A non-parametric test for independence based on symbolic dynamics |
| |
Authors: | Mariano Matilla-Garcí a, |
| |
Affiliation: | aDepartamento de Economía Aplicada Cuantitativa I. Facultad de Ciencias Económicas, UNED. Calle Senda del Rey, 40. CP. 28040, Madrid, Spain |
| |
Abstract: | In the present paper we propose a powerful, yet simple, non-parametric test for independence based on symbolic dynamic analysis. The absence of dependences in the unknown underlying data generating process is studied via symbolic dynamics. This is possible due to the ordering property of real numbers on an interval. Interestingly, the test is closely related to entropy concepts. Apart from being correctly sized, the new test is powerful for realistic finite data sets, and it is easy to use as one does not need to select any free parameter, which sharply contrasts with other tests of independence. In addition, the test is robust in the presence of noise which is one of the most typical cases when dealing with economic time series. |
| |
Keywords: | Entropy Non-parametric test Independence test Symbolic dynamics Chaos |
本文献已被 ScienceDirect 等数据库收录! |
|