The effect of linear filters on dynamic time series with structural change

Abstract

Quite often, when parametric models are tested for structural change, they are fitted to filtered series instead of raw data. Many filters, like those associated with the X-11 seasonal adjustment program, have smoothing properties. Hence, they have a tendency to disguise structural instability. The paper analyzes, both theoretically and via Monte Carlo simulations, the effect of linear filtering on the statistical properties of several tests involving structural change. Historical series of economic activity covering the Great Depression are used to study and illustrate the sensitivity of some tests to the application of seasonal adjustment filters.