xtivreg2 warning with time fixed effects

[Mihir Sharma]

Dear Statalisters,

I am working with an unbalanced panel data (385 observations, T=21, N=21). I am trying to estimate a fixed effects IV regression:

xi: xtivreg2 y var1 var2 i.year (var3  var2_var3 = z z_var3), fe cluster(id)

var1 and var2 are exogenours, var3 is endogenous and I instrument it using z. var2_var3 = var2*var3 and z_var3 = z*var3.
(-xtivreg2 is not allowing me to use -fvarlist- so I am doing interactions manually).

Stata gives the following error message:
Warning: estimated covariance matrix of moment conditions not of full rank. overidentification statistic not reported, and standard errors and model tests should be interpreted with caution. Possible causes: number of clusters insufficient to calculate robust covariance matrix singleton dummy variable (dummy with one 1 and N-1 0s or vice versa) partial option may address the problem.

The error does not come when I remove the year fixed effects i.year. Does anyone know why this is happening and what the warning means?

Many thanks,
Mihir

[Phil Bromiley]

I'm not sure what is happening with the warnings, but it seems like you're trying to identify the model by complex interactions rather than simply exogenous variables at do not appear in the main equation. This is not legit. Obviously, z_var3 is not exogenous since it is created using var3 which is the endogenous variable you're trying to instrument. So you're pretending you have two exogenous instruments for two endogenous variables, but you only have one.

[Mihir Sharma]

Thanks Phil. sorry there was a typo in my code. I meant

xi: xtivreg2 y var1 var2 i.year (var3 var2_var3 = z z_var2), fe cluster(id)

var1 var2 are exogenous and var3 is endogenous. This causes var2_var3 to be endogenous too and so I use z_var2 to instrument for the interaction. but even if I do this- I still get the error. The error is somehow coming in when I add the year fixed effects i.year. any idea why this might be happening?

Many thanks,
Mihir