Re: st: Fixed Effects Model and Year Dummies

[Jed Cohen <jedcohen@vt.edu>]

Hi Drew,
For the model you propose, I would suggest you do some reading on
"treatment effects" and "difference-in-difference" models as this may
be more appropriate for your research. In any case when I say a "year
indicator" i mean you create a group of variables that take a value of
one if the observations falls within the stated year (or year/month
combination) and 0 if the observation is not within this year. You can
then include all of these in your model *except for the first one* and
this is equivalent to a time fixed effect.
In stata this can be done quite crudely as:
gen yearind2000=0
replace yearind2000=1 if year==2000
.
.
.
Hope that helps!
JED

On Fri, Sep 13, 2013 at 2:07 PM, Andrew Reed <drew.reed1@gmail.com> wrote:
> Hi Jed,
>
> Thanks for your response. I am modeling how exchange rates respond to credit rating agency news announcements from September 2000 until November 2011. I have a structural break in October 2006 where all time before coincides with a pre-crisis period and everything after is defined as being part of the European sovereign debt crisis.
>
> What do you mean by indicator variables, however? I believe my friend has run a regression like this, which she showed me last week but I can't quite remember the formulation...I believe had i's at the end of her regression line. If you would be so kind as to maybe describe this process so that I might adopt it as well, I'd be very grateful.
>
> Drew
>
>
>
> Den 13/09/2013 kl. 13.05 skrev Jed Cohen:
>
>> Hi Drew,
>> I would strongy suggest you include year dummies at a minimum, to
>> control for temporal variation in your dependent variable. I don't
>> know what you are modeling, but often a phenomenon varies between
>> years/seasons/months for reasons that are not adequately captured by
>> the explanatory variables in the model. Not including a time fixed
>> effect then leads to ommitted variable bias. you can use
>> xtset panelvar timevar
>> xtreg, fe
>> for this model, though I usually prefer to do this manually with sets
>> of indicator variables, which gives greater control and intuition into
>> the process. Hausman tests can generally be used to compare sets of
>> coefficients, though I think the decision to include time fixed
>> effects should hinge on whether or not you think there is temporal
>> variation that is not due to your set of explanatory variables.
>> Best,
>> JED
>>
>>
>>
>> On Fri, Sep 13, 2013 at 12:11 PM, Drew Reed <drew.reed1@gmail.com> wrote:
>>> Hi,
>>>
>>> I'm running a fixed effects regression on panel data. I am aware of
>>> the fact that fixed effects models control for time invariant
>>> characteristics, like that of country effects, etc. I am wondering,
>>> however, if I should include year dummies in my regression. Are these
>>> dummies time-invariant? Would Stata control for these in using the
>>> xtreg, fe command and are there any tests that can be performed to see
>>> if they should be included?
>>>
>>> Thanks,
>>>
>>> Drew
>>> *
>>> *   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/
>>
>>
>>
>> --
>> ----------------------------------------
>> Jed J. Cohen
>> Graduate Researcher
>> Virginia Tech
>> Dept. of Agricultural and Applied Economics
>> jedcohen@vt.edu
>> *
>> *   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/



-- 
----------------------------------------
Jed J. Cohen
Graduate Researcher
Virginia Tech
Dept. of Agricultural and Applied Economics
jedcohen@vt.edu

*
*   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/