A1836
Title: A novel Markov-switching GAS jump-in-mean model: State-dependent dynamics and risk heterogeneity in U.S. stock indices
Authors: Kuang-Liang Chang - National Sun Yat-sen University (Taiwan) [presenting]
Abstract: A novel Markov-switching GAS Jump-in-Mean model is proposed that integrates a Markov-switching Generalized Autoregressive Score (GAS) process into both the conditional variance of non-jump innovations and the conditional jump intensity. The framework is employed to investigate the state-dependent effects of non-jump risk and jump risk on excess stock returns for the S&P 500, DJIA, and Nasdaq indices. Furthermore, the analysis examines how macroeconomic and financial factors -including economic policy uncertainty (EPU), the VIX, and the U.S. dollar index- influence conditional volatility across different volatility states. The empirical results indicate that: (i) excess stock returns display pronounced Markov-switching and jump dynamics; (ii) the effects of non-jump risk and jump risk are state-dependent and heterogeneous across stock indices; (iii) the contribution of non-jump innovations to conditional variance is stronger in the high-volatility state for all indices, while the contribution of jump innovations to conditional variance is greater in the low-volatility state for the S&P 500 and Nasdaq; and (iv) stock market volatility responds differently to macroeconomic and financial variables across indices, depending on volatility states and the underlying sources of innovation.
