Sieve bootstrap monitoring for change from short to long memory

- A monitoring procedure is proposed to sequentially detect changes from short to long memory processes.
- The proposed sieve bootstrap method is robust for many types of innovation processes.
- The new procedure exhibits competitive power than the existing retrospective test.

This paper proposes a variance ratio statistic to monitor changes from short to long memory processes. The asymptotic distribution is derived under the null hypothesis and the consistency of the monitoring procedure is proven under the alternative hypothesis. A sieve bootstrap approximation method is introduced to determine the critical values. Simulations indicate that the new procedure is quite robust for many types of innovation processes and performs better than the existing retrospective test.