Estimation of portfolio diversification with copulas

• Ano de defesa: 2022
• Autor(a) principal: Paolucci, Arthur Abreu
• Orientador(a): Souza, Rafael Martins de
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Palavras-chave em Inglês: Brazilian markets; Copulas; Financial risks; Mean-variance; Modern portfolio theory
• Link de acesso: https://hdl.handle.net/10438/32657

Portfolio diversification plays a key role in asset allocation, still, its benefits are often overestimated by the classic mean-variance model. Although most publications focus on the estimation error of returns, this study examines the impact of dependence models on diversification. Copulas are explored as an alternative to model some equally weighted portfolios composed of IMA-B 5, IMA-B 5+, IBOV and S&P 500 from 2004 to 2021. They reveal to be useful tools to gain a deeper understanding of the dependence structure of portfolios, however, there is no evidence they reduce estimation error compared to the multivariate normal. Nevertheless, the results provide some interesting insights to both managers and researchers such as Brazilian inflation-linked indices providing little (or no) risk reduction during extreme loss events and portfolio dependence structures changing significantly over time.