With the Dow Jones at an all-time high this week, some people are saying that we’re in another bubble. See this recent story from the Wall Street Journal (subscription required). Because of the painful memories of the housing bubble of 2006 and the tech stock bubble of the late 1990s, such warnings strike real fear in some investors. It is worthwhile, therefore, to ruminate a bit about how bubbles work.
Although bubbles are generally blamed on human emotions (think “irrational exuberance”), the most basic cause of bubbles is not excessive human emotion but deficient human cognition. That is, we human beings have but limited cognitive powers. In particular, our ability to predict future market conditions is severely limited. Hence, when we participate in markets by buying or selling, borrowing or lending, we inevitably do so with imperfect information about the future. Very often our predictions about future values turn out to be wrong, which implies that the prices at which we traded were inaccurate. Without more, however, this mundane circumstance does not produce bubbles. Everything else being equal, we are as likely to undervalue assets as overvalue them, and on average the various mistakes of market participants should cancel out.
But sometimes there are special factors that can lead everyone or almost everyone to overvalue assets. For, from time to time, something very important affecting prices (the kind of things economists call “fundamentals”) starts to charge in a very significant manner that justifies higher prices for a certain class of assets. This is a good thing, not a bad thing, and markets tend to react appropriately: the relevant assets trade at higher prices. So, to take an example, in the mid 1990s, there was a period when the amount of data transmitted over the Internet was doubling every 100 days. This was an extraordinary period to live through. When I was summer associate at Wachtell, Lipton in 1996, none of the attorneys’ computers were connected to the internet, and the firm did not have email. By the time I started full-time in the fall of 1997, all the computers were online. The real gains from these technological changes, everyone knew, were going to be huge, which meant that some internet firms were going to make tremendous amounts of money. Which ones and how much? Of course, no one could really know the answers to such questions. Nothing like the internet had ever happened in human history, and all that anyone who was inclined in invest in internet stocks could do was gather the limited information that was available and make an educated guess. The stock prices of internet firms then reflected the aggregate guesses of everyone participating in the market.
Now, this aggregate guess was of course not going to be exactly right, but was the guess too high or too low? If, on the whole, the guess was too pessimistic—if, for example, the market guessed that internet traffic would continue to double every 100 days for the next year and then in fact it continued to grow at that pace for two years—then, as new information about internet usage and profits to be made from such usage became available, it would eventually have become clear to everyone that internet stock prices were on average too low. The prices would then have risen, and everyone would have been pretty happy about this. Although for a while prices would have been too low, no one fifteen years later would be talking about the “tech trough” of the late 90s; the temporary undervaluation would hardly have been noticed.
But, if, on the other hand, the collective guess of the market about the future value of internet stocks turned out to be too optimistic, then, at some point, when new information made this clear to all, prices would fall. This, of course, is just what happened, and fifteen years later memories of the event still return to haunt those who lost money when the market crashed.
Bubbles thus begin from the unavoidable uncertainty about the magnitude of important, positive changes in economic fundamentals. As is clear from the work of Robert Solow and others, most real economic gains come from technological change, and so the change in fundamentals that triggers a bubble is often a major technological advance. This was certainly the case with the tech bubble. It was also the case with the stock market bubble of the late 1920s. From 1924 to 1929, the American economy experienced unprecedented productivity gains as major industries were rationalized and financial professionals invented new kinds of financial services (such as installment purchases) that opened up new markets. For many consumer durables, this meant that costs were falling and demand was rising; there was thus a lot of money to be made. This was a wholly good thing, and stock prices should have—and did—rise accordingly. But just as with the tech bubble seventy years later, the collective guess of the market as to magnitude of the benefits to be created was too optimistic, and we got a bubble in stock prices that finally burst in 1929. Other famous bubbles, such as the South Seas bubble in England in the 1720s, the conglomerate stock bubble in the 1960s, the Nifty-Fifty bubble in the 1970s, and the Japanese real estate bubble in the 80s all fit this pattern closely.
