A0237
Title: Testing for a breakdown in forecast accuracy under long memory
Authors: Philipp Sibbertsen - University of Hannover (Germany) [presenting]
Jannik Kreye - Leibniz University of Hannover (Germany)
Abstract: A test is proposed to detect a forecast accuracy breakdown in a long memory time series. The proposed method uses a double sup-Wald test against the alternative of a structural break in the mean of the out-of-sample squared error loss series. To address the problem of estimating the long-run variance under long memory, a robust estimator is applied. The breakpoint is determined by a long memory robust CUSUM test. We provide theoretical and simulation evidence on the memory transfer from the time series to the forecast residuals. The finite sample size and power properties of the test are derived in a Monte Carlo simulation. We find that only the fixed forecasting scheme leads to a monotonic power function. The practical relevance of the method is demonstrated by detecting a forecast failure in German electricity prices.
