Dear all,
I want to know
1st
Whether Bivariate Johansan cointegration and Bivariate BEKK GARCH models can be applied on daily data of 7 years..derived from stock market or monthly data is suitable for these..
2nd
query is that before applying Johansan cointegration test the VAR model (just to select lag length to be used in johansan test) that we run is applied on non stationary data or stationary data. because after that johansan test is applied on non stationary data.
3rd
on pair of two series co-integration approved then we run VECM on this pair but what happens in case error term in vecm comes significant but r square value comes very low and residual diagnostic shows arch effect ...means vecm models proved unfit. how we comment on this? whether we say like that even though error term comes significant residuals arch effect and low r square values made our model unfit so we avoid to use this model.
Teena ...
1. For volatility analysis or say time series analysis, large data series is always suitable. If you have daily data, go for it.
2. just simply run var with raw data. then select the appropriate lag structure.
3. Just cross checked all the procedures. if same results are comes out, report it what as it is
Daily data in the seven years period give you enough observations to apply the indicated methods.
- Badges
- Science topic
- Similar topics
- Mathematics
- Applied Mathematics