Dependence Structures in Financial Time Series with High dimensional Vine Copula Models

The recent financial crisis has emphasized the difficulty of modeling dependence structures among multiple financial time series. This study investigates sectoral dependence among major industries in the Tehran Stock Exchange using daily return data for the Chemicals, Automobiles, Cement, Pharmaceuticals, Investments, Steel, Petroleum, and Metallic Minerals indices. To improve risk assessment beyond traditional correlation-based measures, high-dimensional Vine Copula models are employed. For the marginal distributions, all series are modeled using an ARMA (1,1) process for the conditional mean and an EGARCH (1,1) specification for volatility, with standardized Student-t innovations to capture heavy tails and asymmetry. Using data from 2020 to 2024 and implementing the analysis in R software, these marginal models are combined with C-Vine, D-Vine, and R-Vine Copulas to capture nonlinear and tail dependence. The results show that student-t and Copulas dominate at the pairwise level, while the Investment sector emerges as a central node with strong dependence links to the Steel, Petroleum, and Chemical industries. A comparative analysis indicates that the R-Vine Copula provides the most flexible and accurate representation of high-dimensional dependence, outperforming the more restrictive C-Vine and D-Vine structures. These findings offer valuable insights for portfolio diversification and systemic risk management in emerging financial markets.