I would be grateful if anyone could advise on whether I could use the first differenced values of variables not stationary at levels alongside those that are stationary at levels. Is it advisable to use the first differences of all regressors since GMM uses these first differences as instruments.
↧