Hi all
Can I run a Gmm model with a time series dependent variable? I have a sample of 12 countries and years from 2000 to 2018
If the variable is changing over time and across countries, I think so. If the variable is only time-varying, then the case reduces to time-series analysis.
Keep in mind the limitations due to the low n (number of countries).
Please first advise how the series in your dataset would be arranged
Elvis Munyaradzi Ganyaupfu It is a poverty dynamic model. I am applying the model on MENA countries, and unfortunately the poverty data are not available on the national level for all countries. Therefore, my poverty dependent variable is the estimated poverty rate for the whole region. On the other hand, I have two independent variables and three control variables and the data are available on the country level for these variables. What is your advice?
Since your number of countries is lesser than the number of years using standard panel techniques needs caution. Also, for testing stationarity 18 years annual data may be less. You may consider adding more countries in your sample (all countries in the world).
Shreya Biswas Thank you very much. Actually you are right about the fact that the N should be larger than T. Therefore, I have increased the number of N to 33 countries.
If you are dealing with panel data, its not compulsory that you capture all the entire 21 countries in MENA, you can choose the sample countries based on data availability...Also note that , Gmm works for N>T. Since your T>N, you can try using DOLS and FMOLS since they also correct for endogeneity ..
- Badges
- Science topic
- Similar topics
- Statistics
- Time Series