• by Erik Sass , Staff Writer @eriksass1, June 7, 2011

Having harped on the whole social media bubble thing in the past, including the LinkedIn IPO, I now find myself playing devil's advocate in response to an interesting study by academic mathematicians purporting to "prove" that LinkedIn's IPO is a bubble. Basically, I am merely going to point out that -- even if it looks, smells, sounds, and acts like a bubble -- there is really no way to actually know if it is a bubble until it pops.

A Forbes blog post highlighted the study by Cornell's Robert Jarrow, Columbia's Philip Protter, and the Ecole Polytechnique's Younes Kchia, who claim that the huge run up in LinkedIn's stock price -- jumping from $45 to over $100 on the first day of trading -- reflects a classic bubble dynamic. According to Forbes, their methodology is based on studies of numerous other stock price movements, including "generally accepted stock price evolution, along with its expected volatility ranges, and an extrapolation as to how that volatility should behave."

While there is certainly reason to be suspicious of LinkedIn's exuberant performance -- I have voiced my own reservations about LinkedIn in several previous posts -- I would also argue that the mathematicians are promising more than they can realistically deliver with their claim to be able to discover bubbles in real time, before they pop.

It would be great if stock prices behaved like, say, physical phenomena, determined by underlying rules which can be uncovered and described as natural laws. However this is clearly not the case because of the element of human psychology, with its mix of rational and irrational thinking -- difficult enough to understand at the individual level, and far too complex to describe (let alone understand) when multiplied by millions of individuals. Most attempts to reduce mass human behavior to quantifiable patterns inevitably come up short: that's why "social sciences" have never enjoyed the same authoritative weight of natural sciences, especially in terms of predictive value. At best they can offer retrospective explanations of human behavior; when it comes to forecasting the future, it's a whole ‘nother ball game.

Take for example "herd mentality" -- an epithet that's often used to describe behavior in economic bubbles. The concept implies a certain unthinking willingness to do something "just because" lots of other people are doing it, and this, in turn, is supposedly proof of irrational behavior. But "herd mentality" can actually result in rational or at least productive behavior by the individual: real herd animals behave the way they do (sticking together, stampeding, etc.) for very good reasons -- namely, because they are broadly beneficial to the individuals who compose the group. Likewise, doing something "just because" lots of other people are doing it may actually be a great idea, depending on the context.

Turning to concrete examples, there are plenty of instances where stock prices rose suddenly in a generally exuberant atmosphere -- then proved to be quite reasonable in the long term. When Google went public in August 2004 the share prices immediately jumped 17% from $85 to over $100 on the first day of trading, and a week later it was $108, up 27% from the original offering. By the end of the year it had increased 126.8% to $192.79, and in 2005 it increased 115.2% to $414.86 -- and this was actually a modest increase next to some analysts' predictions, which put the target price as high as $2,000!

Subsequent increases have been smaller and more gradual, but they are certainly nothing to sneeze at: in June 2011 Google's stock is trading at around $520. Yes, the company has had its ups and downs; recently, for example, its leadership seems nervous about the rise of social media. But I don't think anyone would argue that investors who got the stock at $85-$100 a share during the IPO and held on to it over the last seven years made a bad decision, or fell victim to "bubbly" thinking.

Of course, LinkedIn isn't Google, operating a very different business model (and on a rather smaller scale, at least right now). And it may well turn out to be over-valued. My point is merely that we won't know until the bubble actually pops, because massive increases in stock valuations and price volatility don't necessarily indicate a bubble in and of themselves. If you could really predict these things with mathematical models, bubbles would never happen in the first place.