8 Specification Error and Structural Change

Everything so far assumed that the model is correctly specified: the right regressors, the right functional form, and the same coefficients for every observation. Applied work consists largely of finding out whether those assumptions hold. This section studies what goes wrong when variables are wrongly omitted or included, and develops the main tests of parameter constancy: the Chow forecast and break tests, recursive residuals, and the RESET test of functional form. Dummy variables, which are both a modelling tool and the machinery behind several of these tests, are treated along the way.

Questions on this section

Stuck on something here? Ask below and it stays attached to this topic.