The trouble with calling market bubbles

The original version of this article first appeared in The Business Times.

Michael Burry, the contrarian investor Christian Bale played in The Big Short, earned his reputation by seeing what almost no one else could: that the “safe” label stamped on mortgage-backed securities in 2007 bore no relation to the rotting loans beneath them. He was right, spectacularly so, and the call became the stuff of financial legend and a movie.

He has also, in the years since, wrongly predicted at least two recessions that never arrived. Which raises an uncomfortable question. Was the first call genius, or was it just lucky timing? And the harder one behind it: we see so many predictions from where the market will be headed to the loud bubble warnings but how many times have they actually been right—and how often have they simply been early, loud, and wrong?

This is not an attempt to question anyone’s skill. Many of the economists, strategists and policymakers who have called market tops too soon are among the most respected voices in finance. The problem is not personal—it is structural. Identifying a bubble in real time, the empirical evidence shows, is one of the hardest problems in all of financial economics, and the behaviour of actually acting on it is harder still. 

Even the definition of a bubble is contested

Strip away the drama and a bubble is simple to describe: a self-reinforcing loop in which the price of an asset can continue to rise beyond reason and the potential of its future more than anything it actually can produce. Momentum overrides discipline. The greater fool theory becomes the entire investment thesis.

Easy to define. Devilishly hard to identify. Even academics cannot agree on the threshold. One widely cited Yale study defined a bubble as a market that doubles and then surrenders most of those gains within five years. Examining 21 developed equity markets across 115 years of data, it found bubbles to be remarkably rare—between just 0.3% and 1.4% of all market periods, far below popular perception.

The more startling finding was what tended to follow a sharp run-up. After a market doubled within one to three years, a crash that erased the gains occurred only about 10% of the time. In 26% of cases, the market simply doubled again. In plain English: a boom is more likely to be followed by a further boom than by a bust. Any warning system tuned to flag every steep rise as a bubble is, by design, a machine for generating false alarms. 

What initially could easily be perceived as so called bubbles often is a reflection of the emergence of new technology, companies or industries that go on to become the mainstream dominant force generating real earnings and economic value. Think of the biggest companies in the world these days and the decade or decades long journey they have been on, from Walmart to Berkshire Hathaway to Google. 

The most famous false alarm

The most celebrated bubble warning in history is also one of the clearest false positives. On 5 December 1996, Federal Reserve chair Alan Greenspan asked aloud whether “irrational exuberance” had “unduly escalated asset values.” Markets flinched and the world heard a central banker calling the top.

He was wrong, at least on timing, which in markets amounts to the same thing. Over the next three years the S&P 500 nearly doubled, returning 31%, 27% and 21%. The Nasdaq tripled. An investor who heeded the warning and sold the day it was uttered would have forfeited a 105% gain before the correction finally arrived in 2000. Greenspan himself cut rates in 1998, pouring fuel on the very fire he had flagged.

Even Robert Shiller’s landmark Irrational Exuberance, published almost exactly at the 2000 peak, made its core argument back in 1996—four years and one enormous rally too early. His cyclically adjusted price-to-earnings (CAPE) ratio has fared no better as a real-time alarm. Wharton’s Jeremy Siegel found that between 1981 and 2015 the CAPE signalled overvaluation in 416 of 422 months. An investor who obeyed it would have sat out one of the greatest bull markets ever recorded.

The crashes nobody flagged

If false alarms are one failure, silent collapses are the other. The 2008 crisis is the defining case. A scathing 2009 paper signed by eight economists, concluded the profession had been collectively blind to the risk building in housing and the plumbing that financed it.

A handful did see it in a way but they were only partially right and often very early. Roubini warned the IMF in 2006 of a coming housing bust; Shiller had flagged the same in 2004. They were dismissed as an outlier. Tellingly, a Boston Fed study found that economists examining the identical data in the mid-2000s reached opposite conclusions. The signals—stretched price-to-income and price-to-rent ratios—were visible, but genuinely contested.

Japan is starker still. At its December 1989 peak, the Nikkei traded at 60 times earnings, its CAPE neared 100—twice the US extreme of 1999—and the land under the Imperial Palace was reckoned to be worth more than all of California. In hindsight, the signals were deafening. In the moment, the prevailing story that Japan was simply different was powerful enough to silence nearly everyone. As the bubble continues to expand, the loudest critics often give up too early and can never time the top.  

What the evidence does—and doesn’t—allow

So can a bubble ever be spotted in advance? The most direct study, “Bubbles for Fama” by Harvard’s Robin Greenwood, Andrei Shleifer and Yang You, offers a qualified yes. A soaring price on its own tells you almost nothing. But sectors that went on to crash shared a fingerprint: surging trading volume, a rush of companies issuing new shares, and price gains accelerating sharply in the final months. Taken together, those signals carried predictive power.

The deeper obstacle is behavioural, not statistical. Eugene Fama’s objection is logical: if a bubble were obvious, rational investors would sell and pop it, so its very persistence is evidence that few truly believe it. The incentives then make cowards of the rest. As Keynes observed, it is better for reputation to fail conventionally than to succeed unconventionally. An IMF study documented that fund managers ride bubbles they privately distrust, because they are paid on relative, not absolute, performance—and standing apart from the crowd is the one truly unforgivable risk.

This is the wisdom and the limit of the Minsky-Kindleberger model, which maps every mania through displacement, boom, euphoria, distress and revulsion. It is a brilliant description of the anatomy of a bubble after the fact. What it cannot do—what nothing can reliably do—is call the turn from euphoria to distress. Which is, of course, the only moment that pays.

Why you should avoid predictions in general

The economist Samuelson famously said economists have predicted nine out of the last five recessions. Recently that probably has deteriorated to ten of the last two recessions. Beware of any expert that likes to predict too much. 

The lesson is not that bubbles do not exist. It is that identifying one in real time, and acting on it more profitably than simply staying invested, is extraordinarily hard—even for the brilliant, the well-resourced and the famous. Being early, again and again, has proven indistinguishable from being wrong. The investor who fled US equities on Greenspan’s word surrendered three years of compounding for a correction that, when it came, still left disciplined holders ahead.

That is a liberating conclusion, not a defeatist one. If the top cannot be reliably timed, the winning behaviour is the unglamorous one: stay invested, diversify so that no single narrative can sink you, rebalance with discipline rather than conviction, and treat your own certainty as the most expensive position in the portfolio. In a game where even the experts cannot see the future, the durable edge is not trying to make a better prediction. It is following a better process.