generated regressor bias

[wardb]

If anyone has code for implementing the estimation procedure described in the following paper

Pagan, Adrian, 1984, "Econometric Issues in the Analysis of Regressions with Generated Regressors," International Economic Review, No. 25, pp. 22X-A1

I'd be most grateful to obtain a copy.

Thanks in advance.

[TomDoan]

I believe that that has largely been superseded by Murphy & Topel, "Estimation and Inference in Two Step Econometric Models", JBES, 1985, pp 370-379. A quick read through Pagan would seem to indicate that all of his cases are special cases of Murphy & Topel. An example of the latter is

grnp304.rpf

(2.12 KiB) Downloaded 201 times

with data file

TableF8-1.txt

(3.37 KiB) Downloaded 141 times

This is from Greene's 6th edition. The corresponding section in Greene's 5th is section 17.7. This is a bit more complicated than a two-step linear estimation since the first step is a logit, but since the gradient for the logit likelihood is only the density x the regressors, it's not that much different.

[wardb]

Thanks very much, Tom, for the helpful information. I'll be alright now.