Artigo Produção Nacional Revisado por pares

Statistical test for Multiple Detrended Cross-Correlation Coefficient

2020; Elsevier BV; Volume: 562; Linguagem: Inglês

10.1016/j.physa.2020.125285

ISSN

1873-2119

Autores

Aloísio Machado da Silva Filho, Gilney Figueira Zebende, Arleys Pereira Nunes de Castro, Everaldo Freitas Guedes,

Tópico(s)

Financial Risk and Volatility Modeling

Resumo

In this paper, we proposed a statistical test for the Multiple Detrended Correlation Coefficient DMCx2. The DMCx2 is a measure of total association between the dependent variable, Y, with other p independent variables, X all with the same length N. DMCx2 is based on the generalization of the ρ(Xi,Xj) cross-correlation coefficient. With this methodology, it is possible to evaluate the statistical significance of DMCx2 for different levels confidence. Findings on this research show that rejection or non-rejection of the null hypothesis of DMCx2 also depends on the size N of the series and the time scale n evaluated. Our findings also show a behavior pattern in the critical values of DMCx2. Fixing the size of the series (N), as the size of the time scale n increases, the critical values tend to increase.

Referência(s)