The model is:
lnGDP = lnINVESTMENT + lnX + E
The data consist of a panel of six countries and 19 year periods for each.
All variables are stationary at I(1) and are cointegrated (two pre-requisites for using FMOLS method).
Should I use the lag of investment (as well as the lag of other independent variables in X vector) as instrumental variables to avoid endogeneity?