Detecting Relevant Changes in Time Series Models

Abstract

Summary Most of the literature on change point analysis by means of hypothesis testing considers hypotheses of the form H0:θ1=θ2 versus H1:θ1≠θ2, where θ1 and θ2 denote parameters of the process before and after a change point. The paper takes a different perspective and investigates the null hypotheses of no relevant changes, i.e. H0:‖θ1−θ2‖⩽Δ, where ‖·‖ is an appropriate norm. This formulation of the testing problem is motivated by the fact that in many applications a modification of the statistical analysis might not be necessary, if the difference between the parameters before and after the change point is small. A general approach to problems of this type is developed which is based on the cumulative sum principle. For the asymptotic analysis weak convergence of the sequential empirical process must be established under the alternative of non-stationarity, and it is shown that the resulting test statistic is asymptotically normally distributed. The results can also be used to establish similarity of the parameters, i.e. H1:‖θ1−θ2‖⩽Δ, at a controlled type 1 error and to estimate the magnitude ‖θ1−θ2‖ of the change with a corresponding confidence interval. Several applications of the methodology are given including tests for relevant changes in the mean, variance, parameter in a linear regression model and distribution function among others. The finite sample properties of the new tests are investigated by means of a simulation study and illustrated by analysing a data example from portfolio management.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 63 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.