The Hindsight Fallacy

The real reason it’s so hard to predict bubbles.

The recent sky-high IPO of LinkedIn, along with eye-popping valuations for other social networking and shopping companies, has raised concerns that we are now in the midst of another technology bubble, this one fueled by excessive investor enthusiasm for all things social. 

No sooner have these concerns been raised, however, than they have been countered by an array of arguments, all of which are variations on the basic claim that this internet boom is unlike the previous one. This debate illustrates one of the central causes of financial bubbles: Although after the fact it seems obvious that prices were irrational and an unhappy end was inevitable, bubbles are neither obvious nor inevitable at the time.

The fact is that financial bubbles throughout history display almost metronomic regularity. Some asset class, often a product of new technology or financial innovation, becomes fashionable, and prices rise rapidly as demand outstrips supply. Excitement over early riches leads to speculation that the normal economic rules have changed, thereby justifying valuations that are based largely on assumptions about future profits. As prices balloon well beyond early investors’ wildest dreams, disagreement over the correct measure of value increases. Then, finally, some adverse event that might normally have little impact triggers a downward spiral of selling, often accompanied by outright panic. 

In hindsight, this pattern is so familiar that it’s hard to believe anyone could have misunderstood what was happening—witness the now-ridiculous-seeming valuations of the 1990s Internet start-ups, and more recently the clearly misguided assumptions about perpetual growth in real estate prices that underpinned the 2008 financial crisis. 

But if bubbles are so obvious, why do they keep happening?

One explanation is that fear and greed are unchanging and inextricable components of human nature—and that no matter how many times we suffer the consequences of financial crises, there will always be a pool of people so blinded by their greed that they cannot see the folly of their ways until it’s too late. 

Another view is the “greater fool than thou” hypothesis, which casts financial bubbles as giant Ponzi schemes. In this view, nobody really believes the skyrocketing valuations, but everything thinks they can sell to someone else who’s prepared to pay even more than they did, on the grounds that they, too, can resell to someone else, and so on. … Eventually, the chain of fools runs out, at which point the whole thing crashes.

Both these descriptions no doubt capture certain aspects of bubbles, but unfortunately neither is much help in spotting one in the making.

The reason is that in the heat of the moment, there is no way to be sure that the explanations being proffered to justify the valuations are wrong. Some start-ups, after all, do go on to become astronomically valuable companies in ways that would have been hard to imagine at the time of their IPOs. 

With Google’s stock at over $500, for example, its IPO price of $85 seems like an absolute bargain.  But many observers at the time thought it was overvalued.  Why? Because at the time, it was inconceivable to just about everyone how much money could be generated by search.  How are we to know that LinkedIn or Facebook or Groupon won’t also go to generate enormous revenues from business models that have yet to be invented?

In fact, we don’t.  Yes, the price to earnings ratio for LinkedIn appears high by historical standards.  But that’s only relevant if its earnings potential also conforms to historical patterns.  If LinkedIn turns out to be an important player in a whole new economy, then its future earnings may be so great that its current price will seem amply justified.  In other words, whether the investors paying top dollar now are visionaries or suckers depends on events that have yet to transpire and that, probably, the founders of the companies themselves have yet to imagine. 

There is really no getting around this problem, which is actually a problem with how we learn from history.

We all know, of course, that “hindsight is 20-20,” but when we say this we’re implying that our failure to predict the outcome resulted simply from not paying attention to relevant information.  After a bubble has burst, that is, we can always find signs, and very often people, pointing out that it was a bubble; thus it always seems that investors could have known what was happening and just chose to ignore it.

But this view of history is actually deeply misleading.  The real benefit of hindsight is that we only invoke it once we know the ending. And at that point, we don’t merely have more information. We also know which of all the information we do have that fits into the story we want to tell—a story that has to end the way we now know it did. 

By analogy, when you’re reading a mystery novel, you don’t necessarily know what to make of events because you don’t know how the story ends. But the novelist knows and is carefully including or excluding characters, scenes, and actions, all based on his or her intentions for the way the various pieces are going to fit together at the end. 

History is told the same way. Historical explanations can only be constructed in retrospect because only once we know the outcome can we decide what to emphasize and what to ignore. If the envisioned social network economy fails to materialize, financial historians will cluck-cluck at all the warning signs that investors overlooked to drive up LinkedIn’s stock; and if it does materialize, those very same actions will later be construed as bold and insightful.

Right now, however, investors don’t have the benefit of knowing which of these future states of the world will be realized. For sure, they can imagine plausible endings, but everything we know about complex social systems tells us that outcomes are in large part random.  Thus many endings are possible, and it’s impossible to know, even in principle, which one of them will transpire.

For the same reason, it’s impossible to know, even in principle, which information to pay attention to—because what is relevant now will only be decided by some future historian, who, like the novelist, can decide what was most relevant in the light of the ending that he or she knows about but that the participants themselves do not.

This is why, when we look back on bubbles, it seems to us that the explanation we have constructed is what really was happening and that investors at the time were falling for some contrived view of the world. Really, it’s the opposite: The confusion we experience in the present is real.  It is the story that we tell afterwards that is contrived.