Extreme Outcomes: Why We Underestimate Acute Risks

It's best to accept that at some point you will lose money and avoid betting too heavily on any one outcome, says Morningstar's Samuel Lee

Samuel Lee, ETF Analyst 15 May, 2013 | 1:25PM
Facebook Twitter LinkedIn

It's a law of nature that bad things happen more often than not. The possible outcomes that humans consider desirable are a sliver of all possible outcomes. It's why genetic mutations usually hurt or kill offspring born with them, why throwing a wrench into a finely turned engine will almost never make it run better, and why stock prices are more prone to sudden collapses than sudden rises.

Consider human history. For all but a fragment of it, about half of all children died before their fifth birthdays. Of the ones who survived, life was nasty, brutish, and short. According to data compiled by economist Angus Maddison, the total world population grew at an annualised rate indistinguishable from zero over the period from 1 AD to 1820 AD. Per capita output grew even slower, if at all. While the data is necessarily sparse and uncertain, the estimates are reasonable. A growth rate of, say, 0.5% annualised over 1,820 years implies growth by a factor of more than 8,750, an implausibly high number. Ergo, past growth rates were likely lower.

Over the millennia, there were pockets of progress, some lasting for centuries. The ones we know of included the Roman Empire in Europe, the Tang and Song dynasties in China, and the Inca, Aztec and Maya civilizations in the Americas. However, these civilisations, for the most part, did not set growth on a permanently higher trajectory. They usually ended in the sacking of great cities and the burning of books. Humanity didn't convincingly break out of this cycle until a couple of centuries ago. Against the yardstick of written history, this era of growth has been short--only a handful of lifetimes.

Even narrowing our scope to the 20th century, our current prosperity looks surprisingly fragile and uneven, the outcome of a series of contingent events that plausibly could have led to far worse outcomes. The future of democratic capitalism was up for grabs during the Great Depression. World War I and World War II devastated Europe twice over and Asia once. The Cold War risked nuclear apocalypse. If we could reroll the 20th century many times, creating slightly different starting conditions each time, some outcomes surely would have led to regress (the probability of such outcomes is unknowable, of course). If you were born to a random family in the first half of the 20th century, chances are you would have experienced privation or total war. Even today, the majority of humanity lives in poverty by the rich world's standards. Bad things happen more often than not.

I don't write to frighten you but to let you know how good you've had it these past few decades. They were the most peaceful, stable, prosperous period humanity has ever experienced. Unsurprisingly, I've found that many investors have what I believe are implausibly optimistic beliefs about the prospects of economic and societal continuity. For example, I've asked several people, "What is the probability you assign to the US government collapsing or disappearing over the next 100 years?" Responses typically expressed the value as "less than 0.0001%." While patriotic, the estimate implies that if we could reroll the next century as many times as possible with slightly different starting conditions, the average number of years to failure would be 100 million (100 years/0.0001%).

A sound principle of forecasting uses historical data to set a baseline probability and moves away from that baseline depending on the strength of new information. History suggests the average major civilisation experiences economic continuity for a few centuries, if that. However, the advent of nuclear weapons has ruled out conventional warfare as a way to wipe out a major civilization. In addition, there haven't been many democratic, capitalist societies, so we're left to make educated guesses. Given how close the US has come to disintegrating over the past couple of centuries, an average life span of 5,000 years seems reasonable. It implies a per-century failure rate of 2%--still uncomfortably high, and far higher than the probability a typical investor would provide without engaging the data.

I believe human nature makes it difficult to accept this argument. Deep down in our bones, we are an overconfident, optimistic bunch. We unjustifiably project the recent past to the far future, and we privilege emotionally vivid experiences above other forms of knowledge. It's why we're so often surprised by the way the future transpires--the future doesn't draw its outcomes from the impoverished imagination of the conventional mind. It's also why so many people find it hard to learn from history.

Learning From History and the Markets Is Hard 

We can see how human nature makes learning from the markets an iffy proposition for the typical investor by contrasting it with a typical learning process. Consider learning to drive. (I confess I don't know how to.) If your car swerves, a course correction is a turn of the steering wheel. If your car is moving too slowly, speeding up is a press of the gas pedal. The right thing to do in each case is crystal clear. Moreover, the feedback is near-instantaneous and frequent. Unsurprisingly, almost anyone can learn how to drive.

