A Dynamic Structural Model for Stock Return Volatility and Trading Volume.

NBER Working Paper No. 4988 January 1995 Asset Pricing

We develop a structural model that uses data on asset returns and trading volume to determine whether the autocorrelation in volatility comes from the fundamental being priced by the trading process or from the trading process itself. In the context of our model, the data suggest that persistent volatility is caused by traders experimenting with different belief systems that are based on both past and estimated future profits.

We introduce adaptive agents, in the spirit of Sargent (1993), whose strategies adapt more slowly than the trading process itself. This leads to positive autocorrelation in volatility and volume during the trading process; the positive autocorrelation is caused by persistence of strategy patterns that are associated with high volatility and high volume.

At a rough level, our model is able to reproduce qualitatively the following features of the data...