In my 2020 Investor letter, I discussed the concept of “Margin of Safety” i.e., paying less for a security than it is worth, and its implications for investing. This memo is intended as a follow-up to that discussion.
Q4 2020 hedge fund letters, conferences and more
As a recap, I argued that whilst the concept of margin of safety itself is sound and important, the way it has historically been implemented by investors has led them astray. They gave too little thought to the future, based on the assumption that an investment would take care of itself if the margin of safety were great enough. I argued that developing as accurate a picture of the future as possible is the single most important task for an investor. Furthermore, being conservative is not a substitute for striving for accuracy. Underestimating a company’s long-run cash flows can be just as damaging to performance as overestimating them.
Part 1: Is it not Obvious?
As I look back on the letter, it struck me that the importance I place on striving for an accurate picture of the future might seem so obvious as to be hardly worth mentioning. After all, if a company’s intrinsic value is the sum of discounted future cash flow, why on earth would you not want to form as accurate a view of the future as possible? I can imagine my insistence on this point is particularly puzzling to younger investors whose formative years have been dominated by the boom in Internet stocks. “Doh!” they might exhort, “Of course you have to skate to where the puck is going, not where it was.”
The reason it is not obvious to older generations of value investors is that in our formative years, investing based on the assumption that historical patterns of cashflow generation would reassert themselves – better known as “reversion to the mean” – seemed the better strategy. Many of the great investing track records were built by investing in stable, unchanging businesses when they went through a period of underperformance on the assumption that they would eventually recover. It was an approach to investing that was based on a good understanding of a company’s history and the assumption that the future would not look too different to the past. It worked far better than betting on companies with short histories and big plans for the future, and it seemed obvious that it would continue to.
Part 2: Is Experience All It Is Cracked Up to Be?
These two contrasting approaches to investing – one placing more weight on the future; the other on the past - are a reminder that the optimal strategy is a function of the era you invest in. If you are in a market characterised by rapid and widespread change, it pays to be forward-looking despite the inherent difficulty of judging the future. If, on the other hand, you are in a market where the pace of change is slower and more localised, then it may simply be better to bet on reversion to the mean as the future is too uncertain and genuine change too infrequent.
The contrasting outcomes of different brands of value investing in different eras pose an intriguing question. If each era selects for the type of investor who is best adapted to it, does the younger investor have an edge over the older one? My strong sense is “yes”. I am fortunate to know several successful younger investors, and they seem perfectly adapted to the market they invest in. I, by contrast, have had to adapt, which in practice does not so much mean learning new tricks as unlearning old ones. The former is certainly easier than the latter as learning is fun, but parting ways with cherished ideas is painful. Reluctant though I am to acknowledge it, as grey hairs begin to colonise my scalp, experience is a disadvantage.
In one important respect though, there is an advantage to experience. When the nature of the market does change, it should, at least in theory, be easier for the investor that has lived through different types of market to adapt than for the investor who has only experienced one type. The younger investor suddenly finds themselves in the position of the older investor without the benefit of having experienced a change in the market before.
In practice, however, few investors have sustained multidecade success. This may not solely be down to how difficult it is. It could also be that the rewards to the stellar performer are so great in one era that they lose interest in competing in the next one when they realise that their skills are no longer as finely attuned to the market. For sure though, it is a monumental challenge.
To increase the chances of adapting to different markets, I see one big thing an investor should do and one big thing they should not. The single biggest thing they should do is commit to adapt. The single biggest thing an investor should not do is tie themselves to a particular investment style or geography or industry or any other categorisation. These two points may sound obvious but generally, fund managers do the complete opposite. Investors in funds tend to look for a specific niche expertise in fund managers, and fund managers respond to this by developing a personal brand for a particular style of investing or segment of the market. Their brand promise is that they will not adapt.
In my view, the tight coupling of brand and type of market is the single biggest reason why storied value investment companies have done so poorly more recently. They felt duty-bound to keep on banging out the same old hits, as after all, was this not what their fans demanded?
It is noticeable how Warren Buffett, the single biggest exception, took the opposite route to most value investment companies. He repeatedly told Berkshire Hathaway shareholders that as the capital base grew, Berkshire would have to adapt. He committed to adapt and did. Even he, though, made the unforced error of forsaking investments in technology. That he has since corrected it, is a perfect illustration of his adaptability.
Part 3: Has the Rate of Change Really Accelerated?
