Serial Autocorrelation Tobit

[altijani]

I have a panel data, and I am running my model like this:

Y = B0 + B1X + error

I ran the model using QLIM procedure, and I need to include the serial autocorrelation to it. The DW test using Proc Autoreg shows significant P-Value as in this examample:

http://support.sas.com/documentation/cdl/en/etsug/60372/HTML/default/viewer.htm#etsug_autoreg_sect00...

I am a STATA geek, and this can be done in STATA in one line as follows (after setting my time variable):

xttobit Y X, robust

Can you please let me know how this should be done in SAS?

Thanks

[gunce_sas]

Hi,

You can estimate random-effects tobit models in PROC QLIM using the new RANDOM statement. A simple example, based on your model, for the syntax would be

PROC QLIM;

MODEL Y = X / censored(lb=0);

RANDOM INT / SUBJECT=id METHOD=HERMITE(QPOINTS=12);

RUN;

Currently, PROC QLIM does not offer an option for obtaining robust standard errors for heteroskedasticity and serial correlation.

Please note that xttobit does not have the “robust” option, either.

If you are willing to specify the nature of the standard errors, then you might estimate your model with PROC NLMIXED by modeling the standard errors explicitly.

Best regards,

Gunce

[altijani]

Thanks Gunce for your post.

Are you sure about the code? I tried it and it gave me this error in red:
ERROR 180-322: Statement is not valid or it is used out of proper order.

[gunce_sas]

Hi,

You must have an older version of SAS/ETS. PROC QLIM started supporting the RANDOM statement starting from version 14.1.

Gunce 

[altijani]

That is right. Any other solution that you might suggest?



Thanks.

[gunce_sas]

Hi,

As I mentioned before, you can use PROC NLMIXED and specify the nature of the standard errors.

Best regards,

Gunce