Dynamic Panel Data Models Flashcards
What is a dynamic model?
Models where your dependent variable in the previous period of time is included as an explanatory variable now
What is the correlation between the transformed lagged dependent variable and the transformed error term in the fixed effects estimator?
What is the problem with the Fixed effects transformation?
This transformation makes every observation in the transformed data endogenous to every other for a given individual
What is the Generalised Methods of Moments estimator?
- Very similar to the 2SLS
- Only difference is GMM allows for heteroskedasticity and autocorrelation within individual unit errors but not across them
How do we represent a dynamic panel model?
Why can’t we instrument dynamic panel data as we would regular panel data?
because yt-1 is on the RHS, we must difference both sides
Is the error term correlated with values of yt-k where k≥2?
No, they all equal 0
How many instruments do we need for t time periods?
t-2
What is the main property of Difference GMM?
How does the system GGM ‘augment’ the difference GMM?
What is the issue with the differenced residuals exhibiting AR(2) behaviour and how do we fix it?
What is the correlation between the differenced residual and the differenced dependent variable (lagged j times)?
What are the properties of the Sargan/Hansen test, and when do we reject the null?
We reject the null when the chi-square value is greater than the critical value, or p > 0.05