I am running VAR(1)-BEKK-GARCH(1,1) model on 5 series.
these are the issues that i'm facing
1. My 3 and 4th series at lag 1 is showing insignificant result hence means its not having any impact from the previous information.(I am working on 5 countries stock indices).
2. while running the Uni variate diagnostic the results of McLeod Li test for series 1 and 5th series is significant. hence not good.
Explain the way through which this problem can be removed.
Are you using Full BEKK or Diagonal BEKK?
The univariate models are simply inputs into the multivariate model, so the lags of the former, significant or otherwise, are not particularly important.
I am using full BEKK. And what about McLeod Li test.
these are the results of the series which are not showing good results.
my series 3 is China. In the Mean Modelling part i.e. VAR(1) at its lag 1 its not giving significant result. So, is it OK. As it can be interpreted that China stock market is not getting affected by its previous day information.
Independence Tests for Series Z1
Test Statistic P-Value
Ljung-Box Q(20) 27.099089 0.1325
McLeod-Li(20) 83.302551 0.0000
Independence Tests for Series Z5
Test Statistic P-Value
Ljung-Box Q(20) 41.21270 0.0035
McLeod-Li(20) 321.78551 0.0000
It is not advisable to use FULL BEKK as it has no mathematical or statistical properties whatsoever.
I recommend that you read the following paper on DCC, which also discusses Full BEKK:
“What they did not tell you about algebraic (non-)existence, mathematical (ir-)regularity and (non-)asymptotic properties of the dynamic conditional correlation (DCC) model, Journal of Risk and Financial Management, 12 (2:61), 1-9.
As I mentioned previously, the tests of independence at the univariate level are not important as they are simply inputs to the second stage in estimating the multivariate conditional volatility models.
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running