1[2 {"id":"C1","text":"Over 1960Q1-2019Q4, a 1 percentage point higher quarterly real GDP growth rate is associated with a 0.26 pp lower change in the unemployment rate (OLS Okun coefficient about -0.256).","results_keys":["effect_estimate","n_analysed","sample_start","sample_end"],"status":"supported","caveat":"Descriptive linear association, not a causal effect; current-vintage revised data."},3 {"id":"C2","text":"The Okun coefficient is significantly negative: the Newey-West (Bartlett, 4 lags) 95% CI is about [-0.326, -0.185].","results_keys":["effect_se","hac_lags","effect_ci_low","effect_ci_high","conclusion"],"status":"supported","caveat":"Normal-approximation CI; SE barely changes with 8 lags (effect_se_hac_lags8)."},4 {"id":"C3","text":"Real GDP growth explains a substantial part of quarterly unemployment changes: residual RMSE 0.250 pp vs 0.325 pp for the mean-only baseline (R^2 about 0.41).","results_keys":["model_metric","baseline_metric","r_squared_full"],"status":"supported","caveat":"In-sample fit only; no out-of-sample evaluation."},5 {"id":"C4","text":"The classical Chow test rejects joint equality of (alpha, beta) across 1960Q1-1983Q4 and 1984Q1-2019Q4 at 5% (F about 6.09, p about 0.003), so the relationship is classified as unstable.","results_keys":["chow_f_stat","chow_p_value","stability_conclusion"],"status":"supported","caveat":"Classical Chow assumes homoskedastic, serially uncorrelated errors with equal variance across regimes; residual variance falls from 0.079 to 0.048. A HAC-robust joint Wald test also rejects (p about 0.005)."},6 {"id":"C5","text":"The instability is not a change in the slope: subperiod slopes are nearly identical (-0.259 vs -0.273; HAC test of slope difference p about 0.85). The point estimates instead show a lower intercept after 1984 (0.254 vs 0.152).","results_keys":["beta_sub1","beta_sub2","hac_wald_slope_diff_p","intercept_sub1","intercept_sub2","hac_wald_intercept_diff_p"],"status":"supported","caveat":"The intercept difference alone is not significant under HAC (p about 0.26); only the joint test rejects, because intercept and slope estimates are correlated. Attributing the break to the intercept is an interpretation of point estimates."}7]8