High-dimensional Multivariate Realized Volatility Forecasting with Community Network Structure
Published in Journal of Business & Economic Statistics (accepted author version), 2026
We introduce the Community Network Heterogeneous Autoregressive model with Quarticity (CNHARQ), which integrates network-based information to enhance multivariate realized volatility forecasting. A Correlation-Based Stochastic Block Model (CBSBM) uncovers latent community structures, reducing parameter dimensionality and improving out-of-sample forecast accuracy over static industry classifications.
Recommended citation: He, Sixuan, and Cheng Liu. (2026). "High-dimensional Multivariate Realized Volatility Forecasting with Community Network Structure." Journal of Business & Economic Statistics. (accepted author version, posted online February 18, 2026) https://doi.org/10.1080/07350015.2026.2632810
