Non-Linear Dependence in Oil Price Behavior

摘要: In this paper, the authors analyze adequacy of GARCH-type models to oil price behavior by applying two types non-parametric tests, Hinich portmanteau test for non-linear dependence and a frequency-dominant time reversibility, reverse based on bispectrum, explore high-order spectrum properties Mexican series. The results suggest strong evidence structure irreversibility. Therefore, it does not comply with i.i.d (independent identically distributed) property. dependence, however, is consistent throughout sample period, as indicated windowed test, suggesting episodic nonlinear dependence. imply that GARCH cannot capture series structure.