Portfolio optimization based on forecasting models using vine copulas: An empirical assessment for the financial crisis

TL;DR

This paper evaluates the effectiveness of vine copulas in modeling dependencies and forecasting returns for portfolio optimization during the 2008-2009 financial crisis, highlighting their superior risk reduction capabilities.

Contribution

It introduces the use of various vine copula structures for portfolio optimization and compares their performance in risk management during a financial crisis.

Findings

Vine copulas outperform simple copulas in risk reduction.

D-vines are more effective than R- and C-vines in reducing conditional VaR.

Student-t drawable vine copulas perform best overall.

Abstract

We employ and examine vine copulas in modeling symmetric and asymmetric dependency structures and forecasting financial returns. We analyze the asset allocations performed during the 2008-2009 financial crisis and test different portfolio strategies such as maximum Sharpe ratio, minimum variance, and minimum conditional Value-at-Risk. We then specify the regular, drawable, and canonical vine copulas, such as the Student-t, Clayton, Frank, Joe, Gumbel, and mixed copulas, and analyze both in-sample and out-of-sample portfolio performances. Out-of-sample portfolio back-testing shows that vine copulas reduce portfolio risk better than simple copulas. Our econometric analysis of the outcomes of the various models shows that in terms of reducing conditional Value-at-Risk, D-vines appear to be better than R- and C-vines. Overall, we find that the Student-t drawable vine copula models perform best with regard to risk reduction, both for the entire period 2005-2012 as well as during the financial crisis.