%, with the same 9.5% growth, lowers it to 36x earnings.

That doesn’t mean the formula is wrong, if you were able to be 100% certain the above numbers hold up for a very long time to come and your holding period is near forever then paying over a 100x earnings would be rational. It is just that you can’t be 100% certain. Since that would imply the company would surpass global GDP at some point. So it's best to just plug in conservative growth numbers at or below nominal GDP growth or use probability trees.

The formula does work well to quickly illustrate why low quality companies deserve to trade at low multiples. For example if you want at least a 10% return buying a 6% ROIC stock growing 3%/year then don’t pay more than 7x earnings. If ROIC is below expected return, then the multiple starts to plummet to 0 if the company reinvests more of its cash back into the business for a long time to come. 

Pricing power or “free growth”

There is one element still missing and that is pricing power. Which is essentially free growth. For example if costs go up and the company raises revenue by the same amount that is basically free growth. But if it were to expand beyond keeping wallet share stable, that growth is limited by its ROIC. 

ROIC is also not a fixed thing. If a company scales up, margins might expand without raising its prices as the company gains economies of scale.

Some businesses have untapped pricing power with Buffett’s See’s candies acquisition being a famous example. I think Buffett raised prices 10%/year for more than a decade, so well above inflation essentially increasing wallet share. Tobacco companies being another famous example. 

You can adjust the max payout formula like this to take pricing power into account:

\((ROIC - epsGrowth + free Growth) \div ROIC\)

How I do it

This is an overview of most of this blog’s current active ideas and how I keep track of my investments (click to enlarge):

(If you want to copy paste it for your own use here is the google sheet link. I am sure excel veterans are cringing at the layout though)

The 2024 and 2026 Low % means how much upside there is in a base case if the company goes into low/no growth mode after 2024 or 2026. The LTM PE range is essentially a range of exit multiples that I think the market is likely to assign in a low/no growth scenario. 

I adjust these numbers up by 1 year each year. So for example for TK group I assume HK$0.34 of EPS in 2024. If earnings level out after this point and the stock trades at 7x earnings, there is 57% upside. If they keep growing to HK$0.45 in 2026 and then level out and the stock trades at 7x earnings there is 107% upside. Plus another 10-30% upside in dividends. 

Why only 7x earnings? Assuming that average market participants needs a higher return of 12% to invest in a controlled Chinese stock, a suboptimal 60% payout, 4% terminal growth then I should pay no more than:

\(7.5 = 1 \div (0.12 - 0.04) \times 0.6\)

I also look at the historical PE ratio, and this stock has only traded above 10x earnings when it was rapidly growing. And I value their net cash position at a low amount, so clearly hitting that 7x earnings multiple with an EV of almost 0 shouldn’t be too difficult.

If I see evidence they will not hit my earning target, all I have to do is lower the EPS estimate. And when I see too many red/light green squares I start selling.

This is why I sold Grupo Catalana Occidente recently. And will close the position here. So far the stock has generated a 30% pretax return and 7% outperformance since mentioning. I think they will not grow faster than 5-6% and forward dividend payout estimates have recently been lowered to a meagre 25% of earnings. I was kind of hoping for the opposite to happen since they have over €1 billion in net cash. What also worries me a bit is that EBIT margins are near historical highs.

Since this is a majority controlled family company there is little chance an activist will be able to do anything here. I really struggle to see how this stock will be valued above 10x earnings for sustained periods of time until they become more capital efficient. Currently it is valued at nearly 8x earnings which I think leaves too little upside. If they do become more capital efficient (paying out >50% of earnings) the stock could rerate to 12x earnings: 

\(12 = 1 \div (0.1 - 0.05) \times 0.6\)

But that looks less and less likely in the near term.

Conclusion

So my method could be summed up as:

1. Find out which growth state the company is in. How much cash is the company paying out if not growing much? Is this likely to change?

2. Only focus on above average companies (ROIC of 15-20%) that are stable or growing, unless it is some kind of special situation.

3. I make my eps estimates no more than 2-3 years out and take a exit multiple by using a combination of the exit multiple formula and looking at where the stock has historically traded. 

4. I plug it in my google spreadsheet watch list so I can compare in an instant short and medium term upside and see how much I am getting paid to wait. I also have a similar watch list of ~100 stocks (they have mostly light green and red squares though). The art here is to not be too conservative in your exit multiple, but also not too delusional. If you pick the right multiple it allows for regular trading around different stocks as they get more expensive/cheaper. While still getting a good return. 

5. There is a qualitative element too. For example the chances of a buyout, possibility of special dividends from net cash positions. Recurring earnings streams get a lower required return (so higher exit multiple). Large customer concentration means lower multiple and historical PE range will be more strongly considered.

6. Some companies will not grow much, but will pay large dividends and can produce a lot of upside just rerating to a modest multiple (Halyk Bank). And some stocks are not necessarily that cheap on near term numbers but carry limited downside and significant optionality in the medium term as they scale up and possibly become very profitable with rapidly rising ROIC (Dada Nexus). 

7. I start selling as all the squares get light green or red, and replace that stock with another one from my watch list (or add to one of my other positions) that has more dark green squares. It is important to not be anchored if the stock has moved up already. Sometimes a stock is a stronger buy after it has moved up 20-30% already. So it is best to not display share price or % return since adding. 

So there you have it. Down with the DCF tyrant. I hope it was helpful. I learned a few things myself writing this all out actually.