Re: Re: st: inconsistent results for two-dimensions fixed effects

[Nahla Betelmal <nahlaib@gmail.com>]

Thanks Kit for the great input. Just to check that I got it right.

if the panel is set to firm and year
to get two dimensions fixed effect using xtreg, the right way is to run :
 xtreg DV IV i.year, fe vce (cluster firm) , but not xtreg DV IV
i.year, fe  because the latter does not control for correlation of
error terms within clusters.

Also, it is feasible to use :  ivreg2  DV IV,cluster (firm year)
IV is independent variable (not instrument)


Finally, reg DV IV i.year i.firm does not provide the same out as the
previous two because cluster-robust VCE is not applicable in this
case. However, it will give the same result as xtreg DV IV i.year, fe

Many thanks again, I highly appreciate your kind help

Nahla


On 16 August 2013 13:33, Christopher Baum <kit.baum@bc.edu> wrote:
> <>
> Michael said
>
>> Lastly, if you are including fixed effects at the industry-level, you
>> don't have to compute clustered standard errors at the same level. You
>> can just use the typical robust standard error estimator. The cluster
>> fixed effects will control for correlation of error-terms within
>> clusters.
>>
>> So I think you should use one of these two commands:
>> reg IV DV i.year i.industry, robust
>> areg IV DV i.year , absorb (industry) robust
>>
>> About the ivreg2 command, it is used for instrumental variables. I
>> think your "IV" stands for independent variable, not instrumental
>> variable, so it is not relevant to your topic. ivreg2 will not help
>> you with a fixed-effects analysis.
>
>
> This is not quite right. First of all, inclusion of fixed effects (FEs) adjusts the conditional mean for, e.g., the industry; it does not deal with non-i..d. errors. You cannot 'just use the typical robust s.e. estimator' in that context; as the documentation for xtreg, fe states, Stock and Watson showed that the standard robust estimator is not consistent in the FE context where there are more than two observations per panel unit. xtreg, fe robust automatically uses the cluster-robust VCE, clustered by panel ID. Your regression 'by hand' using -regress- will not give you the same results as xtreg, fe robust or xtreg, fe cluster(pid).  Also note that areg and xtreg, fe are designed for different assumptions about the data.
>
> FEs by cluster will, thus, not 'control for correlation of error terms within clusters'. That is what the cluster-robust VCE is for.
>
> ivreg2 (SSC, Baum/Schaffer/Stillman) and its sibling xtivreg2 (SSC, Schaffer/Stillman) are potentially relevant in this context, as they  implement two-way clustering in an OLS context (as well as in an instrumental variables = IV context).
>
> Kit
>
> Kit Baum   |   Boston College Economics & DIW Berlin   |   http://ideas.repec.org/e/pba1.html
>                              An Introduction to Stata Programming  |   http://www.stata-press.com/books/isp.html
>   An Introduction to Modern Econometrics Using Stata  |   http://www.stata-press.com/books/imeus.html
>                                                                                                       http://www.crup.com.cn/Item/111779.aspx
>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/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/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/