1[2 {"id":"C1","text":"The AR(p)+seasonal-dummy model has lower out-of-sample RMSE and MAE than the seasonal-naive forecast at h=1, 2 and 4 over the 40 origins.","results_keys":["rmse_snaive_h1","rmse_model_h1","rmse_snaive_h2","rmse_model_h2","rmse_snaive_h4","rmse_model_h4","mae_snaive_h1","mae_model_h1","mae_snaive_h2","mae_model_h2","mae_snaive_h4","mae_model_h4"],"status":"supported","caveat":"Point estimates on 40 origins; current-vintage data."},3 {"id":"C2","text":"Relative RMSE improvement over seasonal naive is about 23% (h=1), 14% (h=2) and 19% (h=4).","results_keys":["rel_improvement_rmse_h1","rel_improvement_rmse_h2","rel_improvement_rmse_h4"],"status":"supported","caveat":"Influenced by 2021-2023 inflation surge, though ranking persists when those targets are excluded."},4 {"id":"C3","text":"At h=4 the mean squared-error loss differential (2.83) is not statistically distinguishable from zero with a Bartlett HAC SE using 3 lags (DM=1.36, p=0.17).","results_keys":["loss_diff_mean_h4","loss_diff_se_h4","hac_lags_h4","hac_kernel_h4","dm_stat_h4","dm_pvalue_h4"],"status":"supported","caveat":"Inference sensitive to HAC kernel/lag choice in a 40-observation sample; longer non-prespecified Bartlett lags give larger statistics."}5]6