Periods of exuberantly increasing asset prices followed by sharp price declines (crashes) are said to have been part of competitive financial markets ever since their inception in late 15th century Antwerp (Schumpeter [1939]). Most accounts of alleged asset price bubbles focus on the detrimental effects of the eventual crash. For example, the housing and credit market bubble that burst in 2007-8 has been claimed to have caused an almost 5% drop in U.S. GDP between the second Quarter of 2008 and 2009, and about a 25% drop in wealth, much of it invested for insurance and retirement purposes. Still, there is no doubt that the economy benefited immensely from the funding available because of the build-up of the bubble. Indeed, it is possible to argue that everyone would have been worse off if the financial securities said to be the cause of the crash had never been allowed in the first place.
But such an argument is difficult to defend when dealing with bubbles that emerge in the field, because we do not have the data to prove the case. That is why experimental economists have long attempted to create bubbles in a controlled environment. One setting where asset prices (in the market for a single asset with stochastic dividend payments and a principal of zero) are often too high and eventually drop back to a rational level was pioneered in Smith et al. [1988]. Since Smith et al., there has been a large series of experimental studies addressing the effects of different design parameters on the magnitude of the pricing bubbles.
It is difficult to make sense of the prices in bubble experiments–it requires that subjects collectively make very large mistakes. It is therefore perhaps not surprising that the bubbles are not robust: the mis-pricing quickly disappears with experience; even if only one-third of the subjects have had prior experience, it fails to re-emerge (Dufwenberg et al. [2005]; re-kindling the bubble requires specific changes in the parameters, see Hussam et al. [2008]).
On a purely theoretical basis, however, arguments can be raised to explain prices. Specifically, it requires risk neutrality to claim that a steady decline is the only viable equilibrium price path. Under risk aversion, many price paths are consistent with equilibrium. This is because markets are generally hugely incomplete. For example, if dividends can take on one of two values, and dividends are paid N times, then there are 2N possible final outcomes, and depending on individual preferences, agents will trade continuously to generate their preferred final wealth; the intermediate trading could create price paths like the ones obtained, with the exception of those price paths for which prices move above the maximum or minimum predicted future dividends. In fact, only the latter provides unequivocal bounds on rational asset prices (Bossaerts [2009]).
