Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • System gmm

    Hi, Hope you are doing well. Sir, I am a bit confused about running system GMM. I have 3 questions but before that I would like to share my dynamic panel data model for a bit clarity.

    Rt = Rt-1 + Rt-1^2 + Mt + St + Ct + Ct-1 + St + St-1 + Dt + Dt-1 + Change in CHt + Change in CHt-1 + Firm dummy + time-specific effects + firm specific effects + error term.

    N=230, T=22

    This is the model used in the base study that I am following.
    • They reported to start with lag levels t-2 and extend till t-5 & t-6 as instruments for regression in differences.
    • They likewise suggested to start with lagged differences t-1 and extend till t-2 for regression in levels.
    Question 1:

    I am not getting whether i have to run lag variable generation command of each variable i.e. gen Ct-1 = lag Ct (_n-1) to follow the model variables and then run system GMM and specifying lags in levels of lagged variables as instruments and their differences. Or, is it that I have to mention independent variables and their lags and they automatically generate the lagged variables and instruments.

    Question 2:

    Can you please help me with stata command based on model and the 2 conditions below that.

    Question 3:

    Panel data is unbalanced and there are within year gaps i.e. values are missing somewhere in between as well not only from beginning and end. How i can run the forward orthogonal command?

    vif of regression model with no lags is 1.88. so there is no multicollinearity. Do i need to regress with lagged value of dependent variable or is it okay without that?


Working...
X