The Belsley, Kuh, and Welch Belsley et al. (1980) diagnostics are applied to the precision estimator in nonlinear models and the properties are explored via simulation. The performance of the diagnostics is examined specifically with the ordered probit model. The performance is examined under several parameterizations and under different degrees of collinearity. The results suggest that the identification of the model’s parameters is indeed related to collinearity of the data and by the parametric functional form of the model itself. The BKW diagnostics appear to be useful in the sense that serious problems with identification can be detected quite easily using the built in vif function of gretl.
Adkins, Lee C.
"Weak Identification in Nonlinear Econometric Models,"
Southern University College of Business E-Journal: Vol. 17:
3, Article 2.
Available at: https://digitalcommons.subr.edu/cbej/vol17/iss3/2