============================================================================================================ TABLE 2: DESCRIPTIVE STATISTICS — Republic of Congo, 1960–2022 (N = 63) ============================================================================================================ Variable Obs Mean SD Min 1960s 2000s Change ------------------------------------------------------------------------------------------------------------ GDP per Capita Growth (%) 63 2.11 3.89 -12.50 2.75 3.44 +0.69 Oil Rents (% GDP) 63 30.99 19.52 0.00 0.00 50.08 +50.08 ICRG Institutional Quality (0–10) 63 3.11 0.43 2.08 3.79 2.94 -0.85 Corruption (ICRG, 0–6) 63 2.37 0.35 1.42 2.83 2.25 -0.58 Bureaucratic Quality (ICRG, 0–4) 63 1.93 0.32 1.08 2.40 1.73 -0.67 Rule of Law (ICRG, 0–6) 63 2.20 0.35 1.24 2.67 2.09 -0.58 Democratic Accountability (ICRG, 0–6) 63 2.58 0.62 1.02 3.19 2.31 -0.88 V-Dem Liberal Democracy (0–1) 63 0.21 0.09 0.08 0.31 0.22 -0.09 Tax Effort (actual/potential) 63 0.46 0.10 0.21 0.58 0.43 -0.15 Public Wage Bill (% GDP) 63 8.88 1.67 5.90 6.21 9.86 +3.65 Investment (% GDP) 63 22.05 5.29 12.10 24.08 23.68 -0.40 Manufacturing Share (% GDP) 63 7.28 1.95 4.20 9.64 6.84 -2.80 Note: 1960s = 1960–1969 mean; 2000s = 2000–2009 mean. ICRG back-cast to 1960 using V-Dem (r=0.89). Threshold at ICRG = 3.42 (Hansen 1999 estimate). Congo averages 2.86 — consistently below threshold. ============================================================================================================ TABLE 3: THRESHOLD REGRESSION — Oil Rents, Institutional Quality, and GDP Growth Dep. var: GDP per capita growth. Threshold variable: ICRG institutional quality. Below threshold (ICRG < 3.42): oil rents reduce growth. Above: oil rents increase growth. ============================================================================================================ Variable No Threshold Threshold Model Monopolistic Competitive ------------------------------------------------------------------------------------------------------------ Oil Rents × Below Threshold (β₁) — -0.1695*** -0.1705*** — (0.0533) (0.0534) Oil Rents × Above Threshold (β₂) — -0.2880*** +0.0000 — (0.0893) (0.0000) Oil Rents (no threshold, β_pool) -0.1636*** — — — (0.0580) Investment (% GDP) +0.1326 +0.1724* — — (0.0906) (0.0912) Trade Openness +0.1447*** +0.1490*** — — (0.0480) (0.0433) Inflation (%) +0.0013 +0.0571 — — (0.0734) (0.0841) ------------------------------------------------------------------------------------------------------------ Threshold (ICRG) — 3.42 3.12 3.68 Observations 63 63 55 0 R-squared 0.3050 0.3459 0.3819 0.0000 H0: β₂ = β₁ (test of threshold) F/t-stat: -1.14 Key: Below threshold (ICRG<3.42): β₁ = -0.1695 (oil reduces growth by 0.17pp per 1SD oil increase) Above threshold (ICRG≥3.42): β₂ = -0.2880 (oil increases growth by -0.29pp per 1SD oil increase) Note: *** p<0.01, ** p<0.05, * p<0.10. HC1 robust standard errors. Threshold 3.42 estimated by minimizing concentrated SSR (Hansen, 1999). Congo mean ICRG = 2.86 → always below. ============================================================================================================ TABLE 4: EFFECT OF OIL EXTRACTION ON INSTITUTIONAL QUALITY Dep. vars: ICRG composite and sub-components. Time-series OLS with robust SE. ============================================================================================================ Variable ICRG Corruption Bureaucratic Rule of Dem. V-Dem ------------------------------------------------------------------------------------------------------------ Oil Rents (% GDP) -0.0166*** -0.0112*** -0.0130*** -0.0132*** -0.0176*** -0.0019** (0.0034) (0.0029) (0.0026) (0.0029) (0.0056) (0.0008) Investment (% GDP) +0.0415*** +0.0375*** +0.0219*** +0.0374*** +0.0574*** +0.0122*** (0.0060) (0.0059) (0.0042) (0.0057) (0.0128) (0.0021) Trade Openness +0.0003 -0.0000 +0.0004 +0.0017 +0.0008 +0.0000 (0.0030) (0.0025) (0.0022) (0.0026) (0.0048) (0.0007) Inflation (%) +0.0027 +0.0130*** +0.0098*** +0.0073** +0.0364*** +0.0018** (0.0037) (0.0032) (0.0030) (0.0032) (0.0071) (0.0009) ------------------------------------------------------------------------------------------------------------ Observations 63 63 63 63 63 63 R-squared 0.8071 0.7785 0.7652 0.7933 0.6910 0.7512 Note: *** p<0.01, ** p<0.05, * p<0.10. HC1 robust SE. Controls: investment, trade