spa_test() now follows Hansen (2005). It previously tested in the wrong
direction (rejecting when the benchmark was best), scaled the statistic by
an extra sqrt(n), and did not use Hansen's consistent recentring. The
hl2005 article is updated: GARCH(1,1) is not rejected under any
realised-variance proxy at the 5% level.
cspa_test() now reports the correct p-value when the null is rejected at
every level (was 1, now 0) or at none (was 0, now 1). The reject
decision was unaffected.
Added tests for cspa_test(), csms(), cspa_test_plot(),
do_prewhiten() and get_legendre().
Refined the package title and DESCRIPTION text per CRAN feedback:
removed the redundant "Tools for" prefix from the title and spelled out
acronyms (MSFE, SPA) on first use.
Initial release. Provides a unified toolkit for out-of-sample forecast dominance testing.
dm_test() — Diebold-Mariano (with Harvey, Leybourne, and Newbold
(1997) small-sample correction).cw_test() — Clark and West (2007) MSFE-adjusted statistic.mse_f_test() — McCracken (2007) MSE-F equal-MSFE statistic for
nested out-of-sample comparison.enc_new() — ENC-NEW encompassing test of Clark and McCracken (2001).gw_test() — Giacomini and White (2006) conditional equal predictive
ability test.spa_test() — Hansen (2005) test for superior predictive ability.cspa_test() — Li, Liao, and Quaedvlieg (2022) test for conditional
superior predictive ability, with the accompanying cspa_test_plot()
visualisation and csms() confidence set for the most superior method.uspa_mh_test() and aspa_mh_test() — uniform and average
multi-horizon SPA tests of Quaedvlieg (2021), implemented with a
moving-block bootstrap.ivx_wald() — Kostakis, Magdalinos, and Stamatogiannis (2015) IVX-Wald
statistic for predictive regressions with persistent regressors.qll_hat() — Elliott and Muller (2006) qLL statistic for testing
parameter time variation.Bundled cm2001, hl2005, gw2006, wg2008, quaedvlieg2021,
llq2022, llq2022_jnj, llq2022_uv_cspa, nrtz2014, rossi2006,
rrz2016, and rz2013 datasets used by the replication vignettes.