Is it correct to remove 1% outliers on stock market returns in before applying ARCH LM Test?
That makes a lot sense sense. Outliers distort the estimations. Therefore, one first needs to detect jumps in financial returns or disturbances using a semi-parametric test for additive jumps suggested by Laurent et al. (2016), and then apply ARCH-LM test to the filtered series. Literature shows that the jump-filtered returns or disturbances also display excess skewness, kurtosis and conditional heteroscedasticity. However, in my opinion, one should not simply remove the outliers and apply the test.
See:
Laurent, S., Lecourt, C., Palm, F.C., 2016. Testing for jumps in GARCH models, a robust approach. Comput. Stat. Data Anal. 100, 383-400.
I recommend removing them, not without first, try using a dummy variable for the dates where the extreme values are presented.
Also, would like to confirm if even after applying mgarch model on around 10 stock market indices in 2-3 cases the post diagnostic autocoorelation / hetroskedicity arch still remains, still can we go ahead and use the model. I am asking since in most return data for stock markets, i am seeing only (1,1) model
Your word would require a dummy in the mean equation OR requires inclusion of dummy in the gatch model
- Badges
- Science topic
- Similar topics
- Correction