.
.
Based on our success in studying static (one-period) asset pricing theory in the laboratory, we have now embarked on an ambitious agenda investigating dynamic asset pricing theory. This is particularly challenging, because the standard model here (the "Lucas" model) has everyone living forever, and assumes the world is stationary. (There are a few other challenges to bring this model to the lab!) We are not only interested in price dynamics, but also in allocation dynamics. The standard model assumes that markets somehow reach an optimal allocation, but leaves out the details. This too poses a formidable problem. In collaboration with Elena Asparouhova and Bill Zame, we have obtained some remarkably promising results. One could say that the standard model is very much "alive." Our prices are excessively volatile, though. But that is not unlike in the real world, and our experimental approach actually allows us to explore potential causes in a much more effective way than using historical field data. Despite excessively volatile prices, allocations are on average close to optimal... Our setup allows us to also look at some foundational aspects of dynamic asset pricing theory, such as "dynamic completeness" (the ability to create any payoff using sophisticated dynamic trading schemes) – which we have started to focus on in separate experiments, in collaboration with Debrah Meloso.
Another important observation from our static experiments, which we replicated in the dynamic experiments, is that the system behaves in a way that is unlike each of its components. Specifically, you cannot take any subject and use his or her choices to "explain" or "rationalize" prices (and allocations) at the market level. E.g., a certain fraction of the population is known to be ambiguity averse. This does not imply that prices behave as if there is a representative agent with ambiguity-averse preferences. (Nor does it imply that they behave as if driven by the complement, namely, agents with ambiguity-insensitive preferences, or even by some average of the two...) This should caution theorists who build asset pricing models that are based on extrapolating preferences observed at the individual level to that of the "representative investor."