نوع مقاله : مقاله پژوهشی
نویسندگان
گروه مدیریت مالی، دانشکده اقتصاد، مدیریت، و حسابداری، واحد کرج، دانشگاه آزاد اسلامی، کرج، ایران.
چکیده
کلیدواژهها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
This study investigates the performance of five equity exchange-traded funds (ETFs) listed on the Tehran Stock Exchange in forecasting daily volatility using various models based on the Heterogeneous Autoregressive (HAR) framework. The primary objective is to assess the impact of Long Memory (LM), Markov Switching (MS), and Jump (J) components on improving model accuracy. The findings reveal that the baseline HAR model alone has limited explanatory power for volatility, while incorporating advanced components—particularly LM and MS—significantly enhances model performance in most cases. Results indicate that long memory plays a pivotal role in most funds and is the most influential factor in improving forecasts for funds with more stable volatility patterns (e.g., “Asas”), whereas the combination of LM and MS yields superior performance for funds with more variable volatility structures (e.g., “Sarv” and “Atlas”). The jump component shows a limited and fund-specific effect, contributing to improvements only in certain cases—most notably when combined with LM and MS in funds such as “Aghas” and “Karadis.” These findings highlight the importance of selecting flexible, combined models tailored to the behavioral characteristics of each fund and suggest that fund managers should focus on identifying high-risk regimes and long-term volatility patterns for enhanced risk management.
کلیدواژهها [English]
)