Modelling Time-Varying Downside Risk

This paper estimates time-varying systematic downside risk using a parametric specification (BEKK model) and a nonparametric procedure (rolling window technique). A sample of Malaysian industry portfolio daily returns reveals that the covariance between portfolio excess returns and excess downside market returns is persistent and there is a significant difference between the average downside risk estimated in the BEKK model and in the rolling window technique. When the downside risk estimated in the BEKK model is smoothed using moving averages, a positive association between the smoothed series and the downside risk estimated in the rolling window technique is observed. This association gets stronger as the smoothing interval gets closer to the length of the rolling window