PANELTHRESH: testing for threshold with panel model

[tavera]

I would like to test for threshold effects and estimate the threshold in a non dynamic panel as decribed in "Threshold effects in non dynamic panels: estimation, testing and inference" by Bruce Hansen, Journal of Econometrics, 93 (1999) p. 345-368. It seems to me that the procedure threshtest.src cannot be used in the case of a panel. Is there another procedure that permits to perform the adequate test ? Or, is it possible to use the threshtest.src by changing some codes ?
Thanks in advance

[TomDoan]

Actually, you can't use THRESHTEST for several reasons: first, it is testing for a complete sample split, while the Hansen 1999 paper is using a set of non-breaking variables as well. Also, THRESHTEST doesn't allow for fixed effects.

The following does the estimation of the one and two threshold models from the Hansen paper (without the bootstrapping for the F-tests). This is actually a correction to Hansen's Gauss and Matlab code which have a slight error. (They drop the final data point in each cross section).

panelthresh.prg

(2.68 KiB) Downloaded 515 times

invest.txt

(579.34 KiB) Downloaded 659 times

[tavera]

Than you very much for this efficient and helpful response