Pre-registration
Chile and Venezuela began the 1999-2023 window at broadly comparable GDP per capita (PPP, constant international dollars). Over the subsequent 24 years Chile continued its institutional-market policy trajectory (fiscal rule, central-bank independence, export-diversified commodity model, secure property rights, trade openness) while Venezuela ran the Chavismo programme (FX controls, price controls, PDVSA politicisation, large-scale nationalisations, monetary-fiscal fusion). By 2023 the log GDP-per-capita-PPP gap between Chile and Venezuela exceeds 1.2 log-points (~230% relative). The bilateral comparison is one of the cleanest natural experiments in Latin American development economics because both economies are commodity-dependent, Spanish-colonial-legacy Latin American states that diverged on policy content, not on initial conditions.
Falsification criterion — what would disprove this
This hypothesis is considered falsified if:
PRIMARY (dispositive): falsified if log_gdp_pc_ppp(CHL, 2023) - log_gdp_pc_ppp(VEN, 2023) < 1.20 (i.e. Chile less than ~230% richer per capita PPP than Venezuela by 2023), OR if the cumulative log-growth gap 1999→2023 (Chile minus Venezuela) is less than 0.60 log-points. A sign reversal would be decisive refutation. INFORMATIVE (not gating): pre-1999 log-gap should be small (|gap_1999| < 0.30) to validate the "began comparable" framing. If the pre-window gap is already large, the post-window divergence is less attributable to policy.
formal test & threshold
test: chile_venezuela_bilateral_cumulative_gap threshold: PRIMARY: log_gdp_pc_ppp(CHL, 2023) - log_gdp_pc_ppp(VEN, 2023) >= 1.20 PRIMARY: cumulative_log_growth(CHL, 1999-2023) - cumulative_log_growth(VEN, 1999-2023) >= 0.60 INFORMATIVE: |log_gdp_pc_ppp(CHL, 1999) - log_gdp_pc_ppp(VEN, 1999)| < 0.30 METHOD_VALID: WDI NY.GDP.PCAP.PP.KD available for both countries across full window.
Method
- Template
descriptive- Clustering
none- Sample
- 2 countries · 1999 – 2023
- Evidence type
- descriptive
Bilateral cumulative log-growth comparison. Primary statistic is the 2023 log-gap vs 1999. The framework's interest is a clean, pre- registered natural experiment — not a DiD identification exercise. Magnitude is dispositive; the story is in per-capita-PPP levels.
Data
| Variable | Source | Transform |
|---|---|---|
log_gdp_pc_ppp outcome | world_bank_wdi:NY.GDP.PCAP.PP.KDtier 2 | log |
● ready · ● pending · ● reconstruct-needed
Detailed result card
Result card — Chile vs Venezuela divergence, 1999–2023
Verdict: SUPPORTED — 2023 log-gap (CHL−VEN) +2.30 (>=1.20). Cumulative growth gap 1999→2023 +1.50 log-points (>=0.60). Chile annualised +2.33%/yr; Venezuela -3.93%/yr.
Headline numbers
- Series used: WDI
NY.GDP.PCAP.KD(constant 2015 USD). - Chile log GDP-pc 1999 → 2023: 9.007 → 9.567; cumulative +0.559 log-points (~+75%); annualised +2.33%/yr.
- Venezuela log GDP-pc 1999 → 2023: 8.210 → 7.266; cumulative -0.944 log-points (~-61%); annualised -3.93%/yr.
- Bilateral log-gap (CHL − VEN): 1999 = +0.798; 2023 = +2.301.
- Cumulative growth gap 1999→2023: +1.503 log-points (~+350% relative).
- Pre-COVID robustness (1999→2019): growth gap +1.248 log-points.
Threshold applied
- PRIMARY (dispositive):
log_gdp_pc(CHL, 2023) − log_gdp_pc(VEN, 2023) >= 1.20ANDcumulative_log_growth(CHL, 1999-2023) − cumulative_log_growth(VEN, 1999-2023) >= 0.60. - INFORMATIVE:
|log_gdp_pc(CHL, 1999) − log_gdp_pc(VEN, 1999)| < 0.30(realised: |+0.798|; informative pass: False).
| Component | Threshold | Realised | Pass | |---|---:|---:|:---:| | Endpoint log-gap 2023 | >= 1.20 | +2.301 | yes | | Cumulative growth gap | >= 0.60 | +1.503 | yes | | Pre-window gap (informative) | < 0.30 | 0.798 | no |
Interpretation
This is a descriptive bilateral comparison; results are pattern matches, not causal identification. We have not constructed a counterfactual Venezuela or controlled for oil prices, terms-of-trade shocks, or US sanctions. The pre-registered claim is that the magnitude of the divergence is so large that policy content is the most plausible single explanation, but rigorous causal attribution would require a synthetic-control or bilateral DiD design. The Chavismo/Maduro programme overlaps with the 2014 oil price collapse and post-2017 US sanctions, so the gap reflects a bundle of self-inflicted policy and external shocks, not a clean policy-only experiment.
Sources
- World Bank WDI
NY.GDP.PCAP.KD(vintage NY.GDP.PCAP.KD@2026-04-28T125340Z.parquet). - Note:
NY.GDP.PCAP.PP.KDreturns NaN for VEN from 2011 onward in the local WDI vintage; the fallback to constant-USD preserves within-country log-growth comparability.
Steelman live concerns
See hypotheses/steelman/chile_vs_venezuela_divergence_1999_2023.md.
Strongest opposing argument
Every hypothesis ships with its charitable opposing argument. The framework earns credibility by handling objections at their strongest, not weakest.
Notes
This is part of the canonical natural-experiment tier (HYPOTHESIS_FRAMEWORK_AUDIT.md §E6 case #6). Runnable today with World Bank WDI data already on disk; primary spec is deliberately simple (bilateral descriptive comparison) to isolate the magnitude of the policy-content divergence from any methodological concerns.