Using structural equation package

[Aakanksha Melkani]

Hello!
I have been trying to figure out ways to run a four equation simultaneous equation model in panel framework (N>T). I came across the "sem" and "gsem" commands on Stata. I tried reading through the manual but it is lengthy and I do not know whether it will really be useful for me.

If someone could answer these few elementary questions I would be in the clear whether or not I should devote my time to the structural equation package right now?
1. Am I going in the right direction by using the structural equation package to solve my problem? Is there some easier, more direct way of solving it?
2. Is the package accessible with Stata 13? I do not have a higher version. May be difficult to get one in the given time.

Any other helpful comments (not necessarily related to sem and gsem) are most welcome.

Thanks!

[Clyde Schechter]

The -sem- and -gsem- commands are both available in Stata 13, though there have been some additions to their functionality in Stata 14.

I don't think anybody can answer #1 without more information about the problem you are trying to solve. Certain types of panel-data structures and questions lend themselves to modeling with -sem- or -gsem-. Some can be done that way but it's kludgy. And some are, for practical purposes, unfeasible in that framework.

[Phil Bromiley]

You can probably want to look at reg3 and maybe SEM/GSEM - indeed you should be able to do a model in SEM that is almost identical to some variants of reg3. You need to check whether just throwing in dummies for panels in reg3 is sufficient for consistent fixed effects estimates - I would think so but am not sure. Of course, reg3 doesn't handle latent variables. If reg3 works for what you want, it is much easier to use than SEM.

If you're considering SEM, you might look at Bollen & Brand "Fixed and Random Effects in Panel Data Using Structural Equations Models" (readily found on line). They take a more positive view of using SEM in panel data than Clyde does. They don't deal with simultaneous equations but they would help you understand the issues. Some SEM variants are available in Stata 13, but Stata 14 is a lot better than 13 for SEM. If you're using a charge card, you can buy and download Stata 14 in minutes.

[Aakanksha Melkani]

Thank you Clyde and Phil. I have dropped that execise now due to lack of time. May come back to it later. I will keep your suggestions in mind though.