Scholarlink

<p>This study proposes a Bayesian semi-parametric framework for forecasting Value-at-Risk (VaR) and expected shortfall, with emphasis on tail risk dynamics driven by forward-looking geopolitical uncertainty. Narrative-based information is incorporated through the Geopolitical Threats Index (GPT), while market variability is captured using range-based volatility measures that provide stable high-frequency inputs. The model accommodates nonlinear and asymmetric dynamics in both the quantile and expected shortfall processes and introduces a threshold extension to capture regime-dependent tail behavior. Five competing model specifications are estimated within a unified Bayesian setting, enabling full probabilistic inference and coherent uncertainty quantification. Empirical analyses for Bitcoin and Brent crude oil futures evaluate predictive performance using loss-based scoring rules and regulatory backtests. The results show that incorporating geopolitical signals and threshold dynamics leads to substantial improvements in tail risk forecasting.</p>

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

Bayesian Tail Risk Forecasting with Geopolitical Narratives and Range-Based Volatility

  • 期刊论文

  • 发表日期：

    2026年1月17日

查看全文
  • Cathy W. S. Chen,
  • Aaron E. P. Lin,
  • Manabu Asai

摘要

This study proposes a Bayesian semi-parametric framework for forecasting Value-at-Risk (VaR) and expected shortfall, with emphasis on tail risk dynamics driven by forward-looking geopolitical uncertainty. Narrative-based information is incorporated through the Geopolitical Threats Index (GPT), while market variability is captured using range-based volatility measures that provide stable high-frequency inputs. The model accommodates nonlinear and asymmetric dynamics in both the quantile and expected shortfall processes and introduces a threshold extension to capture regime-dependent tail behavior. Five competing model specifications are estimated within a unified Bayesian setting, enabling full probabilistic inference and coherent uncertainty quantification. Empirical analyses for Bitcoin and Brent crude oil futures evaluate predictive performance using loss-based scoring rules and regulatory backtests. The results show that incorporating geopolitical signals and threshold dynamics leads to substantial improvements in tail risk forecasting.