A1083
Title: Detection of breaks in weak location time series models with quasi-Fisher scores
Authors: Christian Francq - CREST and University Lille III (France) [presenting]
Jean-Michel Zakoian - CREST (France)
Lorenzo Trapani - University of Leicester (United Kingdom)
Abstract: Based on Godambe's theory of estimating functions, a class of cumulative sum is proposed, CUSUM, statistics to detect breaks in the dynamics of time series under weak assumptions. First, a parametric form for the conditional mean is assumed but makes no specific assumption about the data-generating process (DGP) or even about the other conditional moments. The CUSUM statistics considered depend on a sequence of weights that affect their asymptotic accuracy. Data-driven procedures are proposed for the optimal choice of the sequence of weights, in the sense of Godambe. Modified versions of the tests are also proposed that allow to detect breaks in the dynamics even when the conditional mean is misspecified. The results are illustrated using Monte Carlo experiments and real financial data.