Hi all,
I found a lot of papers that treat the endogenous growth theory issues using system GMM estimation technique by choosing 5 -10 years intervals ...can we apply system GMM for long run?
for example for economic growth i found a lot of paper use 5 year difference to generate the long run using 2 step system GMM
Hi Samia,
There is no doubt that the system GMM estimator is one of the most reliable estimators when it comes to economic growth (or, very long run). This estimator combines the equations in first-differences and equations in levels. Of course, you should first write your growth regression as a dynamic panel data model….
You might use the two-step system GMM estimator with Windmeijer (2005) finite sample corrections for the asymptotic variance.
To avoid modelling business cycle dynamics, you may split your sample period into few (three-, four-, or five-year) periods.
Best wishes,
Dimitar
Dimitar Eftimoski Thanks professor for the reply, so if i split my sample period to 5 years to avoid cycle dynamic, must i combining the Lagged and 5 years difference, or the estimator still combines the equations in first-differences and equations in levels?
Also if i split the period sample of 5 years to avoid cycle dynamic should i find in this case external instruments for my variables taken from endogenous growth theory or it is not mandatory in the case that hansen test results are good ?
Hi Samia,
As I mentioned in my previous post, the system GMM combines equations in first-differences and equations in levels. Endogenous (and predetermined) variables in first-differenced equations are instrumented with their lagged levels, whereas in levels equations - with their lagged first-differences.
You may use external instruments, but such instruments are very hard to find. They are not mandatory.
Best wishes,
Dimitar
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running