IESET.
Hypotheses·growth·korean_institutional_divergence_gdp_gap

Korean division 1945-present delivers a natural experiment: matched populations, geography, pre-war institutions; post-war institutional divergence explains the roughly 20x GDP-per-capita gap.

SUPPORTEDengine/runs/korean_institutional_divergence_gdp_gap

SUPPORTED — Maddison MPD2020 KOR/PRK GDP-pc ratio at 2023 is 33.0x (log-gap +3.50, threshold ≥ 3.00 for SUPPORTED). At 1943 the ratio was 0.68x (log-gap -0.38). Cumulative log-growth: KOR +3.21, PRK -0.67; gap +3.88 log-points over 80 years.

confidence cueThis is a clear pass for the claim as written. It still applies only to this sample, period, and method.

policy briefNeeds review

In ordinary language

Over a long period, do more market-oriented institutions translate into higher income or productivity, once the comparison looks beyond a single success story?

plain answer

The data clearly moved in the predicted direction. Maddison MPD2020 KOR/PRK GDP-pc ratio at 2023 is 33.0x (log-gap +3.50, threshold ≥ 3.00 for SUPPORTED).

why it matters

Growth claims can look convincing in single success stories. This test asks whether the pattern survives a broader comparison.

how the test works

It compares 2 country or place units from 1953 to 2023, using a descriptive design.

what was measured
What we checked
  • Log income pc cost-of-living adjusted
  • Log income pc constant
  • Prk income pc estimate
what this does not prove

A single test is not the whole truth. It narrows the claim under a specific sample, time period, and method. Strong policy conclusions need the pattern to survive nearby tests, alternative data, and serious objections.

verification

No evidence packet has been generated yet.

Results

engine/runs/korean_institutional_divergence_gdp_gap
Loading chart…

Who has skin in the game — schools predicting on this

1 school list this hypothesis as a test of their position. The chips below are school-level scoreboard outcomes, not a second hypothesis verdict.

hypothesis verdict vs scoreboard outcome

The banner verdict judges this hypothesis as written. The scoreboard asks whether each school's polarity-corrected prediction was right. Raw status is not a school win: SUPPORTED supports schools that needed SUPPORTED, but refutes schools that needed REFUTED.

New Institutional Economicswins if SUPPORTEDschool supported

Pre-registration

pre-registered
first-spec commit 4c8ce8e · 2026-07-18T22:11:21Z

Korean division 1945-present delivers a natural experiment: matched populations, geography, pre-war institutions; post-war institutional divergence explains the roughly 20x GDP-per-capita gap.

Falsification criterion — what would disprove this

set before the run · honoured after

This hypothesis is considered falsified if:

PRIMARY (dispositive): the hypothesis is SUPPORTED if log_gdppc(KOR, 2023) − log_gdppc(PRK, 2023) on Maddison MPD2020 is at least ln(20) ≈ 2.996 (i.e. the realised KOR/PRK ratio is at or above the claim's "~20x"). REFUTED if the same log-gap is below ln(10) ≈ 2.303 (less than half the claimed gap, per the original stub language). PARTIAL otherwise (direction correct but magnitude short). INFORMATIVE (colour, non-gating): the pre-window |log-gap| at the realised baseline year (1953 if available, else latest pre-division co-observed year) should be < 0.30. Larger pre-window gaps weaken the "matched starting conditions" premise of the natural-experiment framing but do not flip the verdict. METHOD_VALID: at least one common KOR/PRK observation each in MPD2020 plus a 2023 endpoint via the Maddison Korea-extension; otherwise emit `inconclusive (data gap on …)`.

formal test & threshold
test:      korean_bilateral_gdppc_ratio_dispositive_threshold
threshold: PRIMARY: log_gap_endpoint = log_gdppc(KOR, 2023) − log_gdppc(PRK, 2023); SUPPORTED if log_gap_endpoint ≥ 2.996 (≥ 20× ratio); refuted if log_gap_endpoint < 2.303 (< 10× ratio); partial in between.

Method

Template
descriptive
Clustering
none
Sample
2 countries · 1953 – 2023
Evidence type
descriptive

Bilateral KOR-PRK descriptive comparison. Single-statistic divergence test (~20x GDP-per-capita ratio by 2023). Note: a more comprehensive pattern-match version of this hypothesis exists at north_south_korea_development_divergence_1953_present (multi_metric_checklist). This hypothesis retains the simpler descriptive framing.

