Scholarlink

<p>This study investigates volatility interdependencies in Brazil’s equity market using Factor-Adjusted Networks (FNETS), which integrate factor models with sparse network estimation to disentangle common market effects from firm-specific volatility linkages. The networks are constructed from latent volatilities estimated via Stochastic Volatility (SV) models using Bayesian inference through Integrated Nested Laplace Approximations (INLA), which provide accurate and computationally efficient posterior approximations for high-dimensional models. Focusing on firms within the Bovespa Theoretical Portfolio, our results reveal distinct structures: core stocks with higher portfolio weights exhibit strong systemic linkages, whereas peripheral firms show weaker connections. Methodologically, FNETS capture Granger-causal, contemporaneous, and long-term dependencies, while SV models outperform traditional OHLC/HL and GARCH(1,1) volatility measures, producing lower forecasting errors. These findings contribute to enhanced systemic risk monitoring and offer practical insights for policymakers and investors in emerging markets.</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

Learning volatility structures in emerging markets: a data-driven network approach using stochastic volatility models

  • 期刊论文

  • 发表日期：

    2025年11月28日

查看全文
  • João Pedro M. Franco,
  • Márcio P. Laurini

摘要

This study investigates volatility interdependencies in Brazil’s equity market using Factor-Adjusted Networks (FNETS), which integrate factor models with sparse network estimation to disentangle common market effects from firm-specific volatility linkages. The networks are constructed from latent volatilities estimated via Stochastic Volatility (SV) models using Bayesian inference through Integrated Nested Laplace Approximations (INLA), which provide accurate and computationally efficient posterior approximations for high-dimensional models. Focusing on firms within the Bovespa Theoretical Portfolio, our results reveal distinct structures: core stocks with higher portfolio weights exhibit strong systemic linkages, whereas peripheral firms show weaker connections. Methodologically, FNETS capture Granger-causal, contemporaneous, and long-term dependencies, while SV models outperform traditional OHLC/HL and GARCH(1,1) volatility measures, producing lower forecasting errors. These findings contribute to enhanced systemic risk monitoring and offer practical insights for policymakers and investors in emerging markets.