RE: st: Prais with vce(robust)

["Nick Cox" <n.j.cox@durham.ac.uk>]

I am not surprised that this does not do what you need. 

As the behaviour change in -prais- in Stata 11 was not deliberate, it follows that there is _no_ switch in the code for -prais- that allows this use of version control to get what you want. 

-version- does much less than many people think. Only Stata 10.1 is a complete replica of Stata 10.1, for example. Stata's not like a set of Russian dolls with older versions wrapped inside newer versions. 

There is more explanation at

FAQ     . . . . . . . . . . . . . . .  User-written programs and Stata version
        . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  N. J. Cox
        2/09    I have a certain version of Stata and have come
                across a user-written program that is apparently
                written for a later version of Stata. What are
                my options?
                http://www.stata.com/support/faqs/lang/version2.html

This stuff is a bit complicated, but Brian knows what he is talking about: only doing this within Stata 10.1 will suffice until this is fixed. 

Nick 
n.j.cox@durham.ac.uk 

Thomas Jacobs

I am back again on this and now unable to run prais y x, vce(robust)
even when affixing version 10.1 at the top of the program.  Is there
something else I need to do to ensure it applies the rules for 10.1 as
opposed to 11.1?  I have:

version 10.1
*Tb_Infl_Results_MonthlyLevChng.do

...

prais IISBasis `Liquidity' `InfVar' if TenorYr==BasisTenors[1,`j'], vce(robust)

and I receive:

option vce() not allowed
r(198);

It runs fine if I lift the robust option just as in 11.1

On Tue, Jun 22, 2010 at 8:15 AM, Brian P. Poi <bpoi@stata.com> wrote:

> On Mon, 21 Jun 2010, Thomas Jacobs wrote:
>
>> Brian,
>>
>> Thanks for the update.  In the meantime, am I best off simply using
>> robust without vce (regress y x, robust) or going back to version 10.1
>> in my program?
>>
>> Tom
>>
>
> Tom,
>
> If you need to use Prais-Winsten or Cochrane-Orcutt regression with robust
> standard errors, you'll need to do so in version 10.1 until we get the fix
> out.  Of course, if you just want to do plain regression with robust
> standard errors, you can just use -regress- in Stata 11 instead.  One other
> alternative would be to use -newey- to obtain Newey-West standard errors;
> they are robust to autocorrelation.

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/