Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: Clustering Standard Errors.
From
natasha agarwal <[email protected]>
To
[email protected]
Subject
Re: st: Clustering Standard Errors.
Date
Tue, 15 Jun 2010 10:31:15 +0100
Thanks Austin and Nils
On Tue, Jun 15, 2010 at 8:11 AM, Nils Braakmann
<[email protected]> wrote:
> Ah, thanks Austin. I didn't know that.
>
> On Mon, Jun 14, 2010 at 11:25 PM, Austin Nichols
> <[email protected]> wrote:
>> Nils Braakmann <[email protected]> :
>> No, not the same. The group() approach uses a much more restrictive
>> assumption about independence of errors, and that VCE is actually an
>> ingredient of the two-way cluster-robust VCE. Quoting from
>> http://fmwww.bc.edu/repec/bocode/i/ivreg2.html :
>>
>> The two-way clustered variance-covariance estimator is
>> calculated using 3 different VCEs: one clustered on varname1, the second
>> clustered on varname2, and the third clustered on the intersection of varname1
>> and varname2. Cameron et al. (2006, pp. 8-9) discuss two possible small-sample
>> adjustments using the number of clusters in each category. cgmreg uses one
>> method (adjusting the 3 VCEs separately based on the number of clusters in the
>> categories VCE clusters on); ivreg2 uses the second (adjusting the final 2-way
>> cluster-robust VCE using the smaller of the two numbers of clusters).
>>
>> Whether or not you should use one approach or another depends on many
>> things, including your own most plausible theories about independence
>> of errors and the number of clusters in each dimension; see
>> http://www.stata.com/meeting/boston10/abstracts.html#baum
>>
>> On Mon, Jun 14, 2010 at 3:19 PM, Nils Braakmann
>> <[email protected]> wrote:
>>> Hi Natasha,
>>>
>>> I guess the two approaches should give identical results (although I'm
>>> not sure how two-way clustering is implemented in e.g. xtivreg2). The
>>> -egen ... group()- approach essentially creates all possible
>>> combinations of (in your case) industry and region and clusters on
>>> that. In my understanding, two way clustering should do the same.
>>>
>>> Best,
>>> Nils
>>>
>>> On Mon, Jun 14, 2010 at 4:50 PM, natasha agarwal
>>> <[email protected]> wrote:
>>>> Thanks Nils.
>>>>
>>>> How would you know whether you need to do two-way clustering or
>>>> generate a cluster variable?
>>>>
>>>> Thanks
>>>> Natasha
>> *
>> * 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/