I am using a panel dataset (N=73; T=9). Dataset Timeframe: 2010-2018
In the GMM estimate on the total dataset, the AR(1) and AR(2) values are fine.
But to investigate the impact of the European crisis, I had to split the data (5 Years during and immediately after the crisis, and the subsequent 4 years). But when GMM is run on the second set of data, (2015-2018), in one of the models, AR(1) and AR(2) values were not generated.
Is the result still usable? What are the potential problems of using this specific result?
Atiq you please use dummies to separate the effect. It saves space and time. Thanks
Md. Thasinul Abedin, Are you suggesting that I introduce a dummy ( e.g., 1 for EU Countries) and see the interaction of the Dummy variable with the independent variables? I tried this but this creates some serious problems of multicollinearity if both the dummy and the interaction variables are kept in the model. :(
Just keep the interaction variables to see whether there is any moderation or extremism effect. My suggestion is you please take multiple columns to report the results. In each column, report individual interaction effect. Then in the last column, report the all interaction effects. In each case, don't report the effect of dummies. Just report the interaction effects. Here, the interest is to see either moderation or extremism! If the coefficient of interactions is negative, it will urge for moderation. Thanks
Md. Thasinul Abedin , Thanks a lot for your amazing explanation. I actually have been doing this finally. But I forgot to omit the dummy variable initially. Really appreciate your suggestions.
Check this out: https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&cad=rja&uact=8&ved=2ahUKEwiG0-uH4M34AhVoxYUKHcIoD6EQFnoECBkQAQ&url=https%3A%2F%2Fwww.econstor.eu%2Fbitstream%2F10419%2F73774%2F1%2FIfoWorkingPaper-118.pdf&usg=AOvVaw2s7COMAs8GxtjdDw_TJzIt
Dear Respected, As per my understanding we can use panel OLS to test the impact of a specific period. If GMM estimations results contain the problem of first or second-order autocorrelation the coefficient will be biased. You may check the interactive term to see if it works.
- Badges
- Science topic
- Similar topics
- Filing