Heterogeneous autoregressive model with structural break using nearest neighbor truncation volatility estimators for...

Area of Science:

• Financial econometrics
• Time series analysis
• Volatility modeling

Background:

• High frequency financial data modeling is crucial but challenged by structural breaks causing volatility estimation inconsistencies.
• Existing models struggle with the volatility of financial time series, impacting risk management.

Purpose of the Study:

• To propose a structural break heavy-tailed heterogeneous autoregressive (HAR) volatility model.
• To incorporate jump-robust estimators for enhanced accuracy in financial volatility forecasting.

Main Methods:

• Utilized Bai-Perron sequential multi breakpoints procedure to detect volatility breakpoints.
• Employed nearest neighbor truncation (minimum and median realized volatility) for jump-robust estimators.
• Developed a modified HAR model incorporating structural break dummy variables and jump-robust estimators.

Main Results:

• The modified HAR model demonstrated superior in-sample and out-of-sample forecast performance compared to standard HAR models.
• Improvements in both model structure and volatility estimators led to more accurate forecasts.
• The enhanced model effectively handles structural breaks and abrupt jumps in financial volatility.

Conclusions:

• The proposed structural break heavy-tailed HAR model with jump-robust estimators significantly improves financial volatility forecasting.
• Accurate volatility forecasts are essential for effective risk management and investment portfolio analysis.
• This research provides a more robust econometric tool for analyzing volatile financial markets.