Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • How to include non linear constraints in a seemingly unrelated regression (SUR) model?

    I am a graduate student. I would like to estimate a SUR model of 6 equations with 2 non-linear constraints. The constraint command doesn't work because it is only useful for linear constraints. I would appreciate some help in how to incorporate non linear constraints (such as the one described below) in a regression model.

    My unit of observation is a household h, and my dataset has around 1100 observations. Both my independent and dependent variables are continuous. I want to estimate a seemingly unrelated regression (SUR) of 6 equations for each household:
    W1= a1+ b1X + e1
    W2 = a2 + b2X + e2
    W3 = a3 + b3X + e3
    W4 = a4 + b4X + e4
    W5 = a5 + b5X + e5
    W6 = a6 + b6X + e6

    The SUR will be estimated subject to the two non-linear constraints:

    Constraint 1:
    b1/ ( b1 + b2 + b3) = b4/ (b4 + b5 + b6)

    Constraint 2:
    b2/ ( b1 + b2 + b3) = b5/ (b4 + b5 + b6)

    I have explained above the most simplified version of my model. I found some help online with respect to interval constraints on this link: https://www.stata.com/support/faqs/s...l-constraints/
    But I am not sure if that is the right approach for me and if yes, how can the code can be modified for my question?
Working...
X