Selection Between Models Through Multi-Step-Ahead Forecasting.pdfVIP

Selection Between Models Through Multi-Step-Ahead Forecasting Tucker McElroy1 and David Findley1,∗ U.S. Census Bureau 1 Statistical Research Division, U.S. Census Bureau, 4600 Silver Hill Road, Washington, D.C. 20233-9100 ∗ Corresponding author E-mail address: david.f.findley@ Abstract We develop and show applications of two new test statistics for deciding if one ARIMA model provides significantly better h-step-ahead forecasts than another, as measured by the difference of approximations to their mean square forecast error. The two statistics differ in the variance estimates used for normalization. Both variance estimates are consistent even when the models considered are incorrect. Our main variance estimate is further distinguished by accounting for parameter estimation. The simpler variance estimate, which ignores estimation uncertainty, can be rather straightforwardly calculated for any pair of ARIMA models with the same differencing operator, and its broad consistency property offers improvement to what are known as tests of Diebold and Mariano (1995) type. 1 Statistical Research Division, U.S. Census Bureau, 4600 Silver Hill Road, Washington, D.C. 20233-9100 ∗ Corresponding author. E-mail address: david.f.findley@ Keywords. ARIMA models; Diebold-Mariano tests; Misspecified models; Model selection; Parameter estimation effects; Time series 1 Introduction In this article, we make several contributions to the technology of testing whether two not necessarily correct time series models for an observed series have equal or differing h-step-ahead forecasting ability as assessed by estimates of mean square h-step forecast error. This work is in the tradition of Meese and Rogoff (198