Hi !
I try to estimate a panel VAR model with GMM. Got 3 variables (exports, tarifs and gdp) for 14 sectors and 28 years.
My variables are I(1) and cointegrated. According to Greene and Schlacther (2005) or Sims (1980), it seems that I can't use first differences as it hides long-term relationship between variables.
Is that ok if I run the regression with non stationary variable ?