Estimation of Market Risk Premium for Japan

Abstract

In this paper, the time series of risk aversion parameter is estimated for the Japanese stock market using weekly return data covering 2/7/1973 to 12/27/2000. The time series of risk aversion parameter is estimated with the Time Varying Parameter (EVP) GARCH-M model proposed by Chou, Engle and Kane (1992), which allows for the risk aversion parameter to change over time by modeling the risk aversion parameter to follow a random walk process. The risk aversion parameter is found to range between 3.5 to 2.2. We also find that the risk aversion parameter has not significantly changed over time. This implies that most of the variation in excess return can be explained by the variation in the market (variance) risk. Keywords: GARCH-M, Kalman Filtering, risk aversion, time-varying parameter, volatility.