A NOVEL PERSPECTIVE ON PORTFOLIO DIVERSIFICATION THROUGH THE QQKRLS METHOD: EVIDENCE FROM TURKEY AND BRICS COUNTRIES
Anadolu Üniversitesi İktisadi ve İdari Bilimler Fakültesi Dergisi, cilt.27, sa.2, ss.743-770, 2026 (TRDizin)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 27 Sayı: 2
- Basım Tarihi: 2026
- Doi Numarası: 10.53443/anadoluibfd.1752605
- Dergi Adı: Anadolu Üniversitesi İktisadi ve İdari Bilimler Fakültesi Dergisi
- Derginin Tarandığı İndeksler: TR DİZİN (ULAKBİM)
- Sayfa Sayıları: ss.743-770
- Açık Arşiv Koleksiyonu: AVESİS Açık Erişim Koleksiyonu
- Maltepe Üniversitesi Adresli: Evet
Özet
This study provides a dynamic perspective on portfolio diversification by investigating the nonlinear, quantile-level interactions between the Turkish stock market (BIST100) and the stock indices of BRICS countries. Unlike traditional linear frameworks, the research employs the advanced Quantile-on-Quantile Kernel-Based Regularized Least Squares (QQKRLS) method to capture regime-dependent dependencies across the entire distribution. The empirical findings reveal that correlations fluctuate significantly based on market conditions, with synchronization typically strengthening during bullish phases (upper quantiles) and weakening or turning negative during bearish periods (lower quantiles). Specifically, the QQKRLS analysis identifies the South African (JSE) and Indian (BSE) markets as robust diversification instruments, exhibiting negative tail dependencies that offer critical hedging potential during Turkish market distress. In contrast, the Brazilian (IBOVESPA) and Russian (MOEX) stock markets demonstrate heightened synchronization during extreme growth regimes, while showing independence or negative correlations during bearish trends. Furthermore, the Shanghai Stock Exchange (SSE) presents conditional diversification benefits, particularly in mid-level quantiles. To validate these findings, a standard Quantile-on-Quantile Regression (QQR) analysis was conducted as a robustness check, confirming that the identified state-dependent interactions are qualitatively consistent and robust across different quantile-based estimation frameworks. Finally, an out-of-sample backtest demonstrates that a dynamic, QQKRLS-informed strategy yields superior risk-adjusted performance compared to a standalone BIST100 benchmark. These results underscore the necessity for investors to transition from static allocation models to dynamic, state-dependent diversification strategies to enhance stability in volatile emerging markets.