Hi,

I am using firm-level survey data and I am interested in using the Heckman approach to test for selection. My dependent variable for the selection equation is a firm's import status (importer/non-importer) and that of the main equation is import intensity which as a %, lies between [0,1].

Since the 'svy' command cannot be used the -twostep- option, I am using the ml approach to Heckman. I am still in the process of familiarising myself with the model and from what I understand, the 'lambda' in the bottom panel of the output is the coefficient on the inverse mills' ratio, and is also the product of 'rho' and 'sigma.' However, I am pretty confused on the following two questions

(1) But Stata only reports the coefficient and standard errors of 'lambda', and not the t-values. In that case, how do I test whether that coefficient is significant or not? At first I thought it was a problem with my code but when I consulted the Stata handbook, z-values aren't reported for lambda! (Page 8, https://www.stata.com/manuals13/rheckman.pdf) See screenshot below:

(2) I am also confused on which variable is the coefficient of the IMR? Is it lambda, or /lnsigma (whose t-values are present, which makes testing for significance easier)

(3) My survey data Heckman mle estimation also does not automatically report the "LR test of independent equations" for rho. How can I get around it?Array

I am clearly missing some very obvious things here so I would be very grateful if for any insights! Thank you