Bookmark and Share

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: Logit VAR


From   Nima Darbari <[email protected]>
To   [email protected]
Subject   Re: st: Logit VAR
Date   Thu, 20 Feb 2014 10:31:20 +0100

Thanks for your hints Jorge, unfortunately in Stata binary dependent
variables are not permitted in a VAR but  starting by estimating a VAR
ignoring the binary nature of Y is a good starting point that I hadn't
thought about. About Logistic transformation, I think it is not useful
for binary data unless I first make portfolios of my firms and average
the figure inside each portfolio to get as close as possible to a
continuous variable restricted between zero and one. About sureg, I am
not sure if it is possible when using lagged dependent variables in
the RHS. The other option I have come up with is to use Stata SEM
features for estimating the system by accounting for the correlation
between errors and lagged dependent variables. Stata 12 doesn't accept
binary dependent variables in SEM however, and I would need Stata 13.
And finally I think the ultimate solution  is to use a self written
likelihood function for the system and maximize it.

On Sun, Feb 16, 2014 at 6:52 PM, Jorge Eduardo Pérez Pérez
<[email protected]> wrote:
> If I remember correctly vector autorregressions are available in Stata
> 12. And anyway, VARs can be estimated using -sureg-, or even using
> equation by equation OLS, although -sureg- would give you the proper
> standard errors.
>
> You could start by estimating the VAR ignoring the binary nature of Y,
> using an adjustment for heteroscedasticity, and checking that your
> forecasts for Y lie between 0 and 1.
>
> Or you could use -nlsur-, using a logistic transformation for your
> right hand side on the equation for Y.
>
>
>
>
> --------------------------------------------
> Jorge Eduardo Pérez Pérez
> Graduate Student
> Department of Economics
> Brown University
>
>
> On Fri, Feb 14, 2014 at 7:37 AM, Nima Darbari <[email protected]> wrote:
>> Since I have received no replies to my post which I admit has been
>> quite rare, I thought maybe I have not explained myself well. This is
>> the system I want to estimate:
>>
>> Y(t)=C1+Y(t-1)+X(t-1)+P(t-1)+Z(t-1)+e.Y
>> Z(t)=C2+Z(t-1)+X(t-1)+V(t-1)+Y(t)+e.Z
>> (Y is binary)
>>
>> On Thu, Feb 13, 2014 at 9:38 AM, Nima Darbari <[email protected]> wrote:
>>> I need to estimate a simultaneous dynamic (one lag of each one of the
>>> two dependent variables) system with a binary and a continuous
>>> dependent variable and several exogenous variables. I think it is
>>> possible in Stata 13 via VAR estimation but is there any solution for
>>> it in Stata 12?
>>> *
>>> *   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/
>
> *
> *   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/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index