views:

335

answers:

1

Hey guys,

I'm trying to do a fixed effecs OLS regression, a random effects OLS Regression and a Hausman test to back up my choice for one of those models. Alas, there does not seem to be a lot of information of what the code looks like when you want to do this.

I found for the Hausman test that

proc model data=one out=fiml2;
      endogenous y1 y2;

      y1 = py2 * y2 + px1 * x1 + interc;
      y2 = py1* y1 + pz1 * z1 + d2;

      fit y1 y2 / ols 2sls hausman;
      instruments x1 z1;
   run;

you do something like this. However, I do not have the equations in the middle, which i assume to be the fixed and random effects models?

On an other site I found that PROC TSCSREG automatically displays the Hausman test, unfortunately this does not work either. When I type PROC TSCSREG data = clean; data does not become blue meaning SAS does not recognize this as a type of data input?

proc tscsreg data = clean;
var  nof capm_erm sigma cv fvyrgro meanest tvol bmratio size ab;
run;

I tried this but obviously doesn't work since it does not recognize the data input, I've been searching but I can't seem to find a proper example of how the code of an hausman test looks like.

On the SAS site I neither find the code one has to use to perform a fixed/random effects model.

My data has 1784 observations, 578 different firms (cross section?) and spans over a 2001-2006 period in months. Any help?

A: 

I think stats questions are better answered on http://groups.google.com/group/comp.soft-sys.sas/topics.

have a look here faculty.smu.edu/tfomby/eco5375/data/TSCS.SAS

SAS doesn't work well with panel data, can you get access to stata?

Dirk Nachbar