On the way up, bubbles encourage excessive investment in the bubble sector. On the way down a bursting bubble can create wealth shocks, liquidity shortages, and balance-sheet death-spirals. For both of these reasons, it would be good to be able to identify and pop bubbles. Identifying bubbles isn't easy, however, because, especially when interest rates are low, prices can increase rapidly with small, rational changes in investor expectations. But the difficulty of identifying bubbles is reasonably well known. What I think may be less appreciated is that bubbles are hard to pop even when you know that they exist.
In the lab we can create artificial assets with known dividend streams and thus known fundamental values. Since Vernon Smith's classic experiments (JSTOR), we know that even in these cases efficient markets fail and bubbles are common. Bubbles occur even as uncertainty about the fundamental value diminishes (JSTOR). We also know that once a bubble starts it's difficult to stop. Circuit breakers and brokerage fees (transaction taxes), for example, don't do much to stop bubbles (see King, Smith, Williams, and Van Boening 1993, not online.) Investor education doesn't help (for example telling participants about previous bubbles doesn't help). Even increasing interest rates doesn't do much to stop a bubble already in progress and may increase volatility on net.
Bubbles are also less common with more experienced traders – this is one of the strongest findings. Don't get too excited about this, however, it's experience with bubbles that counts not just trading experience. I once asked Vernon, for example, how the lab evidence generalized to the larger economy. In particular, I asked whether 3 bubble experiences in the lab–the number which seems to be necessary to dampen bubbles–might translate to 3 big bubbles in the real world such as the dot com, commodity and housing bubbles (rather than to experience with your run of the mill bubble in an individual stock). He thought that this was a reasonable inference from the evidence. Thus we may not see too many big bubbles during the trading lifetime of current market participants but experience is a very costly teacher. Can we do better?
The last factor that does seem to make a difference is that bubbles liftoff and reach higher peaks when there's a lot of cash floating around. In theory, this shouldn't matter, fundamental value is fundamental value. If an asset is worth $10 in expected value then it's worth $10 whether you have $20 in your pocket or $200. But in practice bubbles are bigger when cash relative to asset value is high.
Note that the latter experiments are consistent with the Fed having a significant role in bubble inflation (a theory I have not pushed). In other words, rather than identifying and popping bubbles already on the rise, not blowing bubbles in the first place may be easier and more productive.