Investment Uncertainty Is Not Going Away: Understanding Complex Systems Can Help
In the years since the 2008 crisis, I have seen a number of job postings for risk managers, advertisements for risk analytics, and papers on risk management. Judging by this burgeoning industry, risk is well understood and perhaps even under control in the investment process.
The funny thing, though, is that I am not aware of any uncertainty managers. It is funny because in the aftermath of the crisis, many people agreed that statistical approaches fell short of capturing uncertain events. Something needed to be done. Talk of black swans and fat tails filled the air. Then, not much. The proof is in the process, and the investment process seems devoid of analytical approaches to uncertainty.
But there is good news. An approach called “complex systems” can help us get a better handle on investment uncertainty — not just in a conceptual sense but as a part of the investment process used by everyone from analysts to plan sponsors. Its purpose is not so much to prevent the downside but rather to provide its user with an edge. It does so by providing a means of pattern recognition that picks up where statistics and finance leave off.
Uncertainty and Soccer Goals
Economist Frank Knight pointed out many years ago that uncertainty is when we can’t assign probabilities to events in advance. This seems like a simple and obvious statement. It cuts through the very foundation of economics and finance. Both of these disciplines assume away uncertainty. They are built on the bedrock that, as rational actors, we know the probabilities and payoffs of everything that may happen in the future. “Risk” is when one of the lower-probability outcomes occurs.
Think of us sitting at a blackjack table where we have two face cards and the dealer is at 16. The dealer’s next card is a five. We lose — not because we didn’t know what might happen but because we took a risk. Risk means we might be disappointed, but we should never be surprised.
Surprise, therefore, is a good stand-in for uncertainty. The word brings me to soccer. Bear with me for a minute because I think you will see how uncertainty, soccer, and the economy are related. Like most fans, I love to watch soccer goals. I revel in the replays, in seeing the same action over and over again. Why is it that we can watch goal replays repeatedly and not get bored? Soccer goals are thrilling because they surprise us, and they continue to do so even after we have witnessed them several times. This happens because the game is complex, and as a complex system, it has three attributes that produce surprise (i.e., uncertainty):
- Soccer exhibits a high degree of interdependence. Player 1 adapts to Player 2’s moves, which influences Player 3, which then influences Player 1 again. Iterate this through all 22 players on the field, and you quickly lose track of the potential outcomes.
- Soccer demonstrates nonlinearity. A tiny wrong move by one player can result in a winning goal scored in the 89th minute of a World Cup final. Little causes have big effects, and these, of course, are difficult to predict.
- Soccer is path dependent. Every time the ball is touched, it produces a whole new set of possible outcomes. It is impossible to tell from one of those early touches which path the play will take, from a loss of possession to a stadium-shaking goal.
Now combine all three dynamics and have them interact in all sorts of haphazard ways. The result is a key aspect of complex systems. Goals are a manifestation of a soup of complex variables. They emerge from that interaction, making the game’s outcome more than just the sum of each team’s relative skill.
Complexity is like a novelty engine. It is constantly “inventing” new outcomes. We cannot know the probabilities of these outcomes in advance. Uncertainty, therefore, is born of complexity.
Of course, the economy is much more complex than a soccer game and is getting more so all the time. As analysts, our ability to predict outcomes is affected by that complexity. Sometimes the system behaves and acts as if it were predictable. Other times, the three dynamics — interdependence, nonlinearity, and path dependence — throw surprises in our way. Unlike soccer goals, however, these surprises are less welcome. At least if we understood the game better and could recognize patterns in it, our predictions would not founder as often. If we could understand these patterns better than our competition does, this might even give us an investment edge.
From Soccer to Science
Scientists in fields as diverse as ecology, genomics, and traffic engineering are all targeting the same problem: how to understand, and even predict, complex behavior (not behavior in soccer, of course, but of their various systems). DNA, for instance, is not a rigid “blueprint” for cell behavior. Instead, it is part of a system wherein cells react to their environment and trigger DNA reactions, which then trigger more cell reactions, which then influence the environment. Even with a sequenced genome, scientists find it difficult to predict the action of a potential drug molecule in the midst of all this networked interaction. As in soccer, the odds are difficult to calculate in advance. Uncertainty intrudes; surprises happen.
Scientists, of course, don’t give up easily. They engage in pattern recognition. Fortunately, they have been able to identify a set of patterns that tends to reoccur in complex systems: feedback loops, network cascades, evolutionary beauty contests, and the like. One of the most common patterns is called a “positive feedback loop.” It’s not positive because we like it but positive because it is self-reinforcing. Examples abound in nature: The most obvious one is an epidemic. As a virus spreads to more people, there are more people who can spread the virus. Contagion feeds more contagion. Eventually the process loses fuel when most of the vulnerable population has been infected. A graph of the virus’s progression through a population looks like an S curve. It starts slowly, accelerates, peaks, then begins its decline.
This same S-curve dynamic explains everything from the length of a peacock’s feathers, to the population boom of rabbits in Australia, to the pattern of traffic caused by a braking motorist on a California freeway. It is not by any means the only pattern to come out of complex systems, but it is one of the most relevant to the investment analyst.
