I am often asked why I expect value investing (traditionally defined as focusing on buying sectors of the market trading at relatively low multiples of earnings, assets/book value, and cash flows*) to continue to 'work' in the manner it has historically, given that we now live in an age of advanced computing power and widespread information dissemination. Surely, it should be a simple matter these days for algorithms to screen for quantitatively cheap stocks and buy them up, quickly arbitraging away any excess profit opportunity. Low multiple stocks must now all be ones fully deserving of such a low rating, and the opportunity which existed in the past in Ben Graham's day was a function of highly inefficient and unsophisticated markets, and is thus no longer relevant.
This argument is seductive, but fundamentally wrong in my view, because it lacks an understanding of the deeply-rooted psychological and structural institutional realities that drive the inefficiency and make profiting from such stocks much more difficult in practice than in theory - particularly on an institutionalised basis. Quant-driven strategies might at first seem capable of overcoming the former set of biases/constraints (although I discuss why in practice, they often end up doing the opposite), but they cannot overcome the latter structural institutional issues. Furthermore, a lot of the institutionalised barriers exist because of the psychological biases inherent not in the managers, but instead in the underlying client base, which as non-professional investors, are often ill-equipped to understand and deal with the deeply counter-intuitive nature of successful stock market investing.
One of the factors efficient market theorists overlooked is that there is a very sizeable principal-agency gap between the owners of the vast majority of funds that are actively-managed in equity markets, and the individuals/managers/organisations actually responsible for making the investment decisions. EMH is premised on the idea that large numbers of individuals are vying with each other in markets with a profit motive, neutralising each others efforts, but this approach fails to consider not just psychological factors that obstruct this process (something now widely studied by behavourial economists), but also the instituational factors and agency-principal conflicts that also do so.
The psychological obstacles for the investment managers themselves are also tremendous - so tremendous that I would argue that even the majority of self-described 'value' investors today, who believe themselves to be dispassionate long term investors attempting to profiteer from 'Mr Market's' temperament, are in actual fact not really practicing value investing at all, and are instead forming part of the herd that creates the opportunities genuine value investors have to systematically profit.
The fundamental issue underlying all these factors, I believe, is the nature of the payoff patterns deep value stocks typically exhibit, and why. A typical value stock has well-understood and well-publicised problems/issues/risks, and the majority of the time, for individual issues, these well-understood issues do result in subsequently lackluster investment outcomes (usually in the form of protracted periods of stagnant performance that lag go-go market favourates).
However, as a group, over time, they deliver better-than-market returns, because their widely understood problems and lacklustre prospects are excessively rendered in prices, and only base-case outcomes are priced in, while 'tail' possibilities - relatively unlikely events that no-one expects to occur - are generally ignored. And in the relatively infrequent cases where those unlikely events nevertheless do occur, outsized returns are generated which more than offset poor base-case outcomes on other positions. This payoff profile - a high probability of a poor outcome,** and a small probability of disproportionately very good outcome - is an extremely difficult payoff profile to practically exploit for institutions and individuals alike.
Take a stock like Whitehaven Coal (WHC AU), for instance. Between 2010 and early 2016, the stock fell 95% from A$7.00 in 2010 to as low as 30-40c. It went sub $2.00 by 2013, and for the next three years consistently and inexorably fell (after already falling by two-thirds, it then proceed to fall by another 80-85%). During the latter parts of this descent, the company was a leverage thermal coal miner losing money, and coal prices continued to fall. The outlook for thermal coal appeared terrible, as markets were oversupplied and China was moving away from coal towards cleaner sources of energy. A shale gas boom in the USD was also generating cheap gas that was displacing coal demand, and this seemed to be a harbinger of what was to come everywhere. Coal was being disrupted. The Chinese economy was also slowing at the time and many feared an imminent crisis, or at the very least, a continuing slowdown in heavy industry/investment. The most likely outcome, investors believed, was either bankruptcy or a significantly-dilutive capital raising (and perhaps both).
The market was right about that being the most likely outcome at the time. However, it ignored tail risks - less likely events that nevertheless could occur - and in this case a tail risk did indeed come to pass: China's economy suddenly and unexpectedly turned around, boosting energy demand, while at the same time the country undertook massive 'supply side reform' that resulted in a significant quantity of inefficient coal production capacity being shuttered. No one expected such a profound supply-side reform to actually be implemented in China, but it was, and thermal coal prices subsequently doubled. As a result, WHC rallied 10x in 2016 to A$3.00, and has since risen further to A$5.00. The company has now already paid off nearly all its debt and is currently paying out about 30c a year in dividends (almost 1x its share price at the lows), while investing in new projects. That is a 15-fold increase from levels as recently as 25-30 months ago. Those sorts of winners compensate for a lot of cheap stocks that drift sideways/down seemingly forever as part of the 'living dead', or the occasional highly-leverage/risky name that does in fact go to zero.
