Smooth Nonparametric Bernstein Vine Copulas

TL;DR

This paper introduces a nonparametric approach using Bernstein copulas for high-dimensional vine models, eliminating the need for parametric pair-copula selection and demonstrating superior performance in financial data analysis.

Contribution

It presents a novel smooth nonparametric vine copula model that replaces parametric pair-copulas with Bernstein copulas, simplifying model selection and improving accuracy.

Findings

Nonparametric vine copulas outperform parametric models in simulations.

The method effectively captures complex dependencies in financial data.

Smoothness of Bernstein copulas enhances model robustness.

Abstract

We propose to use nonparametric Bernstein copulas as bivariate pair-copulas in high-dimensional vine models. The resulting smooth and nonparametric vine copulas completely obviate the error-prone need for choosing the pair-copulas from parametric copula families. By means of a simulation study and an empirical analysis of financial market data, we show that our proposed smooth nonparametric vine copula model is superior to competing parametric vine models calibrated via Akaike's Information Criterion.