Simultaneous Equation Model (PROC SYSLIN) using Panel Data

[niam]

Hello,

I have a panel dataset, and I would like to run a simultaneous equation model using time and individual fixed effects. It seems that PROC SYSLIN does not support panel data. Is there a solution for this?

My model is:

Y=f(X,Z1)

X=f(Y,Z2)

Both endogenous variables (X and Y) and exogenous variables (Z1 and Z2) are measured for a group of individuals over a period of time.

Thanks for you help in advance.

[niam]

Ok, I figured that instead of running a simultaneous equation model, I can write the reduced form of the equations and use a 2sls method to estimate the model. 

The method is described here:

http://econ.ucsb.edu/~doug/140a/Lectures/18%20Simultaneous%20Equation%20Models%20Two%20Stage%20Least...

But I still wonder is SAS provides a package for 2sls estimation using panel data?

[niam]

Hi, 

Stata has a command for handing this issue. It is xtivreg 

Does SAS have similar capabilities? 

Please, help! I do not want to migrate to SAS after being a user for 8 years! 😞

[mitrov]

Hi,

The MODEL procedure SAS/ETS can estimate simultaneous equations models using 2sls and other methods.

This is an example of a non-linear 2sls estimation:

http://support.sas.com/documentation/cdl/en/etsug/60372/HTML/default/viewer.htm#etsug_model_sect004....

Best

Michele

[niam]

Thank you Michele 

I have a panel dataset. I don't think proc model can handle panel data structures. 

[mitrov · Accepted solution]

Have you looked at the PANEL procedure? There are several estimation methods. E.g., HT is a 2sls. I am not familiar with Stata, I wouldn't know the equivalent of xtivreg.

http://support.sas.com/documentation/cdl/en/etsug/68148/HTML/default/viewer.htm#etsug_panel_syntax12...