<![CDATA[This study examines the systemic risk contributions of four major Indonesian state-owned banks, specifically BCA, BRI, Bank Mandiri, and BNI, to the Jakarta Composite Index (IHSG) using a Conditional Value-at-Risk (CoVaR) framework estimated through a Hybrid LASSO-QRNN approach. The study employs 2,491 daily trading observations from November 2016 to June 2026, sourced from Yahoo Finance, Bank Indonesia, and Investing.com, with out-of-sample evaluation on 375 observations from December 2024 to June 2026. Following the hybrid framework of Syalsabila et al. (2024), Step 1 applies LASSO-Quantile Regression (LASSO-QR) to select macroeconomic contagion amplifiers at quantiles q = 0.05 and q = 0.01, corresponding to 95% and 99% confidence levels respectively, while Step 2 trains a Quantile Regression Neural Network (QRNN) on the selected features to estimate conditional quantiles of system-level returns. The results reveal that BNI is the most systemically important institution, with mean ?CoVaR of -0.920% at 95% confidence and -1.991% at 99% confidence, followed by BRI, Bank Mandiri, and BCA. LASSO-QR retains all five macroeconomic variables in the system-level model, contrasting with the sparse two-to-five variable selection at the institution level. The findings further document BRI, Mandiri, and BNI each show a statistically significant attenuation of ?CoVaR during the 2026 domestic confidence crisisss three banks, consistent with the theoretical prediction that investor confidence deterioration compresses idiosyncratic institution-level risk contributions when market-wide confidence collapses.]]>
Copyrights © 2026