WHC was a value stock. But few people were willing or able to buy/recommend it - even putative 'value' investors. The obstacle was the following: If you were an analyst at a fund management organisation; or a fund manager with clients; or a sell-side analyst covering WHC, if you recommended the stock as a buy to your boss or clients, the most likely outcome was that the stock would continue to go down - potentially to zero - and that you would look like an absolute fool. Furthermore, anyone that recommended or bought the stock at any point on the way down would have suffered terrible losses and also looked stupid for a long time, while for years the bears would have looked prescient and vindicated. Worse, not only do you lose money, but you lost money on something that everyone believed was obviously going to lose you money. So you not only lose money, but you do so in a way that makes you appear foolish, imprudent, and ignorant. That can be enough to get you fired (or your organisation fired by its clients).
You can almost imagine the post-mortem after the obvious and predictable losses come in. John, everyone knew coal was dead. It was all over the papers. China was going to stop using the stuff and the world was moving towards natural gas and solar to ameliorate climate change. So what on earth were you doing buying a highly-leverage thermal coal stock? How could you not be aware of these issues - were you living under a rock? It was obvious this was going to happen. I really can't help but question your judgement after this.
Furthermore, in the comparatively rare instances where a tail event comes in and good money is made, you are apt to be seen as having merely got 'lucky', because by definition, the trigger will have been something wholly unforeseeable. Consider a sell-side analyst that recommend 5 very cheap and beaten-up stocks with obvious problems. The most likely outcome will play out in 4 out of the 5 stocks, and they will probably continue to go sideways/down and underperform the market. For the 5th idea, however, the stock might go up 100-200% - perhaps due to a takeover offer, or some other unexpected development. People will say, John, you're a hopeless stock-picker mate. You got 80% of your calls wrong last year. Sure, Stock #5 came in for you nicely, but that was only because you got lucky - no one could have reasonably expected XYZ to happen. I don't remember seeing you argue anywhere that XYZ was going to happen or was likely, so you were merely right for the wrong reasons".
Meanwhile, the exact opposite is true of market favourates with good outlooks and solid earnings growth. 80% of the time, those stocks meet expectations and the stocks go up 10-20%. Everyone feels great. But 20% of the time, something unexpected happens and the stock is down 50%. An analyst recommending all five such stocks could say "I got 80% of my calls right last year. We just made a mistake on one position. We overlooked that that the competitive dynamics of the industry were set to deteriorate, and should have paid more attention to factors X, Y and Z. Lesson learned. However, our overall process is working as 80% of the time we were right. We just need to tweak our process to incorporate the lessons learned from this one outlier loss".
A consequence of this lobsided payoff dynamic is that institutionalising such an investment process is exceedingly difficult. A good analogy here is politics: Political systems select not for people good at governing, but instead people that are good at getting into (and staying in) power. This often results in people of suboptimal capability running the country. The same is true in financial markets - the investment management business selects for organisations that are good at attracting and retaining client funds, not good at managing them (per se). Even if - as is often the case - the managers themselves are good investors, often their craft is heavily restricted by institutional constraints designed to prevent major performance divergences/surprises.
Good investment performance is of course great for attracting client funds, but more important than actual returns over time, from a business perspective, is the nature of those returns. Most clients prefer predictable, low-volatility returns slightly above average, rather than significantly above-average returns in the long term that come attached with significant volatility and performance variation (which also opens the manager up to being accused of making better returns merely by 'taking more risk'). Consequently, it is not an optimal strategy for fund management organisations seeking to attract and retain the maximum funds under management with minimal performance volatility to base their investment approach on betting on tail outcomes, which the majority of the time will not pay off. It instead pays to stick with the herd - look modestly right most of the time, and very occasionally spectacularly wrong (usually at the same time as everyone else, sparing one the worst of the reputational consequences), instead of vice-versa. Along with fees charged, this is likely the primary reason why the majority of mutual funds have underperformed over time.
Furthermore, it is not just a principal-agency issue - this non-linear payoff profile is also psychologically very difficult for the investment practitioner/analyst themselves, because investing in/recommending such stocks requires one to endure a continuous stream of negative reinforcement most of the time, of which our human psychies are not well adapted to withstand. Indeed, academic research shows that most people's ability to remain rational breaks down in an environment of constant negative reinforcement. Day after day, month after month, and even year after year, the market, friends, associates, the media, and clients are telling you you are wrong and are a fool, and most of the time that judgement will seem vindicated by subsequent outcomes.
It also opens the door to 'false learning': if you base your lessons from experience on outcomes rather than process, most of the time value stocks will generate poor outcomes, and most of the time more expensive growthy stocks will deliver better outcomes, so you are actually apt to learn the wrong lessons and become a worse investor over time with experience, because you will learn through experience to avoid exactly the type of stocks you should be buying (and vice versa).
