Testing for structural change in regression with long memory processes

The paper considers tests for structural change in time series regression models where both regressors and residuals may exhibit long range dependence. The limiting distribution of the test statistic depends on unknown parameters. While the unknown parameters can be consistently estimated and asymptotic critical values obtained by simulation, the paper proposes an alternative approach of approximating the distribution of the test statistic by a bootstrap procedure. The asymptotic validity of bootstrap is shown and the performance of the testing procedure is examined in a simple Monte Carlo experiment.