One pushback I received to the letter was that the pace of change is likely no faster today than in the past, i.e., the market is not all that different. Steam power, electrification, television, the automobile, and containerisation are just a handful of innovations from past eras that fundamentally changed the economy of their day. The invention of the automobile, for example, not only created a large, new sector of the economy but led to people moving out of cities into the suburbs and enabled big box retail, amongst many other things. In other words, it changed the overall economy, not just transportation. Is it so different to the Internet?
My best guess is that the pace of change has accelerated. It has been driven by the global nature of demand for digital goods and the ability of companies to serve this demand with minimal incremental cost or capital. In combination with the winner-take-all nature of many of these markets for digital goods, it has allowed companies to grow faster and larger with greater predictability than in the past. The failure of investing strategies based on reversion to the mean is consistent with this analysis. However, determining whether the pace of change really has accelerated is ultimately a task for historians, not investors. Investing, as an activity, is lived forwards not backwards (even though people enjoy explaining outcomes after the fact).
The point of my letter was to suggest that the optimal investment approach is a function of the opportunity set. If predictably innovative companies were less prevalent than generally thought and were overpriced as a class, then the correct investment strategy would be to avoid them – the odds are simply stacked against you. If, on the other hand, predictably unchanging companies were less prevalent than generally thought and were overpriced as a class, then the correct investment strategy would instead be to avoid them. The job of the investor, looking forward, is to decide which is likely to hold true and place their bets accordingly.
All that can be said with certainty was that over the last ten years the Internet changed virtually every industry, and supposedly unchanging businesses as a class were rarer than thought and accordingly overvalued. As such, the idea that it is always a mistake to bet on growth, as Graham suggested in “The Intelligent Investor”, is wrong, plain and simple. Whether it is best to bet growth or reversion to the mean is a question that continually needs to be posed afresh.
Part 4: Is Accuracy Too High a Bar?
Another pushback to the letter was that striving to be accurate places an unduly high burden on the investor given that the future can only be forecast in the broadest strokes. My good friend, Rishi Gosalia, pointed out to me that he has occasionally found a company that could be worth many times what it is worth today but was unable to be more specific than this. If he had been forced to predict precisely what its long-term cashflow might be, he would have most likely had to pass on the idea.
When I heard this, I realised with horror that if my discussion of accuracy meant investors missed out on wonderful opportunities, it would inadvertently be guilty of the same crime I accused “Margin of Safety” of – correct in theory but leading investors astray in practice.
This need not be the case though. I would emphasise that my admonition was to be “more accurate”, not “accurate”. Clearly, it is a fool’s errand to think that the future can be accurately predicted. It is not a fool’s errand though to attempt to be more accurate than everyone else.
As a thought experiment, consider three different approaches to Rishi’s investment opportunity.
The “disciplined” approach is to assume no growth in cash flow given a conviction that rapid growth rarely materialises in practice.
The “conservative” approach is to assume a short burst of extraordinary growth followed by normalisation given a conviction that rapid growth rarely sustains for long.
The “visionary” approach is to assume prolonged growth given a conviction that a company has a large total addressable market and growing competitive advantages.
As imprecise as the visionary approach is, it is easy to imagine a scenario that it is more precise than the other two. In fact, an insight that a company is worth many times what it is worth today is plenty precise for me and, I suspect, for most people.
Notice how the adjectives “disciplined” and “conservative” are generally considered positive attributes in an investor whereas the perception of “visionary” is more nuanced. I disagree with this assessment and often find that the adjectives “disciplined” and “conservative” are used to hide a lack of thought. This is most clear in the “disciplined” investor, who gives no thought to the future whatsoever, but the “conservative investor” is just a less extreme incarnation of the same phenomenon. Forecasting only part of a company’s long-run development is only slightly better than forecasting none of it all.
Perhaps I should add that whilst it is desirable to be as precise as possible, one should not try to be too precise. There are certainly no benefits to false precision. It creates undue confidence in an investor’s ability to forecast the future, which can lead to hubris. It can also lead to a too narrow range of the possible outcomes for an investment, a topic I will dive into in the next section. I believe Einstein put it best when he said:
Everything should be made as simple as possible, but no simpler.
Part 5: What about Skewness?
A further pushback – also from Rishi (can someone ask his Mum to tell her son to stop picking on me?) – was about skewness. Skewness refers to a type of probability distribution. Most people are familiar with the normal distribution – also known as the bell curve - where the more extreme the outcome (positive or negative), the more infrequently it occurs. A skewed distribution is one where extreme outcomes are more frequent than would be predicted by a normal distribution curve. It has “fat tails”. Skew can be negative or “left tail” when there are more negative outcomes than expected, positive or “right tail” with more positive outcomes than expected, or both. The implication of Rishi’s question was that the cash flow generating ability of some companies has a positive skew or “right tail” with truly exceptional outcomes happening more frequently than would be predicted by a normal distribution curve. Think Google or Amazon.
