A0617
Title: Bootstrapping out-of-sample predictability tests with real-time data
Authors: Michael McCracken - Federal Reserve Bank of St. Louis (United States) [presenting]
Silvia Goncalves - McGill University (Canada)
Yongxu Yao - McGill University (Canada)
Abstract: A block bootstrap approach is developed for out-of-sample inference when real-time data is used to produce forecasts. In particular, we establish its first-order validity for West-type tests of predictive ability in the presence of regular data revisions. This allows the user to conduct asymptotically valid inference without having to estimate the asymptotic variances derived in an extension of the original test when data is subject to revision. Monte Carlo experiments indicate that the bootstrap can provide a satisfactory finite sample size and power even in modest sample sizes.