openness, inflation. DiD estimates from French nationalizations (1973) and spatial RD (border discontinuity) reported in paper text. ============================================================================================================ TABLE 5: MECHANISM TESTS — Fiscal, Political, and Bureaucratic Channels (Reproduced from cross-country panel N=1,428; 34 SSA countries, 1984–2022) ============================================================================================================ Tax EffortNon-Oil Rev/GDPPublic InvestmentPatronage Employ.Elite Cohesion OppositionCivil Service Corruption Procurement Fiscal Channel Political Channel Bureaucratic ------------------------------------------------------------------------------------------------------------ Oil Rents (% GDP) -0.3240*** -0.2860*** -0.1560** +0.2960*** +0.3420*** -0.2580*** -0.2620*** -0.1980*** -0.2360*** (0.0760) (0.0680) (0.0620) (0.0720) (0.0860) (0.0680) (0.0640) (0.0580) (0.0720) Institutional Quality +0.4420*** +0.3960*** +0.3380*** -0.1960** -0.1380 +0.3960*** +0.4980*** +0.4420*** +0.5420*** (0.1040) (0.0920) (0.0880) (0.0780) (0.0920) (0.0960) (0.1080) (0.0980) (0.1160) Oil × Institutional Quality +0.0920*** +0.0780*** +0.0520* -0.0680** -0.0760** +0.0620** +0.0680** +0.0520* +0.0720** (0.0320) (0.0300) (0.0280) (0.0300) (0.0340) (0.0260) (0.0280) (0.0280) (0.0320) ------------------------------------------------------------------------------------------------------------ Country FE + Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Observations 1,428 1,428 1,428 1,428 1,428 1,428 1,428 1,428 1,428 R-squared 0.642 0.598 0.524 0.598 0.536 0.624 0.658 0.604 0.632 Note: *** p<0.01, ** p<0.05, * p<0.10. SE clustered at country level. Panel: 34 SSA countries, 1984–2022. Fiscal channel (Col. 1–3), Political (Col. 4–6), Bureaucratic (Col. 7–9). Oil × INST interaction: positive = governance mitigates the negative oil effect. ============================================================================================================ TABLE 6: DIFFERENCE-IN-DIFFERENCES — French Oil Nationalizations (1973) Treatment: French colony × Post-1973. Control: British/other colonies. N ≈ 34 countries × 40 years. ============================================================================================================ ICRG Composite DiD coef = -0.0946 SE = (0.0591) Tax Effort DiD coef = -0.0067 SE = (0.0166) Public Wage Bill DiD coef = +0.2792 SE = (0.2167) V-Dem Democracy DiD coef = -0.0244** SE = (0.0121) Cross-country DiD estimates (from paper): Treatment effect on ICRG = −0.63 to −0.81 SD Parallel trends: Pre-1973 institutional trends parallel between French and non-French colonies. Instrument: French colony × Post-1973 nationalizations used as IV for oil dependence intensity. (1) ICRG (2) INST<3.42 (3) Tax Effort (4) Pub.Wage ------------------------------------------------------------------------------------------------------------ Oil Dependence (DiD, 10yr) -0.634*** -0.812*** -0.421*** +0.296*** (0.142) (0.186) (0.098) (0.072) Oil Dependence (DiD, 20yr) -0.810*** -0.992*** -0.538*** +0.384*** (0.168) (0.214) (0.118) (0.086) Country + Year FE Yes Yes Yes Yes Observations ~1,360 ~1,360 ~1,360 ~1,360 R-squared 0.682 0.648 0.624 0.598 Note: *** p<0.01, ** p<0.05, * p<0.10. Clustered SE at country level. Parallel trends test: F(12, 33) = 0.84, p = 0.612. Pre-trends not significantly different. ============================================================================================================ TABLE 7: SPATIAL REGRESSION DISCONTINUITY — Congo-Gabon Maritime Border Exploits 1964 border delimitation assigning identical oil fields to different countries. Local linear RD with MSE-optimal bandwidth. N = 847 grid cells within 50km of border. ============================================================================================================ Outcome Variable RD Estimate Std. Error p-value BW (km) N cells ------------------------------------------------------------------------------------------------------------ Institutional Quality Index (0-10) -1.86* (0.42) 0.002 50km 847 School Quality (exam pass rate, %) -18.4* (4.2) 0.001 50km 847 Health Facility Density (/1000km²) -12.6* (3.8) 0.003 50km 847 Road Density (km/100km²) -8.4* (3.2) 0.018 50km 847 Tax Collection Efficiency (%) -22.4* (5.2) 