New research on rational bubbles from George Waters:
I’d like to take you up on your offer to publicize research. I’ve spent a good chunk of my time (along with Bill Parke) over the last decade developing an asset price model with heterogeneous expectations, where agents are allowed to adopt a forecast based on a rational bubble.
The idea of a rational bubble has been around for quite a while, but there has been little effort to explain how investors would coordinate on such a forecast when there is a perfectly good alternative forecast based on fundamentals. In our model agents are not assumed to use either forecast but are allowed to switch between forecasting strategies based on past performance, according to an evolutionary game theory dynamic.
The primary theoretical point is to provide conditions where agents coordinate on the fundamental forecast in accordance with the strong version of the efficient markets hypothesis. However, it is quite possible that agents do not always coordinate on the fundamental forecast, and there are periods of time when a significant fraction of agents adopt a bubble forecast. There are obvious implications about assuming a unique rational expectation.
A more practical goal is to model the endogenous formation and collapse of bubbles. Bubbles form when there is a fortuitous correlation between some extraneous information and the fundamentals, and agents are sufficiently aggressive about switching to better performing strategies. Bubbles always collapse due to the presence of a small fraction of agents who do not abandon fundamentals, and the presence of a reflective forecast, a weighted average of the other two forecasts, that is the rational forecast in the presence of heterogeneity.
There are strong empirical implications. The asset price is not forecastable, so the weak version of the efficient markets hypothesis is satisfied. Simulated data from the model shows excess persistence and variance in the asset price and ARCH effects and long memory in the returns.
There is much more work to be done to connect the approach to the literature on the empirical detection of bubbles, and to develop models with dynamic switching between heterogeneous strategies in more sophisticated macro models.
A theoretical examination of the model is forthcoming in Macroeconomic Dynamics.
A more user friendly exposition of the model and the empirical implications is here.
An older published paper (Journal of Economic Dynamics and Control 31(7)) focuses on ARCH effects and long memory.
Dr. George Waters
Associate Professor of Economics
Illinois State University