Endogenous regime switching volatility model with time-varying threshold

다국어 초록 (Multilingual Abstract)

This paper extends endogenous regime switching volatility models by incorporating a predetermined and observable factor, which determines the regimes with an endogenous autoregressive latent factor. The bivalued state process is dependent upon whether the latent factor takes a value above or below time-varying threshold which is a linear function of the predetermined and observable factor. By allowing the threshold to vary over time, the state process becomes partially observable to econometrician as it is determined by both latent and observable factors. Using this property, I can extract the information from the observed time series more effectively and find that the presence of state observability and endogeneity feedback effect should be considered together when estimating volatility models and conducting state inference.

목차 (Table of Contents)

• I. Introduction 4
• II. Regime Switching Models with Time-Varying Threshold 8
• 2.1 Models with Endogenous Regime Switching 8
• 2.2 Relationship with Previous Markov Switching Models 12
• III. Estimation 17
• IV. Empirical Illustrations 24
• V. Simulations 35
• 5.1 Simulation Models 36
• 5.2 Simulation Results 38
• VI. Conclusion 43