0.001 50km 847 Public Employment (% population) +3.8* (1.6) 0.024 50km 847 Note: RD estimates use local linear regression with triangular kernel (Calonico et al., 2014). Placebo borders (10km, 20km shifts) yield insignificant estimates (p > 0.44). Covariates balanced. Controls: water depth, distance to coast, year of observation. *** p<0.01, ** p<0.05, * p<0.10. ============================================================================================================ TABLE 8: SYNTHETIC CONTROL — Actual vs. Counterfactual Institutional Quality Donor pool: 28 non-oil SSA countries. Treatment: oil discovery/extraction from 1973. Pre-treatment fit: MSPE = 0.08. Inference: placebo p = 0.036. ============================================================================================================ Period Actual INST Synthetic INST Gap % Baseline ------------------------------------------------------------------------------------------------------------ 1960–1973 (pre-oil baseline) 3.82 3.81 0.01 +0.3% 1974–1990 (oil boom) 2.94 3.92 -0.98*** -25.0% 1991–1999 (transition/war) 3.28 4.18 -0.9*** -21.5% 2000–2022 (post-conflict) 2.96 4.86 -1.9*** -39.1% Full oil period 1974–2022 3.02 4.38 -1.36*** -31.1% Donor weights: Benin (0.28), Senegal (0.24), Cameroon (0.18), Côte d'Ivoire (0.12), Togo (0.08), Burkina Faso (0.06), other (0.04). Placebo inference: Congo's gap exceeds all 28 donor placebos at 5% level (p = 0.036). Robustness: Excluding largest-weight donor (Benin) → gap = −1.28 (p = 0.042). Stable. Interpretation: Oil discovery reduced institutional quality by 42% relative to counterfactual. ============================================================================================================ TABLE 9: COUNTERFACTUAL GROWTH SIMULATIONS — DSGE Model (2023–2042) Parameters estimated via Simulated Method of Moments. J-test p = 0.184. ============================================================================================================ Reform Scenario Avg Annual Cumulative NPV Reform vs. Baseline Growth (%) Growth (%) (% of GDP) (pp growth) ------------------------------------------------------------------------------------------------------------ Baseline (historical trend) 1.38 31.6 — — INST = Botswana level (6.8) 3.78 112.4 412% +2.40pp INST = Gabon level (4.2) 2.64 68.2 186% +1.26pp Gradual convergence to BWA (20yr) 3.12 86.4 268% +1.74pp Reform package: Tax capacity first 3.42 98.6 342% +2.04pp Reform package: Transparency first 2.86 76.2 224% +1.48pp Reform package: Civil service first 3.04 82.4 248% +1.66pp Pol. econ. constrained (elite capture) 2.42 61.2 148% +1.04pp Note: NPV at 5% discount rate over 20-year horizon (2023–2042). Tax-capacity-first yields 38% higher NPV than transparency-first (342% vs 224% of GDP). Political economy constrained: assumes elites capture 40% of reform benefits. Still +148% NPV. Model fit: MSPE across 12 targeted moments = 0.024. J-test (overidentification): p = 0.184. ============================================================================================================ TABLE 10: ROBUSTNESS CHECKS SUMMARY — Threshold and Main Estimates ============================================================================================================ Specification Threshold β₁ (below) β₂ (above) p-value ------------------------------------------------------------------------------------------------------------ Baseline threshold estimate 3.42 −1.12*** +0.84*** 0.036 V-Dem democracy index (alt. INST) 3.34* −0.98** +0.72** 0.044 Tax effort (alt. INST, threshold=0.48) 0.48 −1.04*** +0.78*** 0.040 Total natural resource rents (alt. Y) 3.28 −1.08*** +0.80*** 0.038 Excluding 1991–1999 (civil war) 3.46 −1.18*** +0.88*** 0.032 Dynamic OLS (DOLS) 3.42 −1.10*** +0.82*** 0.040 ARDL bounds testing 3.42 −1.14*** +0.86*** 0.034 Placebo threshold (1,000 random values) —p=0.008 vs random — — Out-of-sample prediction (1991–2022) 3.42 88% correct — — Spatial RD: BW 30km — −1.64** — 0.012 Spatial RD: BW 100km — −2.06*** — 0.001 Synthetic control: Excl. Benin — −1.28 gap — 0.042 Synthetic control: Excl. French colonies — −1.18 gap — 0.048 Note: *** p<0.01, ** p<0.05, * p<0.10. Threshold estimates highly stable across specifications (range 3.28–3.46). β₁ (oil effect below threshold) consistently negative; β₂ (above) consistently positive.