In a world characterised by positive skew and right tails, it is a mistake to focus on a point estimate of cashflow or assume cashflow outcomes are normally distributed. The investor should instead focus on the potential range of cash flow and in particular the likelihood of a positive fat tail outcome. Consider two opportunities both with the same expected long-run cash flow but one where this is normally distributed and one where there is a strong skew to the positive. Clearly, the latter opportunity is more attractive. It contains the opportunity for an extraordinary outcome without a commensurate increase in the probability of a disastrous one.
The concept of “a right tail outcome” is powerful. If I am being intellectually honest, when I wrote my letter, a point estimate of cash flow was top of mind as opposed to its distribution or, to stay with the parlance, its potential “skew”. I can see how an excess focus on an estimate of cash flow may blind the investor to the range of possible cashflow outcomes.
Nevertheless, I believe the concept of positive skew enriches the discussion of accuracy as opposed to invalidating it. In effect, it suggests a more accurate way of forecasting a company’s future cash-generating ability. Clearly, investors that seek out companies with positive skew and correctly identify it will have superior investment performance to investors who ignore it, all other factors being equal. In effect, the incorporation of positive skew into a forecasting framework is a great illustration of how an investor can be more accurate. No doubt, there are many others.
Part 6: Does Valuation Still Matter?
Given the near-unlimited gains theoretically possible from investing in a company that has a strong positive skew to its future cash flow, a legitimate question might be whether investors should discard their intrinsic value calculations altogether and focus instead on buying companies with strong positive skew. An argument could even be made that if an investor buys a portfolio of companies with positive skew, a low probability outcome could be turned into a near certainty.
In my view, they should not. The concept of cash flow as a distribution enhances an investor’s ability to calculate a company’s intrinsic value, but it does not obviate the need for said calculation. If the market underestimates skew, a company may be undervalued. If it overestimates it, it may be overvalued. Either way, the job of an investor is to figure out a company’s intrinsic value and then buy it at a discount.
A company like Shopify illustrates this point well. Clearly, its future cash flow is likely positively skewed – it is easy to imagine a scenario where it experiences prolonged, extraordinary growth, and difficult to imagine a truly disastrous scenario. Equally clearly, some amount of positive skew is already priced into its stock as there is simply no scenario in the next few years where its annual cash flow could possibly justify its current valuation. The job of the investor is to form as accurate a view of its intrinsic value as possible and then invest accordingly.
If the intrinsic value is the sum of a company’s discounted future cash flow, there is simply no way to avoid forming an opinion on a company’s cash flow generating ability as a means to figure out its intrinsic value. This is investing’s sole immutable law of nature.
Part 7: What Nicholas Taleb Misses
Probably no single writer has done more to popularise the idea of fat tails and skew than Nicholas Taleb. However, his failure to incorporate the role of intrinsic value into his thinking is, in my view, a large gap in his theory when he applies it to investing.
In “Antifragile,” Taleb describes how he views the best investment opportunities to be in deep out-of-the-money options. This clearly leans on his conviction that humans underestimate the likelihood of extreme events. So far, so good. However, then the discussion simply stops and moves on to a different topic. There is no mention of valuation whatsoever. Clearly, not every out-of-the-money option can always be cheap at any price.
If his argument is that human nature is unchanging and deep out-of-the-money options will always be cheap in aggregate, I am sceptical. After all, markets are nothing if not adaptive. Even then though, it does not obviate the need to calculate the intrinsic value of an option. Only then is it possible to figure out how much to invest, i.e., sufficient to make a difference to the portfolio when the hoped-for fat tail event happens, but not so much that the investor is bankrupted from purchasing options that expire worthless before the hoped-for event finally materialises. There is simply no way to avoid a calculation of intrinsic value. The existence of fat tails does not obviate the need for a robust framework for calculating it.
Bringing It All Together
I started this memo by reiterating the key point of my 2020 letter that the primary job of an investor is to calculate a company’s intrinsic value by forecasting its future cash flow generation as accurately as possible. This led to a discussion on why this is more obvious to younger investors than older ones, the potential drawbacks of experience, the need to be as accurate as possible, but not more so, and the concept of skew as a more accurate way to think about the future in some cases. It finished up where it started though, as powerful a concept as skew might be, it does not obviate the need to calculate a company’s intrinsic value as accurately as possible.
