| Summary: | 碩士 === 國立中央大學 === 財務金融研究所 === 98 === Generalizing the component GARCH by Engle and Rangel (2008), this paper proposes a new modeling and forecasting strategy for systemic risk both in the short term and long run. By utilizing the orthogonally decomposed stationary regularity series from real quarterly GDP and CPI by EMD (Empirical Mode Decomposition), an empirical adaptive decomposition method that entertains nonlinear and nonstationary time series, we demonstrate the close coupling relationship between long run stock market volatility and the business cycle fluctuations. As these component series preserve the most primary information in the macroeconomic state variables sampled at lower frequencies, the long run component volatility is capable of generating regime shift behaviors in daily volatility without resorting to Markov switching or other regime switching mechanisms. Moreover, prediction of future volatility at various horizons is easy within the framework by taking advantage of the stable cyclical pattern of these orthogonalized macro series. Our empirical applications in hedging and evaluating VaR reveals that incorporating information from lower frequency macroeconomic fundamentals did provide incremental value toward the modeling of long run risks.
|
|---|
