Prediction in locally stationary time series

Document pages: 42 pages

Abstract: We develop an estimator for the high-dimensional covariance matrix of alocally stationary process with a smoothly varying trend and use this statisticto derive consistent predictors in non-stationary time series. In contrast tothe currently available methods for this problem the predictor developed heredoes not rely on fitting an autoregressive model and does not require avanishing trend. The finite sample properties of the new methodology areillustrated by means of a simulation study and a financial indices study.