Do bubbles and time-varying risk premiums affect stock prices? a Kalman filter approach

This paper separates the validity of the specification of the fundamental stock price model from the implications of bubbles. The time-varying risk premium model (Poterba and Summers, 1986) is used to explicitly derive the misspecification component. We construct a state-space model and use Kalman filter to estimate the relationships between the observable price/dividend and the unobservable bubbles/misspecification. The model is applied to CRSP and S&P 500 data. The results show that the fundamental price model does not describe the market prices well. The time-varying risk premium is important in explaining stock price movements. No significant evidence of bubbles is found.