From Science to Investing
Like a scientist, the analyst tries to understand and predict the behavior of a system through pattern recognition. In an analyst’s case, the relevant system is our highly networked and complex economy. Like systems in nature, the economy sometimes exhibits a stable, predictable pattern. For instance, suppose sales growth for a company has averaged between 4% and 6% the past five years. There is every reason to believe that the environment will not change much in in the sixth year, so the analyst uses the average of previous growth to predict a growth rate of 5%. I am not knocking linear extrapolation. A lot of times this works.
The key, though, is to understand the role of uncertainty. Complexity means that systems do not necessarily behave in a stable manner. Their underlying patterns can shift, sometimes abruptly. The best example of this is the behavior of credit losses during the housing boom. In 2005, subprime mortgage delinquencies had been declining for years and analysts had every reason to believe that this decline might, at worse, flatten out in 2006. They figured that house price appreciation might slow but that it should not be a problem as long as there was no precipitous drop, and with unemployment low, there was no reason to expect such an event. The stable behavior of the system, however, masked an underlying positive feedback loop at work.
A look at state-by-state data from subprime loan pools showed that the Midwest states had delinquencies in the teens, whereas California’s was slightly below 2%. Unemployment and the relative growth rate of regional economies could not explain this yawning gap. The explanation lay elsewhere, in the link between home price appreciation and how borrowers reacted to falling behind on loans. In states with high price appreciation, a borrower’s home equity was constantly rising. Imagine a subprime borrower just about to fall behind on her mortgage payments. A call comes in from a mortgage broker allowing the borrower to refinance her mortgage and cash out that newly minted equity, using it to make payments on the new loan. Presto! Delinquency avoided.
As “cash-out” refinances spread, delinquency rates dropped, leading lenders to relax underwriting standards. The lower standards brought in new borrowers, who, in turn, increased demand for housing, causing house prices to rise further. The more house prices rose, the more delinquencies and loan standards fell, leading to further rising house prices. It was a textbook positive feedback loop. The implication of the loop’s existence was profound. Cash-out refinances were curing delinquencies. Without price appreciation, there would be no newly minted equity to cash out. In contrast, flat home prices would lead not to stable delinquency rates but to a spike of California delinquencies closer to the Midwest’s levels — from around 2% to the mid-teens.
The vast majority of subprime credit and stock analysts never expected this abrupt shift in the “system’s” behavior. An understanding of complex systems patterns might have allowed them to recognize the underlying pattern at work. Analysts who did so would not have merely avoided a potential problem: they would have gained an edge over the competition.
A Process for Capturing Uncertainty
It is tempting to believe that the 2008 crisis was just a one-time event, one in which uncertainty briefly mattered and, perhaps, a complex systems approach was relevant. This black swan or even fat tail view of the problem leads us to think of uncertainty as an exception rather than the rule. This “exception” view is misleading. Our economy is a complex system, and it does not stop being one, not even in periods when it behaves in a stable, predictable manner. The truth is that complexity is always with us, and it is always producing uncertainty. There are many areas in which I see complex systems patterns at work today.
For instance, corporate profits raced to new peaks after the 2008 crisis, even though economic recovery was lackluster. What explains this? Perhaps a positive feedback loop, one in which competition is restrained as CEOs seek the certainty of buyback-led EPS growth over uncertain capital expansion. The less expansion, the less one company might have to compete for share with another: less competition, higher margins, higher stock prices, more reward to focus on share buybacks.
Another example is volatility: In asset class after asset class, we have lower volatility than we have seen in years or even decades. Markets appear highly certain about the future, even though most would agree that the real economy remains uncertain. What accounts for this dynamic? Network science, a branch of complex systems, argues that as the external influences on a network fall, the network becomes more homogenous in its beliefs. Something like this may be happening as a result of the dominant influence of US Federal Reserve rate guidance on markets. The more the Fed influences the network, the more homogenous the beliefs about the future, the lower the volatility, the greater the credibility of the Fed, and the more it influences the network.
The process I have in mind searches actively for these types of patterns. It “scans” the investment ecosystem for them, recognizes them, and then tags them. The goal is not necessarily to make big bets on still-uncertain outcomes. Instead, this scan-and-recognize approach allows analysts to track the evolution of a pattern in real time, to understand its behavior so that as data come in, the portfolio can adapt to possible outcomes. This is uncertainty management.
Frank Knight argued that although the economy is rife with uncertainty, some people could grasp it better than others. They could see the “categories” of what might happen. Until now, we lacked a process for carrying this out. With complex systems, however, we have closed the “Knight” loop — from understanding the nature of uncertainty to actually being able to do something about it.
The uncertainty management process is not perfect, and it will probably never become so. An analyst’s edge, however, does not emerge from perfection. It emerges from being better at something, sometimes even in a small way. Nonlinearity rules. A tiny advantage can result in a goal scored. We want to become the emergent phenomenon of the winning team. To do so, it helps to develop an awareness of complex patterns and to use this awareness to better understand uncertainty.
If you liked this post, don’t forget to subscribe to the Enterprising Investor.
All posts are the opinion of the author. As such, they should not be construed as investment advice, nor do the opinions expressed necessarily reflect the views of CFA Institute or the author’s employer.
Image credit: ©iStockPhoto.com/Hong Li