Central Tehran Branch, Islamic Azad University, Tehran, Iran
10.22034/jsmta.2026.24972.1234
Abstract
This paper develops a dynamic extension of the principal component copula framework for high-dimensional systemic risk monitoring. The proposed model lets the eigenvalues of the copula dependence matrix evolve over time through a score-driven dynamic specification, following a generalized autoregressive score framework for observation-driven parameter updating. We derive tractable expressions for the high-dimensional copula density and develop an efficient two-stage estimation procedure combining marginal volatility filtering with pseudo-maximum likelihood estimation, together with a parametric bootstrap for the estimation error that propagates between the two stages. The model facilitates the computation of systemic risk measures in high-dimensional settings, with explicit algorithms given under the Student's t copula. An empirical application to fifty global systemically important banks shows that the proposed framework significantly outperforms static principal component copulas and several alternative dynamic dependence models in capturing time-varying tail risk spillovers, confirmed by formal statistical tests. During periods of financial stress, the model reveals sharp increases in systemic risk contributions, reported with bootstrapped confidence intervals, that are undetected by static alternatives. We discuss the economic significance of these findings for capital allocation and early warning systems, along with the model's limitations, including the assumption of constant eigenvectors and the tractability-versus-tail-dependence trade-off across copula families.
Ghajari, A. (2026). Dynamic principal component copulas with time-varying tail dependence for high-dimensional systemic risk monitoring. Journal of Statistical Modelling: Theory and Applications, (), 1-27. doi: 10.22034/jsmta.2026.24972.1234
MLA
Ghajari, A. . "Dynamic principal component copulas with time-varying tail dependence for high-dimensional systemic risk monitoring", Journal of Statistical Modelling: Theory and Applications, , , 2026, 1-27. doi: 10.22034/jsmta.2026.24972.1234
HARVARD
Ghajari, A. (2026). 'Dynamic principal component copulas with time-varying tail dependence for high-dimensional systemic risk monitoring', Journal of Statistical Modelling: Theory and Applications, (), pp. 1-27. doi: 10.22034/jsmta.2026.24972.1234
CHICAGO
A. Ghajari, "Dynamic principal component copulas with time-varying tail dependence for high-dimensional systemic risk monitoring," Journal of Statistical Modelling: Theory and Applications, (2026): 1-27, doi: 10.22034/jsmta.2026.24972.1234
VANCOUVER
Ghajari, A. Dynamic principal component copulas with time-varying tail dependence for high-dimensional systemic risk monitoring. Journal of Statistical Modelling: Theory and Applications, 2026; (): 1-27. doi: 10.22034/jsmta.2026.24972.1234
)