Hammer and Nail: My Issue With DCF

Why I usually don't use DCF, and why I think it's of little practical use.

Sound valuation is the core building block of intelligent investing. If we can roughly understand what an asset is worth, we can profit from price irrationalities that develop as a result of human misjudgment. This idea sounds simple, but I’ve found it to be incredibly difficult in practice. 

The intrinsic value of a company is, as all investors already know, the present value of future streams of cash flow, discounted at an appropriate rate. For some companies, valuations like Net Asset Value, Earnings Power, or Liquidation Value are more suitable – but even for these, DCF is still a component. Even in liquidation, we are still getting cash which we have to discount at an appropriate rate (theoretically). 

This DCF definition is nice and simple (and undoubtedly true for most companies). If we can appropriately project what the future streams of cash will look like and anticipate the timing of these streams, we can determine the range of intrinsic values for a company. 

Since John Burr Williams first explained the concept and formula in his Theory of Investment Value in 1938, DCF has permeated through every part of the financial world – investment banks, hedge funds, corporate M&A, private equity, and so on. People sit down and build out these complex spreadsheets where cost of capital is accurate to the hundredth decimal point and growth rates are derived through complex economic and financial models. Although DCF is undoubtedly accurate (and true) in theory, I think it’s of secondary help in analyzing and valuing businesses for the long term. Here’s why:

1: The Future

As Howard Marks said, “there are few if any facts regarding the future, and the vast majority of our theorizing about the future consists of extrapolating from past patterns and trends.”

DCF, by nature, requires us to make assumptions and judgements about the future. We have to sit and think of what future cash flows will look like five, seven, or even ten years from now. The range of possibilities is already incredibly wide five years out, trying to project ANYTHING (not just cash flows) ten years into the future makes very little sense to me. For someone valuing local newspapers in the 1970s or 80s (and even the 1990s), very few could have anticipated the extent of the internet’s disruption. Local newspapers went from regional monopolies with strong customer captivity, highly recurring revenues, and pricing power to obsolete, inferior products. Those that didn’t manage to transition or retain their market share simply dwindled into bankruptcy (or just became mediocre investments that chipped away at the the purchasing power of its shareholders).

Try this example:

Look at a company you aren’t familiar with and read its last three years of annual reports. You want the company to have 1) a 20+ year history, 2) stable earnings and cash flows, and 3) any qualities that you might want to have in a good company. You can listen to earnings calls, look through investor presentations, and talk to management, but you are only allowed to look at the last three years. Now, with all of that information, try to project what the PAST ten years of cash flows looked like. You can look at past economic and demographic trends – just not at the past cash flows of the company or its competitors (and not at the price history of shares). I can assure you that, almost always, your projections are going to be far off the mark, and may even differ by over 100%. 

So, if we can’t even adequately reverse engineer the company’s PAST streams of cash flows (where you have access to market and economic conditions and a glimpse into three years of the future), what makes us think that we can accurately project the next ten years? 

If even management’s guidance is often off by a lot (think Snapchat), what makes us think that we can accurately predict the future with a high enough certainty?

2: Subject to Change + Compounds Errors

One has to look no further than Peloton or Teladoc to see just how uncertain (and irrational) projections may become. Enjoying a COVID-related demand spike, Peloton produced solid results which pushed analysts to extrapolate such results into the future, without much regard to the underlying fundamentals or the temporary tailwinds that’ll likely simmer down post-COVID. The moment the poor results came in, all the complex models came crumbling down in the face of reality, and analysts then had to go ahead and revise their projections (only to likely be wrong yet again).

Yes, it’s easy to see such mistakes in the rearview mirror, but it’s also a great example of just how shaky projections may be. If you make assumptions that are too aggressive, any mistake or unanticipated event may completely destroy your initial thesis and lead to permanent loss of capital. If we have to constantly update our valuations to match recent developments, what’s the point of even making them (and acting upon them) in the first place?

What’s more, any single error in judgment in your valuation is compounded as you project further into the future. A difference in growth of as little as 2-3% a year can lead to substantial differences in intrinsic value over ten years. Let’s say a company earns $1mm in FCF and has grown at a historic rate of 15%. Now let’s say we have two analysts who want to be conservative and assume lower growth. One assumes an average of 10% and another an average of 13% (yes, projecting 13% growth for 10 years is aggressive, but let’s just assume it’s fine in this case).

