A0627
Title: Multistep forecast averaging with stochastic and deterministic trends
Authors: Mohitosh Kejriwal - Purdue University (United States)
Linh Nguyen - Purdue University (United States)
Xuewen Yu - Fudan University (China) [presenting]
Abstract: A new approach to constructing multistep combination forecasts in a nonstationary framework with stochastic and deterministic trends is presented. Existing forecast combination approaches in the stationary setup typically target the in-sample asymptotic mean squared error (AMSE) relying on its approximate equivalence with the asymptotic forecast risk (AFR). Such equivalence, however, breaks down in a nonstationary setup. Combination forecasts are developed based on minimizing an Accumulated Prediction Errors (APE) criterion that directly targets the AFR and remains valid whether the time series is stationary or not. It is shown that the performance of APE-weighted forecasts is close to that of the optimal, infeasible combination forecasts. Monte Carlo experiments are used to demonstrate the finite sample efficacy of the proposed procedure relative to Mallows/Cross-Validation weighting that targets the AMSE. An application to forecasting US macroeconomic time series demonstrates the relevance of the proposed method in practice.
