作者: Igor Viveiros Melo Souza , Valderio Anselmo Reisen , Glaura da Conceição Franco , Pascal Bondon
DOI: 10.1080/07350015.2016.1251442
关键词:
摘要: ABSTRACTThis article proposes a method to estimate the degree of cointegration in bivariate series and suggests test statistic for testing noncointegration based on determinant spectral density matrix frequencies close zero. In study, are assumed be I(d), 0 < d ⩽ 1, with parameter supposed known. this context, order integration error is I(d − b), b ∈ [0, d]. Besides, dth difference power function b. The proposed estimator obtained here performing regression logged set Fourier frequencies. Under null hypothesis noncointegration, expressions bias variance were derived its consistency property was also obtained. asymptotic normality estimator, under Gaussian non-Gaussian innovations, established. A Monte Carlo study performed showed that suggested test...