Impose constraints on the cross-equation with a vector of zeros.

[Max Zur]

Hello Stata Community,
as part of a seminar i am supposed to replicate parts of the following paper: "Agnello, L.; Sousa, R.M, 2014: How does fiscal consolidation impact on income inequality? Rev. Income Wealth 2014, 60, 702-726." (https://onlinelibrary.wiley.com/doi/10.1111/roiw.12004).

The authors use the seemingly unrelated regression estimator (in Stata the command "sureg") to investigate the impact of fiscal consolidation on inequality.
So the authors use the following panel regression system:
y_it=X_itb+a_i+u_it
where y_it=(y_it^(gini)net, y_it^(gini)gross), X_it=(X₁^net, X₂^gross) and b=(b₁^net, b₂^gross).
Additionally dummy variables are used to capture the phases of fiscal consolidation.
Since, according to the authors, this consolidation only affects the gini-net, the authors impose cross-equation restrictions on the vector of coefficients b. As they write "Formally, the coefficients associated to the dummy variables in the gross income inequality equation are assumed to be equal to zero, that is, b₂^gross = {b₂|0_D} where 0_D is the vector of zeros."


I know the "constraint define" command, but I'm not sure how to specify exactly the constraint (vector of zeros for all b₂) correctly in Stata.

I really hope you can help me, and I'm so grateful for any help you can give me. Have a lovely weekend, everyone!

[George Ford]

would that be the same as leaving out gininet in the gross equation?