Your are here: Home» Publications» Working paper


Do High-Frequency Data Improve Multivariate Volatility Forecasting for Investors with Different Investment Horizons?


E2022018                                                         2022-10-25

Limin Yu                     Zhuo Huang

China Center for Economic Research

National School of Development

Peking University


Abstract This study investigates the role of high-frequency data in multivariate volatility forecasting for investors with different investment horizons. We use six multivariate volatility models with high-frequency and low-frequency data for a sample of 10 Dow Jones stocks and evaluate the performance of forecast volatility based on both statistical and economic methods. In our statistical evaluation, we find that high-frequency data significantly enhance forecast accuracy over the daily horizon, but this improvement is dampened when longer horizons are used. In our economic evaluation, we find that high-frequency data cannot improve all economic benefits under the short and long horizons. The economic benefits of using high-frequency data depend on the evaluation framework.

Key Words: Realized Volatility, Covariance Matrix Forecasting, Investment Horizon, Statistical and Economic Evaluation