Robust_Variance_GLS - A function to estimate robust variances for parameters estimated using the GLS function.
It estimates the "sandwich" variance but for GLS models. Unlike the sandwich estimator it also accounts for covariance accross time. It can be used when the covariance structure is misspecified to make the covariance of the parameters "robust" to covariance misspecification . The code currently works for balanced panels and the code for unbalanced panels is forthcoming.
Any suggestions are appreciated. Please reach out if you want to discuss the code or if there any issues. The code is relatively new and has only been tested for longitudinal studies. I am still unclear if it will work for clustered data or for non-continuous dependent variable. The code will be updated when those become clear.
Few Notes:
- Change your working directory and data
- Specify number of subjects(N)
- number of measurement occurences (n)
- number of parameters (p)
- model (model)
- data (data)
- y (level)
- Id (ID)
- time (time)