Title: Stock/bond volatility/correlation on macro factors in China: Based on GARCH-MIDAS
Authors: Qian Chen - Peking University Shenzhen Campus (China) [presenting]
Chen Chen - Peking University Shenzhen Campus (China)
Abstract: The GARCH-MIDAS-X models are applied to China stock and bond market in attempt to examine the power of low-frequency macro factors in predicting high-frequency market volatility. The results confirm the significant relationship between the macro variables and the long run volatility. Specifically, the long-run component of GARCH-MIDAS model incorporating industrial added value growth rate (IP) accounts for around 30\% of total conditional volatility of Chinas stock market and bond market. The study also finds that, industrial added value is a better predictor than the producer price index, which may be due to the fact that the China economy is still in the development stage and the market is more sensitive to economic growth than inflation. The out-of-sample forecasts of GARCH-MIDAS-X models improve with longer horizons. Though for the stock market, GARCH-MIDAS-RV still performs best in semi-annual horizon; for bond, GARCH-MIDAS with IP volatility outperforms other models in semi-annual horizon. DCC-MIDAS-X is also applied to study the relationship between the macro factors and the stockbond correlation. The results suggest a weaker effect of the macro factors, which may be due to the absence of inter-market macro-strategy investors in China.