I'm working on a project investigating the impact of childhood malnutrition on scholastic performance using a panel dataset of 5 survey rounds, 2000 households distributed in 5 regions. I am trying to conduct a Chow test after my analysis to work out whether all coefficient slopes should be estimated separately for the 5 regions separately. One way to do this would be to add interaction terms for each variable and test a joint significance test that all the interaction terms are equal to 0. However, as I have quite a few variables (30-40), this is quite unwieldy with 5 regions and would require 4 separate interaction terms per variable in my regression. Another way to do this for cross-sectional analysis is to estimate the equations separately, and then obtain the respective RSS (residual sum of squares) for each equation, which, when summed, would give the total RSS of the unrestricted model RSS. I could then compare this to the restricted model where all regions are pooled via an F test.

I am using a panel random effects model with instrumental variables and clustered robust standard errors. (XTIVREG, re, vce(cluster panelid)) I am using the 'small' command as this allows me to obtain T and F statistics instead of Z and chi-squared values. However, on trying out the 'etreturn list' command, I have not been able to identify the RSS with panel random effects and instrumental variables (XTIVREG) and can only obtain the R^2 value. Is there a way to obtain the RSS for panel random effects?