# Contract (P0)

Question. Causal effect of one extra training hour per employee (programme year) on next-year productivity growth (pp per hour), simulated firm data. Estimands. (1) Baseline: unadjusted OLS slope of productivity_growth on training_hours (association, not causal). (2) Target: coefficient on training_hours in a linear model conditioning on the pre-programme covariate skill_index; causal under selection-on-observables (conditional on skill_index, hours as-good-as random), linearity and homogeneous effect, no other confounders (no other variables exist). Outputs. out/results.json (naive_estimate, naive_ci_low/high, baseline_metric, model_metric, effect_estimate, effect_se, effect_ci_low/high, skill_index_coefficient, conclusion, task_id, feasibility, feasibility_reason, data_issues, n_input_rows, n_analysed, random_seed); out/figure1.png; out/table1.csv (model, estimate, se, ci_low, ci_high); out/report.md (6 sections); out/analysis.py; out/claims.json. Information rules. Only data/firms.csv. skill_index is pre-treatment (valid control); productivity_growth is post-treatment outcome. No imputation of other variables. Invalidity. Missing columns / no variation in hours -> infeasible. Unobserved confounders, nonlinearity in skill, or invalid treatment values would bias the estimate.