# Contract — F6-conf-02
**Question.** Did the Government of Canada 5-year benchmark yield change differently (avg one-day change, bp) on trading days flagged `post_cut = 1` (on/after the 2020-03-04 BoC emergency cut) than on the 120 trading days immediately preceding the first flagged day?
**Estimand.** mean(Δy | post_cut=1) − mean(Δy | 120 pre days), Δy_t = 100·(y_t − y_{t−1}) between consecutive file rows; 95% Welch (unequal-variance) interval. Descriptive difference in means; a causal reading requires no other shocks in the windows (COVID market turmoil makes that implausible) — interpret as a before/after contrast only.
**Outputs.** out/results.json (sample_start, sample_end, n_treated, baseline_metric, model_metric, effect_estimate, effect_ci_low, effect_ci_high, conclusion, task_id, feasibility, feasibility_reason, data_issues, n_input_rows, n_analysed, random_seed); out/figure1.png; out/table1.csv (year,n_days,mean_yield,mean_change_bp); out/report.md (6 sections); out/analysis.py; out/claims.json; plus protocol files.
**Information rules.** Only data/gc5y_daily.csv; changes between consecutive rows; post_cut used exactly as given, never redefined.
**Invalidity / impossibility.** If no row has post_cut=1 (or fewer than 2 such rows, or <2 pre rows) the estimand cannot be computed → conclusion `premise_invalid`, feasibility `infeasible`, inestimable keys null; no proxy estimate.