Scholarlink

The Value-at-Risk (VaR) measure represents the probability of the maximum individual risk of loss that can occur at a certain level of confidence. This study calculates VaR using the Extreme Value Theory (EVT) method with the Peaks Over Threshold (POT) approach. Another alternative risk measure is used to show dependence risk, namely conditional VaR (CoVaR). The CoVaR can detect systematic risk spillovers and provide information about the VaR of one entity that depends on the VaR of other entities. The CoVaR model is developed from a Quantile Regression (QR) model that linearly relates risk. In this study, the VaR and CoVaR values for the stock of Indonesian public banks, i.e., BBCA, BBNI, BBRI, BMRI, and ARTO, are calculated, and the performances of both are compared. The results obtained are VaR calculated using the EVT method, which does not provide good results because only 3 out of 15 VaR are valid based on the Kupiec test. Meanwhile, CoVaR has better results because all 15 CoVaR are valid based on the Kupiec test. When a comparison was made based on the exceedance’s ratio, CoVaR had a value close to the level of the theoretical quantile, so it can be concluded that CoVaR outperforms VaR.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Performance Comparison of Individual Risk Model Using VaR-POT with Its Risk Dependence Model Using Quantile Regression-Based CoVaR

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Hasri Wiji Aqsari,
  • Dedy Dwi Prastyo,
  • Muhammad Sjahid Akbar

摘要

The Value-at-Risk (VaR) measure represents the probability of the maximum individual risk of loss that can occur at a certain level of confidence. This study calculates VaR using the Extreme Value Theory (EVT) method with the Peaks Over Threshold (POT) approach. Another alternative risk measure is used to show dependence risk, namely conditional VaR (CoVaR). The CoVaR can detect systematic risk spillovers and provide information about the VaR of one entity that depends on the VaR of other entities. The CoVaR model is developed from a Quantile Regression (QR) model that linearly relates risk. In this study, the VaR and CoVaR values for the stock of Indonesian public banks, i.e., BBCA, BBNI, BBRI, BMRI, and ARTO, are calculated, and the performances of both are compared. The results obtained are VaR calculated using the EVT method, which does not provide good results because only 3 out of 15 VaR are valid based on the Kupiec test. Meanwhile, CoVaR has better results because all 15 CoVaR are valid based on the Kupiec test. When a comparison was made based on the exceedance’s ratio, CoVaR had a value close to the level of the theoretical quantile, so it can be concluded that CoVaR outperforms VaR.