Some people argue that human beings have systematic biases that, in situations of the kind described above, encourage us to err on the side of optimism (the aptly-named optimism bias, for example). I have some sympathy with this view, but I note two important qualifications. First, human cognitive biases come into play in this story only in response to a genuine, positive change in economic fundamentals. You can’t suffer from optimism bias until you have something to be optimistic about. Second, as I noted above, whenever the market takes too pessimistic a view of a positive change in fundamentals and prices are too low for too long, we barely take notice of the undervaluation even after it has become apparent. I am not aware of there even being a word to describe such market phenomena, and I don’t think this is because such things never happen. Rather, our failing to note such situations arises from the endowment effect, which makes us feel the pain of losses more keenly than the pleasure of gains. People who, under the dubious influence of behavioral economics, think that bubbles are wholly or even primarily due to human susceptibility to cognitive biases may themselves be deceived by a cognitive bias here.
This understanding of bubbles also implies that attempts to prevent bubbles are futile. Bubbles arise because human cognitive powers are limited. Even if all cognitive biases could be eliminated, as long as people can buy and sell based on their expectations about the future, and as long as the human ability to predict the future is limited, it will sometimes happen that people will be—even perfectly reasonably—mistakenly optimistic for prolonged periods of time. When this happens, the price of a class of assets rises too high—and then crashes when new evidence becomes available and allows people finally to realize their collective mistake. Until we abolish buying and selling under conditions of uncertainty or else become omniscient, from time to time there will be bubbles.
Furthermore, because bubbles arise when there is a prolonged period of reasonable error about changing fundamentals, it is impossible for us to know when we are in a bubble. This perplexes some people, but there is no mystery here. All we are saying is that, while there remain reasonable grounds for thinking that a change in fundamentals justifies higher asset prices, people cannot know for sure that such higher prices are unjustified. While the bubble—that is, the reasonable error—persists, some people will (with more or less evidence to support their view) suspect that we are in a bubble, and some people will be subjectively certain that we are in a bubble (subjectively certain in the sense that such certainty is compatible with actually being wrong). These people will bid down or even sell short the affected assets (thus limiting the size of the bubble and so performing an important public service)—but this is not the same as knowing that we are in bubble. We know we are in a bubble only when the reasonable error about changing fundamentals is dispelled—that is, only when new evidence that the change in fundamentals will not be as great as people thought becomes so conclusive that almost everyone believes that prices have risen too high. At this point, almost everyone sells, which means that asset prices crash. Hence, as soon as we know we are in a bubble, the bubble bursts. In hindsight, then, we can see that prices rose more than was ultimately justified by the change in fundamentals and then crashed again when the true magnitude of the change in fundamentals became apparent. Thus, bubbles are knowable as such only in retrospect.
But the key word here is knowable. Some very astute economists have developed models to detect bubbles, and these have real value. But the thing to understand about them is that, in most such models, the key assumption is that, if certain financial ratios are far off historical norms, then we’re in a bubble. For example, the Shiller P/E is the ratio of the price of the S&P 500 stocks to their ten-year inflation-adjusted earnings. For the period from 1881 to present, the median value of the Shiller P/E was about 15.91. Right before the crash in 1929, it hit what was then an all-time high of about 30, a level that was not seen again until the last days of the tech bubble, when the ratio climbed to 44.20 in December of 1999. Nowadays it stands at about 25.43, which is quite high by historical standards, but probably not dangerously so, especially given the very low interest rates that we justifiably expect to continue for some time to come (lower interest rates generally mean higher stock prices). In any case, the key thing to grasp here is that, even if the Shiller P/E reaches extraordinary heights, this would only suggest, not show, that we’re in a stock market bubble: an alternative explanation would be that changes in fundamentals actually justify the extraordinarily high prices. How plausible that alternative explanation is will depend on the facts of the particular case. As long as most players in the market believe that positive changes in fundamentals justify the higher prices, it is very difficult to say that someone who thinks otherwise—even if he turns out to be right—really knew (as opposed to shrewdly guessed) that we were in a bubble.
TrackBack URL for this entry:
Links to weblogs that reference Is There a Stock Market Bubble?: