A Robust Modified Cox Test for Time Series Models

We reexamine the modified Cox test by Kim(2005) and propose a robust modified Cox test that does not require normality assumption. This approach can extend its applicability to more various time series models from asymmetric conditional distributions or thicker tail distributions than normal distribution. Some small finite sample simulation experiments show that this proposed test performs well in different sample sizes. The empirical applications to two time series data sets are also performed.

1. Introduction

2. A Robust Modified Cox Test

3. Simulation Experiments

4. Empirical Applications under Nonnormality

5. Conclusions

References

(0)

(0)

(0)

(0)