# Contract — F1-conf-03

Question. Association between firm leverage at end of year t and default in year t+1, using only info known at end of t. Estimand. b1 in OLS LPM default_next ~ 1 + leverage + cash_ratio + log_sales, on firm-years with observed outcome; SE clustered by firm_id. Meaningful as a descriptive conditional association (not causal); LPM assumes linear conditional mean; cluster SE assume independence across firms and enough clusters (~300). Outputs. out/results.json (task_id, n_analysed, n_firms, predictors_used, variables_excluded, beta_leverage, se_leverage_cluster, r2, default_rate, feasibility, feasibility_reason, data_issues, n_input_rows, random_seed); out/analytic.csv (firm_id, year, predictors, default_next); out/exclusions.csv (step, rule, n_before, n_after, n_removed, rationale; chained); out/report.md (6 sections); out/analysis.py; out/claims.json. Information rules. Predictor allowed only if known at end of t (codebook). rating_2025 is known in 2025 and built from realised defaults through 2022 -> leakage, must be excluded. Rows with missing default_next (origin 2022) excluded, not imputed. Only data/firm_panel.csv. Invalidity. Using rating_2025; imputing outcomes; double-counting duplicated rows; no observed outcomes -> infeasible.