Data

VariableSourceTransform
log_gdp_pc_ppp
outcome
maddison:gdppc_ppptier 3
log
log_gdp_pc_constant
outcome
world_bank_wdi:NY.GDP.PCAP.KDtier 2
log
prk_gdp_pc_estimate
outcome
bank_of_korea:north_korea_gni_per_capitatier 1
log
life_expectancy_at_birth
outcome
world_bank_wdi:SP.DYN.LE00.INtier 2
level
log_population
control
world_bank_wdi:SP.POP.TOTLtier 2
log

● ready  · ● pending  · ● reconstruct-needed

Detailed result card

Korean institutional divergence and the GDP-per-capita gap

Verdict: SUPPORTED — Maddison MPD2020 KOR/PRK GDP-pc ratio at 2023 is 33.0x (log-gap +3.50, threshold ≥ 3.00 for SUPPORTED). At 1943 the ratio was 0.68x (log-gap -0.38). Cumulative log-growth: KOR +3.21, PRK -0.67; gap +3.88 log-points over 80 years.

Headline numbers

  • Endpoint (2023, Maddison): KOR $41,800 vs PRK $1,265; ratio 33.0x; log-gap +3.50.
  • Pre-window (1943, Maddison): KOR $1,686 vs PRK $2,462; ratio 0.68x; log-gap -0.38.
  • Cumulative log-growth 1943→2023: KOR +3.21, PRK -0.67; gap +3.88 log-points over 80 years (annualised: KOR +4.01%/yr; PRK -0.83%/yr).
  • Life expectancy at birth (WDI, 2024): KOR 83.6y vs PRK 73.7y; gap +9.9 years.
  • Bank of Korea cross-check (DPRK GNI reconstruction, 2023): KOR/PRK ratio 40.3x.

Threshold applied

  • PRIMARY (dispositive): log_gdppc(KOR, 2023) − log_gdppc(PRK, 2023) determines the verdict. SUPPORTED if ≥ ln(20) ≈ 3.00 (i.e. ≥ 20× ratio, matching the claim's '~20x'); refuted if < ln(10) ≈ 2.30 (less than half the claimed gap, per the spec); partial otherwise.
  • INFORMATIVE: pre-window |log-gap| at 1943 should be < 0.30 (realised 0.38; informative pass: False). | Component | Threshold | Realised | Pass | |---|---:|---:|:---:| | Endpoint log-gap (SUPPORTED) | ≥ 3.00 | +3.50 | yes | | Endpoint log-gap (not refuted) | ≥ 2.30 | +3.50 | yes | | Pre-window |log-gap| (informative) | < 0.30 | 0.38 | no |

Method

Bilateral KOR–PRK descriptive comparison. Primary GDP source is Maddison MPD2020 (Bolt & van Zanden 2020), the canonical long-run cross-country panel; values are in 2011 international $. The Maddison Korea-extension series (2020–2023) provides the endpoint year. Bank of Korea's DPRK GNI reconstruction (2020–2023) is reported as a cross-check. WDI is used for life expectancy (NY.GDP.PCAP.KD has no PRK observations and therefore cannot serve as the primary GDP series). Verdict logic is dispositive on the endpoint ratio: < 10× → refuted (less than half the claimed ~20×); ≥ 20× → supported; between → partial. The pre-window gap at 1953 colours method validity (large pre-1953 gaps would weaken the natural-experiment framing) but does not gate the verdict.

Caveats

  • This is a descriptive bilateral comparison, not causal identification. The pre-1945 administrative division of Korea was not random with respect to industrial endowments — Japan-built heavy industry and most hydroelectric capacity sat in the north, favouring the DPRK at independence. The fact that the south overtook anyway strengthens the institutionalist read but does not establish causation.
  • Pre-window choice: the spec's nominal baseline is 1953 (post-armistice). MPD2020 has no PRK observations 1944–1989, so the replication uses 1943 (latest_pre_division_1943) as the realised pre-window. This is the best published anchor for the 'matched starting conditions' premise; readers who want a 1953 baseline must rely on Bank of Korea's reconstruction, which only covers 2020–2023.
  • DPRK GDP is hard to measure. Maddison's PRK series is itself a reconstruction (Lee, Bolt, others), so figures past 1990 have wider error bars than the KOR series.
  • Post-1945 paths differ in foreign aid, security alignment, and 1960s policy choices, so divergence is the institutions × aid × alignment bundle. Claim authors who treat this as a clean natural experiment overstate identification (per the spec's disclosure block).

Sources

  • Maddison MPD2020 (mpd2020).
  • Maddison Korea extension (mpd2020_korea_extension, 2020–2023).
  • Bank of Korea DPRK GNI reconstruction (dprk_gdp_reconstruction, 2020–2023).
  • World Bank WDI NY.GDP.PCAP.KD (KOR cross-check; PRK all-NaN).
  • World Bank WDI SP.DYN.LE00.IN (life expectancy at birth).

Notes

Seeded from an institutionalist (AJR) claim that the Korean division is a natural experiment vindicating institutions-cause-growth. Descriptive divergence test; data quality on PRK side requires human review before pre-registration.

Authored framework. Read the transparency note.