I am trying to estimate a system of two equations where the error terms are correlated and the
coeficients of the regressors are also correlated. To do that I am using the Seemingly unrelated
regression code, sureg. However I am having trouble trying to program the constraints. The problem
is that I have the following system of equations:
y1= ac + bX + cZ + (d1 + d2)W + e1 +e2 (equation 1)
y2 =nc + gD + kcZ + kd2W + e3 + ke2 (equation 2)
where ac, nc are the constants, (b,c,d1,d2) are the parameters to be estimated in equation 1, (y1)
is the dependent variable for equation 1, (X,Z,W) are the independent variables for equation 1 and
e1+e2 is the error term for equation 1.
For equation 2:
y2 - dependent variable
(g,k) are the parameters to be estimated. Note that (c,d2) are also parameters in equation 1 and 2
at the same time. So the coefficients are correlated in both equations.
(D,Z,W) are the dependent variables for equation 2. Note also that some dependent variables are the same
e3+ke2 is the error term, which is also correlated with equation 1.
So I am using sureg (Seemingly unrelated regression) and I need to program the constraints but for
that I need to specify variables like d1 that are just going to be estimated after and are part of
coeficients and not generated from the data, so I cant use gen. So there is a way to define global
variables to program the constraints and then these variables will be estimated inside the sureg
regressions?
thanks,
Priscilla Medeiros
Note that some of the dependent variables are the
Thanks,
Priscilla Medeiros
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/