Dear Stata listers,
I need to estimate the coefficients - wage ealsticities - for a system
of two equations - female and male labor demand equations - and the
constraints are: (i) cross-wage elasticities in the two equtions
should be equal and (ii) the wage elasticities in each equation should
add up to one. My data set includes a panel of 132094 firms over an
eight-year period.
If I pool the data, I can use the following method:
constraint define 1 2 3 ...
global eq1 (depvar indvars)
global eq2 (depvar indvars)
sureg $eq1 $eq2, const(1 2 3)
However, if I want to use xt commands to control for fixed effects, I
found the following SJ paper:
Blackwell, J. Lloyd (2005) "Estimation and testing of fixed-effect
panel-data systems" The Stata Journal 5, Number 2, pp. 202-207
In this paper the commands XTPCSE and XTGLS are introduced to estimate
a system of equations using panel data. From the example provided in
the paper I understand that these commands are good for panels with
small number of panel ids as the example introduces dummies to control
for fixed effects. This is not possible in my case as I have 132094
panel ids. I can use XTREG with FE option treating the data as
suggested by the paper - putting the two equations data into one
equation. However, I still need to address two issues: (i) the minor
issue is to get rid of general constant term when XTREG, FE is used
and instead introduce two separate constants for each panel - there is
no [noconstant] option for XTREG; (ii) the major issue is to impose
the constraints. I have all variables in log and would be grateful if
someone could guide me on imposing them implicitly.
Furthermore, there is another command, XTMIXED, which I am reading an
SJ paper about it now:
Marchenko, Yulia (2006) "Estimating variance components in Stata" The
Stata Journal 6, Number 1, pp. 1-21
It seems that I will face the same two issues I mentioned above, if I
use this command.
Any help would be much appreciated.
I need to estimate the coefficients - wage ealsticities - for a system
of two equations - female and male labor demand equations - and the
constraints are: (i) cross-wage elasticities in the two equtions
should be equal and (ii) the wage elasticities in each equation should
add up to one. My data set includes a panel of 132094 firms over an
eight-year period.
If I pool the data, I can use the following method:
constraint define 1 2 3 ...
global eq1 (depvar indvars)
global eq2 (depvar indvars)
sureg $eq1 $eq2, const(1 2 3)
However, if I want to use xt commands to control for fixed effects, I
found the following SJ paper:
Blackwell, J. Lloyd (2005) "Estimation and testing of fixed-effect
panel-data systems" The Stata Journal 5, Number 2, pp. 202-207
In this paper the commands XTPCSE and XTGLS are introduced to estimate
a system of equations using panel data. From the example provided in
the paper I understand that these commands are good for panels with
small number of panel ids as the example introduces dummies to control
for fixed effects. This is not possible in my case as I have 132094
panel ids. I can use XTREG with FE option treating the data as
suggested by the paper - putting the two equations data into one
equation. However, I still need to address two issues: (i) the minor
issue is to get rid of general constant term when XTREG, FE is used
and instead introduce two separate constants for each panel - there is
no [noconstant] option for XTREG; (ii) the major issue is to impose
the constraints. I have all variables in log and would be grateful if
someone could guide me on imposing them implicitly.
Furthermore, there is another command, XTMIXED, which I am reading an
SJ paper about it now:
Marchenko, Yulia (2006) "Estimating variance components in Stata" The
Stata Journal 6, Number 1, pp. 1-21
It seems that I will face the same two issues I mentioned above, if I
use this command.
Any help would be much appreciated.
Comment