 In ten years, Analyst 1 (assuming 10% growth) will end up with FCF of $2.6mm. Analyst 2 will end up with $3.4mm after growth – a difference of over 30%. Let’s say both assume a terminal multiple of 10x, so Analyst 1 values the company at $26m and Analyst 2 values the company at $34mm. Now, let’s say the company is trading at a valuation of $26mm. Analyst 1 will think that the security is fairly priced and moves on while analyst two will say that it is 24% undervalued and buys it. Who’s right and who’s wrong? Is Analyst 2 wrong for assuming growth that’s too aggressive, or is Analyst 1 wrong for being too conservative? Either one can think that they are right, yet both end up with very different values.

This instance can obviously be avoided by demanding a larger margin of safety (say around 50%), but again one may have a MoS of 50% while another only 10%. Value, at the end of the day, is in the eye of the beholder. I, however, don't have that big of a problem with such projections. Sound understanding of the business and stable FCF can more or less help arrive at a rough ballpark valuation (although I’d stick to five years maximum). I have a much larger issue with the terminal growth and discount rates.

Bruce Greenwald provides us with a fantastic example:

Assume the growth rate is 4% and the discount rate is 8%. The terminal value will then be a multiple of 25x FCF. If our growth and discount rate are off by as little as 1%, the multiple can vary from a high of 50x (7% COC and 5% growth) and 16.7x (9% COC and 3% growth). Using low-quality, arbitrary figures like discount rate, cost of capital, and perpetual growth rate to make up the bulk of your valuation will, ultimately, lead to inaccurate valuations and poor decision making. If I ever do choose to use DCF, I almost always stick with multiples.

3: Inherent Biases

Perhaps this one is just me, but I have found that I have inherent behavioral and psychological biases when trying to project FCF. When we are looking at a company, we usually have an end goal of investing in it. So, when we sit down, we are subconsciously pushing ourselves to make projections that will allow us to invest with a large enough margin of safety. 

Especially when I really like the company, I find myself making assumptions that I know aren’t conservative enough (and that I wouldn’t be making if I didn't like the company) – and yet I make them anyway. Think about it. If you project strong growth and a higher exit multiple because you like the company, the final valuation is very often going to be significantly higher than current valuations. You might think you’ve found a gem, but instead you have just tricked yourself into buying the company at current valuations. 

We only have to look at the countless manias throughout history to see how quickly and easily people go about justifying valuations. I saw one analyst (I don’t remember who exactly) highlight that Tesla is undervalued by assuming 20% growth and a 40x multiple. It’s very easy to engineer models to fit whatever narrative you want to pursue. You might even be doing it yourself (subconsciously) without realizing it. 

Final Thoughts

Truly successful long-term investing, in my opinion, comes from superior understanding of the business, its management, and its sustainable competitive advantages. At the end of the day, long-term performance is driven by the performance of the underlying business. You can make all the projections you want and build out many models, but nothing can replace genuine, superior understanding of the economics of the business. If you find something you genuinely like, it’s not that hard to come up with an appropriate price you’re willing to pay. 

When I read through annual reports, presentations, and earnings call transcripts, I usually already have a range of values in my head that I think would be appropriate for the business. Then, I go ahead and look at the price. Sometimes, I’ll be pleasantly surprised to find that the shares are trading significantly lower because of some short-term issues. Usually, I’ll find that it’s fairly priced or severely overvalued, in which case I just move on. It’s not that hard to think of an appropriate price to pay.

For Kaspi.kz, for example, I’ve never tried to project earnings or FCF. Why? Because I didn’t have to. It was so obvious that the company was drastically mispriced given both growth and quality. Although some may see this as a less accurate approach — I’d argue that it is far more reliable than plugging everything into spreadsheets. Instead of building models, I had to compensate for the uncertainty in value by focusing on the business. It forced me to evaluate management, competitive advantages, and underlying economics a lot more carefully, and in the process gave me a much better understanding of the business and what I think it’s worth.

Superior valuation ultimately stems from superior understanding of the business. I couldn’t agree more with Howard Marks’s concept of “second-level thinking.” Many focus too much on valuations, thinking that a large enough MOS compensates for their ignorance towards understanding the underlying business. They end up sticking everything they can into DCF and basing a large portion of their thesis on these low-quality assumptions. As Abraham Maslow said, “to a man who only has a hammer, everything he encounters begins to look like a nail.”

To me, nothing will replace a solid grasp of the underlying business, no matter how good one is at making these projections.