OLS regression and clustered standard errors

[Melika]

Hello, 

I have a panel data of less than 100 observations. The dependent variable is stock market returns and independent variables are firm characteristics (some binary variables and some numerical values). I have multiple observations for several of the firms in my sample so I need to run an OLS regression with standard errors that are clustered by firm. So far I have the following code but I am not sure if proc genmod actually accomplishes what I am trying to do. I would really appreciate any help!

proc genmod DATA=firmreg;
class firm_id;
model return = var1 var2 var3 var4 var5 var6 var7;
repeated subject = firm_id/type=ind;
run;
quit;

[StatDave]

Yes, that code will fit a regression model which assumes that the response is normally distributed, and use the Generalized Estimating Equations (GEE) method to provide standard errors that account for the correlation due to clustering within firms. However, note that validity of the GEE method assumes a large number of clusters (approximately 100 or more). Another approach would be to fit a random effects model using PROC MIXED or GLIMMIX.