Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

# Re: st: IV vs 2SLS

 From David Greenberg To statalist@hsphsun2.harvard.edu Subject Re: st: IV vs 2SLS Date Thu, 2 Aug 2012 20:44:44 -0400

```using 2sls gets the standard errors right. David

On Thu, Aug 2, 2012 at 8:31 PM, Shikha Sinha <shikha.sinha414@gmail.com> wrote:
> Thanks, David, but which is the preferred one and what are the advantages?
>
> On Thu, Aug 2, 2012 at 5:29 PM, David Greenberg <dg4@nyu.edu> wrote:
>> There is a reason for the difference in standard errors. When you
>> predict x3_hat and use the predicted value in your second equation,
>> Stata doesn't know that x3_hat is not an observed variable, but an
>> estimate with some uncertainty attached to it. It treats this variable
>> as observed, and without measurement error. Consequently, this
>> procedure underestimates the standard errors of estimates obtained
>> using the second strategy. David Greenberg, Sociology Department, New
>> York University
>>
>> On Thu, Aug 2, 2012 at 8:16 PM, Shikha Sinha <shikha.sinha414@gmail.com> wrote:
>>> Dear all,
>>>
>>> What is the difference between IV and 2sls and which is the preferred
>>>
>>> (a) ivreg2 y x1 x2 (x3=z)
>>>
>>> (b) reg x3 z x1 x2
>>> predict x3_hat
>>> reg y x3_hat x1 x2
>>>
>>> By employing (a) and (b) I get similar coeff, but standard errors are
>>> different.
>>>
>>> Thanks,
>>> Shikha
>>> *
>>> *   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/
> *
> *   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/
*
*   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/
```