Testing for structural change in a long-memory environment

Long-memory time-series analysis is apt to be applied to economic time series which extend over many years, in which circumstances the possibility of structural breaks is likely to be entertained. Tests for a change in parameter values at a given time point are proposed in linear regression models with long-memory errors. Existing tests based on the assumption of serially independent or weakly dependent errors will typically be invalid in such an environment. The tests are derived in case of certain nonstochastic and stochastic regressors, and are given large-sample justification. A small Monte Carlo study of finite-sample behaviour is included.