I want to analyse; short and long term, effects of independent variables on dependent variable for my project. I am using panel data for 25 years. For normal analysis panel regression (OLS) will do. But, how can I check for long-term and short term?
There are several panel cointegration methods and that will take care of the long term. For the short term, all variables are stationary, you should use error- correction model that includes a linear combination of the level terms or ecm(t-1). That,s it!
There are several panel cointegration methods and that will take care of the long term. For the short term, all variables are stationary, you should use error- correction model that includes a linear combination of the level terms or ecm(t-1). That,s it!
Absolutely agree with both professors.
By employing an Ecm it will give u both short run and long run coefficients. Another possible solution is to estimate one model with ols which wil give u theoretically the short term effect-coefficient and then the same model with FMOLS which will give u the long run coefficients and thus the long term effects.
Agree with comments above.
You can check this video to see step by step procedure on STATA also, hope it helps!
https://www.youtube.com/watch?v=YI-r8P_2ZoE&list=PLSmDYXCpxDHqA-YvQeW0AdpE5fvuNfxWk
In addition to co-integration and error-correction frameworks, you might also consider the state- and flow-adjustment models of Houthakker and Taylor (Taylor, L.D. and Houthakker, H. S., Consumer Demand in the United States, Springer-Verlag, 2010, Chapter 13). Good luck in your research.
Econometricians often are too method-oriented. They take a set of data (the more the better) and throw it into a (modern) estimation program. Before doing so, one should analyse the problem and try to find a rough model in which way variables affect each other. One should also analyse the information content of the data and possible relations between them (by diagrams and correlations). After that you may choose which data (transformation) and which estimation method to use, and I would take the simplest possible. One can go to more refined programs, if the results suggest to do so.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression