When is a Bubble a Bubble?
This post was originally published at the Institute for New Economic Thinking
Bubbles have become a major focus of discussion in today’s financial markets. But very few people actually define what they mean when describing this financial phenomenon.
In a recent Harvard Business Review blog post, Markus Brunnermeier, an economist at Princeton University and a member of the Institute for New Economic Thinking’s Advisory Board, had a go at it. Brunnermeier defines the leading characteristics of bubbles thusly:
Bubbles are typically associated with dramatic asset price increases followed by a collapse. Bubbles arise if the price exceeds the asset’s fundamental value.
Well, that’s part of it. It certainly describes a characteristic of bubbles – namely that they represent a massive fundamental departure from the asset’s underlying value.
But does that give us everything?
Bubbles also are about trend following behavior that develops positive feedback effects. Larry Summers and colleagues wrote a famous paper in 1990 that set out in simple terms this kind of trend following feedback dynamic. Didier Sornette has recently done the same, though in a very, very complicated way.
Even though households have learned something from the two 50% bear markets in recent memory, in light of the recently rising stock market many now feel compelled to play the game. Money managers were taught the same lesson regarding potential loss from those two bear markets, and they also are now worried that the Fed will take the punchbowl away. But they, too, feel pressured by the past rising trend in stock prices to “play the game.”
It is this feedback effect from a steeply rising trend in past stock prices that is the hallmark of a bubble. In the United States, we are in an incipient bubble stage in which households and money managers are tentative, cynical, self-aware trend chasers. It is the unwavering corporate net purchase of equities regardless of valuations that hold these less resolute players in the game.
There is another important feature of bubbles – namely, that the acceleration of price as the object of the bubble (whether it be equities, bonds, real estate, Dutch tulips, or dotcom companies) goes way beyond the asset’s underlying value as the bubble itself matures and intensifies.
This is something the French economist Maurice Allais has analyzed. Allais noted that when the rate of past price appreciation in a market is rising not only does the memory of the rate of appreciation rise in a lagged fashion, but the market’s collective memory becomes shorter. Which means that in each successive time period a higher and higher weight in the effective memory is given to the most recent and higher rate of change instance. This mechanism makes the adaptively based expected return explosive.
I think that this idea also seems to be behaviorally familiar. Now when you put all of these plausible mechanisms together you can make an adaptive model very explosive. That is what happened in the 1990s and is one reason why the bubble went parabolic even though the real interest rate was always above the historic average and money and credit did not grow faster than nominal GDP overall for the first eight years of the ten-year bubble. Now all this modeling focuses on the dynamics of euphoric return expectations.
But there is also a parallel dynamic based on adaptive behavior that focuses on the prospective risk of loss, which can be measured via downside volatility. That adds to the rate of ascent of expectations of risk adjusted returns, as opposed to just euphoric returns.
There is a reductio ad absurdum that captures the reality of this behavioral effect. Suppose that stock prices begin to go up every day, initially perhaps by very little, but every day. Then people with adaptive behavior will come to think there is less and less risk of a price decline and therefore a loss. If the pattern of no downside action persists for a period equal to the effective memory of market participants they will eventually come to believe that stocks cannot go down. If people think they cannot lose they will not sell. In a market where some will buy but where no one will sell, prices must levitate forever.
Now, the most astounding fact about the great bubble decade of the 1990s was that there were 84 months in a row in which the market did not fall by more than ten percent. The previous high in this figure was 28 months. So naturally people began to think that not only were returns so high that you could become rich quick by participating in the stock market; but there was seemingly no risk in chasing such quick riches.
I only mention this because while I think the economics literature from Keynes onward is very good on the propensity of markets to greatly overshoot and undershoot the fundamentals, economics per se does not adequately explain what makes the dynamics of bubbles more than an overshoot. In other words, what makes them recursive, explosive, parabolic? That is the difference between real bubbles and mere waves of pessimism and optimism that move markets all the time even when there is no rational basis.
To get a full measure of this one has to enter into the realm of psychology and neuroscience. That’s where the definition lies. Bubbles, like so much else, are too important to be left to the realm of economics alone.
Marshall is a former investment manager with 30 years of experience. He is now Director of Institutional Partnerships at the Institute for New Economic Thinking.