Re: st: Setting some coefficients to zero in a panel probit regression

[Maarten Buis <maartenlbuis@gmail.com>]

Here is an example:

*------------------ begin example ------------------
sysuse auto, clear
gen byte miss = missing(rep78)
replace rep78 = 0 if miss
logit foreign i.miss#c.rep78 weight
*------------------- end example -------------------
(For more on examples I sent to the Statalist see:
http://www.maartenbuis.nl/example_faq )


Hope this helps,
Maarten


On Mon, Apr 8, 2013 at 12:42 PM, Nick Baker <nb419@cam.ac.uk> wrote:
> Thank you very much for your help Maarten,
> I have implemented the first part of the method, but am unsure about the
> second aspect: "Add the lag an interaction term between the indicotor
> variable and the lagged variable". Do I use the interaction terms as
> regressors instead of the lagged returns or do I use the sum of the
> interaction terms and the lagged returns as regressors?
>
> Many thanks for you help,
>
> Nick Baker
>
> On Apr 8 2013, Maarten Buis wrote:
>
>> On Sun, Apr 7, 2013 at 4:17 PM, Nick Baker wrote:
>>>
>>> I am estimating the model with 12 lagged monthly returns. The
>>> coefficients on lagged returns are assumed to be equal across funds, with
>>> the exception of those cases where a fund has fewer than 12 historical
>>> returns.
>>>
>>> In such a case the authors set the coefficients on lagged returns equal
>>> to zero if the corresponding return is unobserved.
>>>
>>> I am entirely stumped as to how to implement such a procedure in Stata
>>> and was wondering if anyone else had encountered a similar problem.
>>
>>
>> Say the problem is just the 12th lag, then you create an indicator
>> variable whether or not the lag should be included in your model. Than
>> replace that lagged variable with any non-missing constant (e.g. 0)
>> when it should not be included in your model. Add the lag an
>> interaction term between the indicotor variable and the lagged
>> variable. The coefficient for the lagged term when it should not be
>> included in your model is automatically dropped from your model (i.e.
>> set to 0) because the lagged variable will than be constant and thus
>> colliniear with the constant. You can easily extend this strategy to
>> the 11th lag, 10th lag, etc.
>>
>> However, I don't think this is a good idea. You added 12 lags because
>> you wanted to adjust your coefficients for 12 lags. Now you get some
>> mixture of adjusting for 12 lags, 11 lags, 10 lags, etc., so who knows
>> what your results mean?
>>
>> Hope this helps,
>> Maarten
>>
>> ---------------------------------
>> Maarten L. Buis
>> WZB
>> Reichpietschufer 50
>> 10785 Berlin
>> Germany
>>
>> http://www.maartenbuis.nl
>> ---------------------------------
>> *
>> *   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/



-- 
---------------------------------
Maarten L. Buis
WZB
Reichpietschufer 50
10785 Berlin
Germany

http://www.maartenbuis.nl
---------------------------------
*
*   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/