(Joint with David F. Hendry)
We consider model selection for non-linear dynamic equations with possible location shifts, applied to re-analyze an empirical model of real wages in the UK over 1860-2004, updated to 2005-2011 for forecast evaluation. After an automatic search over more variables than observations based on a general class of non-linear-in-the-variables functions augmented by impulse-indicator saturation, simplified to a congruent terminal model, an encompassing test can be implemented against an investigator’s preferred non-linear function. As that is often a non-linear in the parameters threshold model, the approach can only be semi-automatic.