본 논문에서는 비정상적 시계열이 갖는 장기균형관계를 검정하기 위한 검정방법을 연구하였다. 기존의 우도비 (LR) 장기균형검정에서는 동일 분포를 가정하기 때문에 자료가 갖는 이분산성에서 발생하는 문제를 다루지 못했다. 여기서는 이분산성을 고려하여 이분산성에 ...

본 논문에서는 비정상적 시계열이 갖는 장기균형관계를 검정하기 위한 검정방법을 연구하였다. 기존의 우도비 (LR) 장기균형검정에서는 동일 분포를 가정하기 때문에 자료가 갖는 이분산성에서 발생하는 문제를 다루지 못했다. 여기서는 이분산성을 고려하여 이분산성에 일치하는 분산추정량에 기초한 라그랑지 승수 (LM) 검정통계량과 Wald 통계량을 제시하였다. 모의 시뮬레이션의 결과 이분산성이 존재할 때 기존의 검정 방식에 비하여 우수한 성과를 보임을 밝혔다.

This paper considers statistical inference on the cointegration rank in vector error correction models, which
is robust to heteroscedastic errors. As the likelihood ratio (LR) statistic assumes identically distributed
errors, we develop the Lagrange ...

This paper considers statistical inference on the cointegration rank in vector error correction models, which
is robust to heteroscedastic errors. As the likelihood ratio (LR) statistic assumes identically distributed
errors, we develop the Lagrange multiplier (LM) and Wald statistics for the cointegration rank using
the heteroscedasticity robust covariance estimator. The asymptotic distributions of the LM and Wald
statistics follow the nonstandard distribution, which has been found by Johansen (1991).
Simulation evidence indicates that the proposed tests improve
the performance of the cointegration rank test.

This paper considers statistical inference on the cointegration rank in vector error correction models, which
is robust to heteroscedastic errors. As the likelihood ratio (LR) statistic assumes identically distributed
errors, we develop the Lagrange ...

This paper considers statistical inference on the cointegration rank in vector error correction models, which
is robust to heteroscedastic errors. As the likelihood ratio (LR) statistic assumes identically distributed
errors, we develop the Lagrange multiplier (LM) and Wald statistics for the cointegration rank using
the heteroscedasticity robust covariance estimator. The asymptotic distributions of the LM and Wald
statistics follow the nonstandard distribution, which has been found by Johansen (1991).
Simulation evidence indicates that the proposed tests improve
the performance of the cointegration rank test.

Economic variables show time-varying variance and clustered volatility, and the empirical studies
have found that the volatility can be explained by the near-IGARCH process. As the LR statistic for
cointegration rank assumes identically distributed ...

Economic variables show time-varying variance and clustered volatility, and the empirical studies
have found that the volatility can be explained by the near-IGARCH process. As the LR statistic for
cointegration rank assumes identically distributed errors, the statistical inference on cointegration rank
can be affected by conditionally heteroscedastic errors. This study proposes the LM statistic for cointegration rank,
which involves the heteroscedasticity consistent covariance estimator. The asymptotic distribution of the LM statistic
follows the nonstandard distribution in the same way as the LR statistic. The size performance of the LM statistic
exhibit heteroscedasticity robustness even when the errors are near-IGARCH.