# Contract (F2-smoke-01)
- Question: linear association between y and x in a synthetic firm panel.
- Estimand: population slope β of the linear projection y = α + βx + ε (a projection coefficient, not necessarily causal). Meaningful under i.i.d. sampling of firms (independence across firms, arbitrary dependence within firm), finite second moments, Var(x)>0; cluster SE inference requires a reasonably large number of clusters.
- Outputs: out/results.json (task_id, beta_hat, alpha_hat, se_ols, se_cluster, beta_ci_low, beta_ci_high, n_analysed, n_clusters, feasibility, feasibility_reason, data_issues, n_input_rows, random_seed), out/report.md (Question; Data and exclusions; Method; Results; Uncertainty; Limitations and open issues), out/analysis.py, out/claims.json.
- Information rule: only data/panel.csv.
- Invalidity: missing/garbled columns, too few clusters (<~20) for cluster-robust inference, no variation in x, using other inputs.