For the majority of tasks, people figure out what works and what doesn't, and they do it well enough to get by. However, their rate of learning is constrained by two factors:

1) The efficiency with which people integrate new information;

2) The rate at which feedback arrives, and its quality.

In most cases, people are efficient learners because they're dealing with simple cause-and-effect relationships, and the feedback they receive is both high-quality and frequent. It plays to the brain's inherent strengths.

Investing, on the other hand, is an unnatural activity that demands a cold, probabilistic mindset that few people can master. As a result, most investors do not efficiently integrate information from the markets because the brain is hyperactively spinning coherent stories out of all the data presented to it, stories that largely happen to be false. Most investors don't realise this, and believe that they are "learning" from the markets by observing patterns and constructing stories.

The evidence is everywhere, if you know where to look. It's easy to find a lot of very persuasive-sounding people talking about the markets and why they're doing what they're doing. It's not so easy to find people who can actually predict what the market will do. What this tells me is that many investors are fooling themselves or learning the wrong things.

The most corrosive of these "wrong things" is the belief that past performance strongly indicates future performance. Investors make the seemingly logical inference that if they want high returns, they should buy things with high returns and avoid things with low returns. In the short run, their beliefs are validated, as rising prices attract return-chasing investors, causing prices to rise even further, and so on. However, eventually their collective behaviour pushes prices far beyond fair value, leading to reversal. The correct inference is that while returns are self-fulfilling over the short run, in the long run, extreme price movements up or down tend to reverse themselves.

Many who have invested long enough know how these cycles of momentum and reversal work. In fact, I'd say most professional investors understand them, because it's impossible not to run into examples over the course of an investing career. See Apple's (AAPL) recent rise and fall. (The investors who don't learn from them are lost causes, doomed to lose money.) However, there is another factor that catches even many professional investors unaware: long-wave market cycles.

I'm not talking about the kind of cycle you can plot on a chart and attempt to divine from price movements. I'm using Bridgewater Associates founder Ray Dalio's definition: a sequence of logical events that repeat. The financial crisis was the culmination of such a cycle, the build-up of leverage over the past three decades. The last of its kind occurred in the US during the 1930s. "Long-wave" events occur once--perhaps never--in a lifetime and are rarely anticipated by investors because those who experienced them are long dead. In other words, the rate and quality of the market's feedback is both slow and low, meaning most people never learn how to deal with long-wave events.

Given half a chance, humans will ignore such long-wave dangers because they judge the likelihood of events happening based on how vividly they can imagine them, and they can most easily imagine events they've gone through themselves. The only way around this is to supplement experience with the study of history and to seriously consider that what's happened before can happen again. You will look like a loon most of the time. But safely navigating the future will require independent thinking, because the average investor with a conventional portfolio will almost certainly experience devastating losses at some point.

Lost Decades 

The 2000s were an object lesson in long-wave risk. Many investors are now under the impression that they've lived through a bear market, surely the worst of their lifetimes. While the tech bust and the financial crisis were extraordinary, aggressive central bank easing and massive government deficit spending ensured that the extraordinary circumstances didn't produce an extraordinary decade of returns or volatility. In Exhibit 1, I've produced the cumulated returns of a simple 60/40 S&P 500/intermediate government bond portfolio, rebalanced at year-end. The numbers are adjusted for inflation, and the chart's Y-axis is on the logarithmic scale, meaning each uptick represents a 10-fold increase. By these lights, the past decade looks unremarkable. You only had to go back to the 1970s to find worse drawdowns and longer periods of underperformance.

The most striking thing about this graph is how clumpy returns are over decade-long spans. The 1950s, 1980s and 1990s account for almost all of the portfolio's post-World War II returns. Imagine if you retired during the 1960s or 1970s: Your portfolio would have gone nowhere, with tremendous volatility. By the late 1970s, valuations for both stocks and bonds were at generational lows. It was a tremendous transfer of real wealth from retirees, who were drawing down their assets, to savers, who went on to enjoy outsized returns for the next two decades. 

