# Contract — F2-conf-04

Question. Okun's law in US quarterly data: slope β of Δu_t = α + β g_t + e_t, 1960Q1–2019Q4 (240 quarters).

• u_t = mean of 3 monthly UNRATE values in quarter t (pp); Δu_t = u_t − u_{t−1}.
• g_t = 100·ln(GDPC1_t/GDPC1_{t−1}), quarter-on-quarter, NOT annualised.
• 1960Q1 differences use 1959Q4 levels. Estimands. β (linear-projection slope, OLS), α, R². Interpretation as a projection coefficient requires only finite second moments and stationarity of (Δu, g); HAC inference requires weak dependence (mixing) of g_t·e_t. Uncertainty. conventional iid SE (baseline) and Newey–West Bartlett HAC SE with justified bandwidth; 95% CI from HAC SE; okun_negative_5pct = β̂<0 and |β̂/se_hac|>1.96. Outputs. out/results.json keys: task_id, beta_hat, alpha_hat, r_squared, se_iid, se_hac, hac_lags, beta_ci_low, beta_ci_high, okun_negative_5pct, mean_growth, first_quarter, last_quarter, n_analysed, feasibility, feasibility_reason, data_issues, n_input_rows, random_seed. Also out/report.md (6 sections), out/analysis.py, out/claims.json, out/figure1.png, out/exclusions.csv. Availability rules. Only data/ files; current vintage; no observations outside 1960Q1–2019Q4 except 1959Q4 levels for differencing; do not extend window. Invalidating conditions. Missing monthly UNRATE or GDPC1 values within 1959Q4–2019Q4; incomplete quarters; checksum mismatch; using annualised growth.