Can anyone help me to carry out mean group analysis and pooled mean group analysis. I have used Microfit and Eviews before. Appreciate if I can get some advice on how to use these panel data methods in Microfit, Eviews and STATA.
Hi there,
I have done this in R before. I think it would be very easy for you to do this in R as many time-series methods are available in this program and it is a simple matter of creating a looping algorithm to estimate individual models for each group/individual. I used this approach to estimate a cointegrating equation and error correction model about a year ago - it worked very well as i automated all tests (e.g. Phillips-Peron).
R is a free statistical programming language with great online support and a huge library base for different statistical methods. It is also easy to program it yourself.
I have not done it in E-views, not sure if it is available... Hope this helps as an alternative solution.
-Daniel
Hi Gregg,
Thank you for the advice. I have some basic knowledge of R. However, am not quite sure how to implement PMG and MG algorithm. Does R support panel data estimation? If it does, then I would need your guidance on coding matters for the analysis.
I am familiar with Mfit. But Mfit does not support PMG or MG under ARDL framework.
Looking forward to your advice.
Ron
Hi Ronald,
R supports any estimation you can code really.. The PMG and MG estimator is simply repeated estimations of a regression/model for each individual. You could do this manually but if you had say >30 individuals it would certainly become tedious. The PMG model is a good option for time series data because each group/individual may have a different level of serial correlation - forcing the same level on all responses causes all results to be potentially biased whilst using individual responses allows those for which your model assumptions are correct to remain unbiased. You can even allow for different levels of serial correlations between groups. So it is a simple idea applied many times essentially and the summary is done using standard data summarisation - ideally not using the mean of parameter estimates but the resultant empirical distribution of parameter estimates for each of your groups.
If your model is a standard model which R handles automatically it would be very easy and would look something like this:
(after initialising data)
for(i in 1:length(y)){
reg= "ESTIMATE MODEL"
results.list[[i]]=summary(reg)
}
If you want to do a panel data estimation instead (single model) then you may need to code the model directly. If it is a maximum likelihood method and the likelihood function is available this should be easy enough - I can help you with this if needed but would need the likelihood function. If it is another model from the literature you might be able to ask the authors how they coded their model. If it is an easy model (such as single vector cointegrating equation model) i would be happy to help - i need to redo my code for a paper i want to do anyway. If it is something more complex someone may still have made available a library for model estimation in R, in which case it should, again, be easy.
Someone else may know more and be able to suggest programs which can do this for you (and perhaps with automated testing as well!)...
So what regression/model do you want to estimate exactly?
Also would a panel Vector AutoRegression (VAR) model suit???
Hi Gregg,
thanks a lot. I am interested panel VAR model. I'm attaching a paper. Is it possible the model in R? Please give me your email and we can discuss this further. I need your help on Dynamic Panel estimation. Maybe we can work on a paper together. Many thanks.
Thank you very much. This is indeed helpful. Can we perform this in EViews?
Loganstahn ! There is typographical mistake it code, The installation code is 'ssc install xtpmg' not xtpmfg
How do I get the dataset as used by Pesaran et al. in their work. I want to replicate the work for clearer understanding. Reference will be noted. Thank you.
Hi dear
you can use STATA to estimate Pooled mean group (PMG), Mean Group (MG) and Dynamic Fixed effect (DFE) by installing xtpmg on STATA.
to install it write in command editor of STATA: findit xtpmg and press enter and follow the installation procedure.
after installing you can use help xtpmg t learn how to estimate.
regards
Dr. Morteza Molaei
Dear Researchers, I do need how to run Panel ARDL model in Eviews. Thank you
Hi Dear,
How can we apply Panel NARDL model in Eviews or Stata?
Thank you
MG considers heterogeneity in both slope and constant, while PMG only assumes fix effect heterogeneity in the long-run specification! it means, PMG uses ML to get the long-run equation, but MG estimates different coefficients (constant and slope) for each panel separately!
Does anyone know if you can run the MG estimator in Eviews? (Hausman test rejects null hypothesis therefore PMG is not efficient, and i need to use the MG estimator.)
With ssc install xtpmg, you can run
Pooled mean group (PMG), Mean Group (MG) and Dynamic Fixed effect (DFE) on STATA.
Adding full as option gives details on Short run coefficients.
Convergence is achieve quicky with PMG and DFE than MG.
Best,
The Pooled Mean Group is an intermediate estimator that allows the short-term parameters to differ between groups while imposing equality of the long-term coefficients between groups.