Lost decades (and ruined retirements) are not unusual events. Out of the past eight decades, starting with 1931, three (the 1930s, 1970s, and 2000s) showed returns of less than 1% annualised after inflation. Counting taxes and management fees, those returns would have been reduced to zero or worse. A three-in-eight chance of experiencing a lost decade doesn't sound too great.

Severing the Left Tail 

Frankly, most of us will not do a great job timing the huge shifts. A good strategic portfolio for the typical investor should presume that the future is largely unknowable and that various types of disaster will eventually occur, whether it's boom or bust, inflation or deflation. In other words, I believe investors should consider owning a risk-parity portfolio. The key to understanding risk parity is to realise that two great economic forces determine most of an asset class' behaviour: economic growth and inflation. These produce four possible economic scenarios:

  • Rising inflation
  • Falling inflation
  • Rising growth
  • Falling growth


More precisely, only the unexpected changes matter. If the market expects inflation to be 10% next quarter, and inflation turns out to be 10%, asset prices will not budge because expected inflation is already factored into the price. Conventional portfolios performed poorly during the financial crisis because they were overly reliant on rising economic growth and rising inflation. The diversifier of choice back then was commodities. Unfortunately, 2008 was a perfect storm of falling inflation and falling economic growth. The only asset class that could hedge against such a scenario was nominal Treasury bonds.

Asset classes react differently to each economic scenario. It rarely matters what kind of equities you hold--small or large cap, growth or value, quality or junk--if economic growth turns out to be lower than expected, equity prices will fall. I'm not claiming equity styles don't offer diversification. Over long periods of time, different styles can produce very different returns, especially when starting valuations are very different. However, over a lifetime, your exposures to different macroeconomic environments will affect your outcomes the most.

A controversial method I advocate is to obey trend-following signals. Such strategies are pro-cyclical, cutting exposure to an asset when its price falls below its moving average, usually measured over 200 days, and adding exposure when an asset's price rises above its moving average. At first glance, there's no fundamental reason why it should work. Both dyed-in-the-wool value investors, a hugely successful group, and many academics dismiss it as voodoo. However, during the past couple of decades, academics have chipped away at the efficient-markets dogma, and many acknowledge that markets do display cycles of self-perpetuating fear and greed. (Tobias Moskowitz, Yao Hua Ooi and Lasse H. Pedersen wrote the most influential study on this subject.)

Trend-following strategies have produced excess risk-adjusted returns in virtually every market studied over long periods. There are a couple of theories as to why they work. The first is behavioural. Investors are herd animals, stampeding in and out of markets at the same time because of irrational behavioural biases. The second is fundamental. Central banks often manage interest rates and exchange rates to achieve macroeconomic stability. Their interventions are often countercyclical, preventing prices from fully reflecting the market's opinion of fair value.

Exploiting trends requires paying higher transaction fees, incurring more taxable gains, and suffering from occasionally eye-popping tracking error to conventional benchmarks. Despite these costs, trend-following provides cheap insurance against a lost-decade scenario.

The markets will at some point suffer a prolonged bear market for any number of reasons. The average investor is certain to lose a lot of money at some point in the future. It's best to acknowledge this fact and avoid betting too heavily on any one outcome. The implications are unsettling, but not nearly as much as losing everything. Bad things happen.

A version of this article ran in the February 2013 issue of Morningstar ETFInvestor.

The information contained within is for educational and informational purposes ONLY. It is not intended nor should it be considered an invitation or inducement to buy or sell a security or securities noted within nor should it be viewed as a communication intended to persuade or incite you to buy or sell security or securities noted within. Any commentary provided is the opinion of the author and should not be considered a personalised recommendation. The information contained within should not be a person's sole basis for making an investment decision. Please contact your financial professional before making an investment decision.

Facebook Twitter LinkedIn

About Author

Samuel Lee, ETF Analyst  -

© Copyright 2024 Morningstar, Inc. All rights reserved.

Terms of Use        Privacy Policy        Modern Slavery Statement        Cookie Settings        Disclosures