نوع مقاله : مقاله علمی پژوهشی
نویسندگان
1 دانشیار، گروه مالی و بانکداری، دانشکدۀ مدیریت و حسابداری، دانشگاه علامه طباطبائی، تهران، ایران.
2 استادیار، گروه مالی و بانکداری، دانشکدۀ مدیریت و حسابداری، دانشگاه علامه طباطبائی، تهران، ایران.
3 دانشجوی دکتری، گروه مدیریت مالی، پردیس تحصیلات تکمیلی، دانشگاه علامه طباطبائی، تهران، ایران
چکیده
کلیدواژهها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
Objective
In the field of stock portfolio optimization, various risk measures have been used under the title of "classic risk measures" in previous studies. The correct estimation of portfolio risk and its management plays a crucial role in preserving investors' wealth. Failure to accurately estimate portfolio risk has been one of the major causes of financial crises in recent decades. According to prospect theory, investors exhibit a heightened sensitivity to losses and thus assign more weight to them. This leads to the underestimation of risk by common risk measures, such as Value at Risk, Expected Shortfall, Variance, and others. This study examines stock portfolio optimization with a behavioral finance approach under loss aversion conditions in the Tehran Stock Exchange, comparing the performance of this approach with 11 other risk measures in the classic financial field by considering different time horizons (short-term, medium-term, and long-term) and portfolios of varying sizes (small, medium, and large).
Methods
The performance of the risk measure based on loss aversion, developed from Fulga's (2016) model, is evaluated alongside 11 other classic risk criteria in portfolio optimization, using daily data from 178 stock companies. Non-parametric tests (Kruskal-Wallis test and one-sided Wilcoxon test) are applied for evaluation. Instead of limiting the comparison to portfolio returns and risk, this study utilizes 26 different criteria for a comprehensive performance comparison. Historical simulation methods are used to calculate the risk measures. The reason for using this method is twofold: first, all the risk measures used in this research can be calculated using this approach; and second, since this method makes no assumptions about data distribution, it is superior to other risk calculation methods. Fulga's model, based on the downside of portfolio return distributions, specifically simulates loss-averse behavior and offers a more accurate risk evaluation, incorporating criteria like Conditional Value at Risk (CVaR) and the level of negative deviation from a reference point (LPM).
Results
Depending on the threshold value or reference point in prospect theory, the performance of the loss aversion-based risk measure can vary. When the threshold is zero or greater (e.g., 0%, +2%, and +4%), the loss aversion-based criterion demonstrates significantly better performance in portfolio optimization compared to other risk measures. However, when the threshold value is negative (e.g., -2% and -4%), no particular advantage is observed over other criteria.
Conclusion
In situations where it is possible to estimate parameters for the loss aversion coefficient and the threshold value, the performance of the behaviorally-based risk measure (under conditions of loss aversion) in portfolio optimization proves to be superior to other classical financial measures in real market conditions (where the threshold value is zero or higher). Additionally, as the threshold value increases above zero, the results of this study indicate that the "standard deviation" measure demonstrates better performance compared to other classical risk measures.
کلیدواژهها [English]
)