i am having a panel of nine countries with some 4 variables. and my data is of 20 years. Since there are many missing values in the series it is difficult to sort the data and make it a balanced panel. As daily or weekly data is good for volatility studies i tried to get daily and weekly data. Can i proceed with this or go for monthly data?. My objective is to test spillover effects and volatility...
Hi
Yes, always one can run though, first the unbalanced daily and weekly panel data are itself comes with complexity, high-frequency data with full of noise. It will provide the output bt one would need meaningful output for interpretation. Though such a huge data set of high frequency always comes with full of missing value and will difficult to set the relationship. If previous studies employed this type of data then go a head otherwise monthly data makes much more sense to proceed with. You can model any volatility modelling to check the spill over effect with Panel unbalanced-data set, though needs due care.
Hi Roy,
I am not sure if Garch can work with panel. I am thinking is only for Time series.
Dear Ashiq
you can do it, but with the sample of 20 years through weeklly or daily or monthy you can test the GARCH for panel, but this will put you to difficultly. So I suggest you separate data into 4 sample for 4 countries and test it by time series techniques.
Best
Canh
I am dealing with a garch estimation on panel data in stata but I do not find any way to perform this estimation. Maybe with stata17, this is feasable, I have no details however.
- Badges
- Science topic
- Similar topics
- Statistics
- Time Series