Dear all,

I am having trouble using endogenous treatment model (eregress function).

Basically I want to look at whether searching for jobs through headhunters has implications for individuals’ move across functions (change in function). Obviously unobserved factors may affect whether individuals search for jobs through headhunters and also affect their move across functions. So, we believe that it is appropriate to treat search through headhunters as an endogenous treatment. I created an instrument, which is the proportion of other searches in that country that take place by search firm and are trying to run an endogenous treatment model using ‘eregress’ function. The problem I have encountered is that as I ran the model, the correlation between the error terms of the selection equation and outcome equation is nearly one. And I am not entirely sure the reason why. In the attached log file, the first model is what I think is the correct model. As you can see, it takes some time to converge and the correlation of the error terms is almost 1. In the second model, where we exclude all the controls from the selection equation, the correlation goes down. One reason for this might be because the instrument we are not using is not very strong.

But It would be great to hear from you why this might be the case. Thank you very much in advance.