Factor Garch, Regime-Switching and the Term Structure of Interest Rates

摘要: The presence of time-changing variance (heteroskedasticity) in financial timeseries is often cited as the cause fat-tailedness unconditional distribution series. However, many researchers have found that, after allowing for heteroskedastic behaviour, conditional distributions remain fat-tailed. Consequently, one approach adopted by applied econometricians has been to postulate a fat-tailed distribution. In multivariate context, few such offer tractable solutions which accurately capture deviations from normality. taken this paper model dynamics covariance matrix with parsimonious regime-switching factor GARCH model. loading switches within finite state-space according value an unobserved Markov state variable. process then mixture normals. Fat-tails are explicitly generated structural breaks or changes regime. We develop some theoretical properties models, and filters inference about chain, well maximum likelihood estimation via EM algorithm. fit our daily term structure US interest rates apply one-step ahead distributional forecasts simple portfolio risk management context.