Many so-called 'value' investors fall into this trap. Most value investors these days do a lot of things that are actually the antithesis of true value investing as described above: they focus on buying good businesses with good outlooks trading at 'reasonable' valuations (read full/high but not absurd multiples), and they invest in concentrated portfolios. This is the antithesis of exploiting the market's tendency to overprice the best businesses with the best outlooks and underprice the worst businesses with the worst outlooks; and it focuses - just like the market - only on the base-case, most-likely outcome, and generally ignores tail risks. And yet it is changes of opinion, driven by unexpected events, which drives the vast majority of the big moves (and returns/losses) in markets. The fundamental issue underlying this dynamic is that investors systematically overestimate their ability to predict the future, and are therefore prone to overconfidence and excessive extrapolation.
These factors are structural and endemic to markets and cannot and will not go away, in my opinion. Even quant strategies find them difficult to exploit, because underlying investors into quant funds may not have the tolerance for significant and prolonged performance deviations. Much of the observable and documented trending/momentum in markets exists because end-investors allocate more money to markets/funds/countries/sectors etc that have recently performed well, and sell those that have recently performed poorly (because the outlooks, which appear close to certain to occur to investors at the time, are believed to be respectively bright and dire, and continuing realised investment returns/outcomes seem to be vindicating those beliefs).
These pro-cyclical adjustments simply prolong existing momentum, as managers holding cheap stocks are forced to sell them to fund redemptions, while growthy in-vogue sectors receive the inflows which further drive up valuations when invested. This acts to perpetuate recent performance disparities that trigger yet further portfolio reallocations, and so forth. This can go on a long time - for periods of years that can exceed the patience of the typical investor/asset allocator. This is likely why Greenblatt notes that a significant portion of the best performing funds over the course of a decade usually spend at least three years in the bottom quartile of performance, and why Grantham notes that while GMO's funds typically make it through a full cycle with above-average returns at below-average risk, they do so 'not necessarily with the same clients we started with'.
If you are trying to run a stable investment management organisation with a large number of employees and stable earnings, you simply cannot run an investment management business that can lose 80% of its assets in a cycle (through redemptions), and then regain all of that back and more when the cycle turns. You will go out of business during a downturn and/or lose all your key employees. Many value funds folded during the dot.com bubble for this reason. Only a small number of niche performance-based funds have managed to do so. Usually, they are lean operations owned and run by the key investment staff that have sizeable amounts of their own wealth invested in firm's funds which allow them to withstand FuM drawdowns/performance deviations.
Because of this, most quant funds are equally ill-adapted to exploit these value inefficiencies - particularly because the 'machine-learning' algorithms used are apt to look for short term correlations, which actually results in the models betting on the most likely short to medium term outcome, rather than rare/outlier events - the exact dynamic which drives market inefficiency in the first place. Indeed, it is arguable that quant strategies are structurally pro-cyclical and momentum-oriented for exactly this reason, and it is worth observing that every time algorithmic methods have been used in markets historically, it has lead to a major crash as 'six sigma' events occur that the models didn't/couldn't predict because it had not happened in markets before (for instance, the GFC was caused by flawed agency MBS pricing/rating models and flawed VaR models by leveraged investment banks that underestimated potential volatility in these instruments; and the 1987 stock market crash by algorithmic portfolio insurance). In that respect, the growing presence of quant/automated trading strategies is actually likely to amplify the opportunity for genuine value investors, rather than arbitrage it away.
In short, I am optimistic that opportunities for genuine value investors (as opposed to the large number of swarming would-be Buffett copycats that don't truly understand why Buffett does what he does and don't really understand the true essence of value investing) will remain prevalent, probably indefinitely. However, the size of the opportunity that exists is inversely proportional to the ease with which it can be exploited by institutions on an industrial scale. If it were easy to systematise and exploit at scale, the opportunity would go away, but I believe that is unlikely to happen.
*Many putative value investors today seek to redefine value investing to be based more on qualitative factors such as business quality, management, competitive advantage, growth potential, ROE, etc, relying on many Buffettisms that are liberally taken out of context, and believe quantitative value strategies to be naive/outdated. They focus more on 'intrinsic value', which means whatever the relevant investor thinks a stock is subjectively worth, and which can be used (and routinely is) to justify almost any valuation of good quality businesses. However, they forget from where the notion that value investing 'works' derives from: the long term statistically-documented tendency for quantitatively low-multiple stocks, as a group, to outperform quantitatively high-multiple stocks, as a group. They substitute statistically-demonstrated facts for unsubstantiated beliefs.
The truth is that there is no statistical evidence that high quality stocks - however they are measured - systematically outperform. In fact, there is evidence to the contrary, and it makes sense why: investors overestimate their ability to predict future growth and business quality, and underestimate the capacity for change. Consequently, investors systematically overpay for growth and quality. The problem is that when aspirant 'value' investors come to implement the philosophy, they notice that all the cheap businesses have problems of one type of another, and so avoid them. They end up seeking quality instead of value, and forget that you are rewarded in markets not for identifying and owning good companies, but instead for identifying and exploiting mispricings. It turns out markets are 'too efficient' at pricing in growth and quality - it is too well recognised so they overpay for it.
**As noted, in most cases, the 'poor' base case outcomes I am referring to are a long period of stagnant share price performance lagging market favourites, rather than outright absolute losses. WHC was an example of a stock with a particularly wide range of potential outcomes.