# Contract (P0)

Question: At end of month t (48 origins 2022-05..2026-04), how accurately can Canadian monthly real GDP y/y growth for month t+h (h=1,3) be forecast using only data published by then; does a model adding LFS employment y/y growth beat the naive forecast? Estimands: out-of-sample RMSE and MAE (pp) per horizon for (a) naive = gdp_yoy(t-2) and (b) one chosen model; relative improvement 1 - RMSE_model/RMSE_naive per horizon. Meaningful under: current-vintage data (simulates publication timing only, not first releases); fixed origin set; expanding window. Information rules: at origin t, GDP columns usable only for months <= t-2; employment only <= t-1. Applies to estimation sample, model selection, transforms, plug-in values; training targets must be observed (<= t-2). Outputs: out/forecasts.csv (192 rows: origin_date,target_date,horizon,forecast,actual,model,last_target_date_used,last_predictor_date_used); out/results.json (keys per TASK.md table); out/report.md (6 sections); out/analysis.py; out/claims.json; plus state files, exclusions.csv. Invalid/impossible if: any forecast uses data beyond the availability limits; targets missing for some origin (last target 2026-04+3=2026-07, available); data hash mismatch.