IESET.
Hypotheses·monetary·household_debt_minsky_cycle_2008

Post-Keynesian / Minsky reading: across advanced economies in the decade preceding the 2008 GFC, the rise in household credit-to-GDP was the leading indicator of subsequent banking-sector distress.

The hypothesis tests, in a panel of OECD countries 1995-2010, whether the change in household credit-to-GDP over a five-year window predicts Laeven-Valencia-classified banking-crisis onset 2007-2010, with a predictive coefficient material in magnitude and surviving conditioning on standard macro controls. A market-liberal "subprime-as-policy- failure" reading attributes the same pattern to a US-specific Community Reinvestment Act / GSE channel; the cross-country panel test discriminates by showing the relationship holds across countries with very different mortgage policy regimes.

REFUTEDengine/runs/household_debt_minsky_cycle_2008

REFUTED — coef=+0.3377 (sign opposite claim -), p=0.00971

confidence cueThis test cuts against the claim as written or misses its pre-declared threshold.

policy briefClear refutation

In ordinary language

In plain terms, this asks whether household credit income 5y change is actually linked to better or worse banking crisis indicator from 1995 to 2010.

plain answer

The data did not support the prediction. coef=+0.3377 (sign opposite claim -), p=0.00971

why it matters

This matters because monetary claims should change belief only when they survive a pre-declared empirical test.

how the test works

It compares 21 country or place units from 1995 to 2010, using a panel fe design, with fixed effects for country.

what was measured
What changed
  • Household credit income 5y change
  • Household credit income level
What we checked
  • Banking crisis indicator
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

7 input datasets, 0 unresolved missing series, provenance status: reproducible hash verified.

Results

engine/runs/household_debt_minsky_cycle_2008
1007550250199520032010USAGBRIRLESPNLDDNKFRA
illustrative sketch · run pending
No coefficients yet. When the model fires, this chart will show banking_crisis_indicator across 21 sampled countries over 1995–2010.
The shapes above are stylised — none of the lines are real data.
Placeholder for household_debt_minsky_cycle_2008. Published chart will be generated from engine/runs/household_debt_minsky_cycle_2008/chart_data.json.

Who has skin in the game — schools predicting on this

2 schools 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.

Marxian Political Economywins if SUPPORTEDschool refutedPost-Keynesianwins if SUPPORTEDschool refuted

Pre-registration

registration ordering unverified
first-spec commit 4c8ce8e · 2026-07-18T22:11:21Z
run generated · 2026-06-29T17:54:16Z
Run timestamp predates this path's first git-add commit (rebase, rename, or pre-git local run). Spec hash is still the path's first-add commit — not repository HEAD — but ordering is not a clean pre-registration proof.

Post-Keynesian / Minsky reading: across advanced economies in the decade preceding the 2008 GFC, the rise in household credit-to-GDP was the leading indicator of subsequent banking-sector distress. The hypothesis tests, in a panel of OECD countries 1995-2010, whether the change in household credit-to-GDP over a five-year window predicts Laeven-Valencia-classified banking-crisis onset 2007-2010, with a predictive coefficient material in magnitude and surviving conditioning on standard macro controls. A market-liberal "subprime-as-policy- failure" reading attributes the same pattern to a US-specific Community Reinvestment Act / GSE channel; the cross-country panel test discriminates by showing the relationship holds across countries with very different mortgage policy regimes.

Falsification criterion — what would disprove this

set before the run · honoured after

This hypothesis is considered falsified if:

Not supported if (a) the coefficient on household-credit-GDP five-year change is statistically zero or negative as a crisis predictor (p > 0.10), OR (b) the predictive power disappears after conditioning on real house-price change (i.e. household credit predicts only because it is collinear with the housing boom and adds no independent information), OR (c) the AUC of household-credit change as a crisis predictor is below 0.65 in the OECD pre-2008 panel, indicating it is not a useful predictor relative to alternatives like current-account deficit.

formal test & threshold
test:      panel_logit_household_credit_change_2007_2010_crisis_with_house_price_conditioning
threshold: coef(household_credit_change) > 0 at p < 0.10 AND coefficient survives real-house-price conditioning AND AUC of household-credit change >= 0.65

Method

Template
panel_fe
Fixed effects
country
Clustering
country
Sample
21 countries · 1995 – 2010
Evidence type
associational

Panel logit / linear-probability regression of 2007-2010 banking-crisis-onset indicator on lagged household-credit-GDP five-year change, conditioning on real house-price change and macro controls. Country FE only (year FE would absorb the crisis- onset year). Robustness: Schularick-Taylor JST long-run banking crisis sample for cross-validation, and ROC-curve report of AUC for credit-change vs alternatives as crisis predictor.

Data

VariableSourceTransform
banking_crisis_indicator
outcome
bis:WS_CREDIT_GAPtier 2
indicator
household_credit_gdp_5y_change
treatment
world_bank_wdi:FS.AST.PRVT.GD.ZStier 2
diff_5y
household_credit_gdp_level
treatment
world_bank_wdi:FS.AST.PRVT.GD.ZStier 2
level
real_house_price_5y_change
control
bis:WS_SPPtier 2
diff_5y
current_account_balance_gdp
control
world_bank_wdi:BN.CAB.XOKA.GD.ZStier 2
level
real_gdp_growth_5y_avg
control
world_bank_wdi:NY.GDP.MKTP.KD.ZGtier 2
avg_5y
short_term_interest_rate
control
oecd:OECD.SDD.STEStier 2
level
log_population
control
world_bank_wdi:SP.POP.TOTLtier 2
log

● ready  · ● pending  · ● reconstruct-needed

Detailed result card

Result card — household_debt_minsky_cycle_2008

Verdict: REFUTED — coef=+0.3377 (sign opposite claim -), p=0.00971

Pre-registration

  • Claim: Post-Keynesian / Minsky reading: across advanced economies in the decade preceding the 2008 GFC, the rise in household credit-to-GDP was the leading indicator of subsequent banking-sector distress. The hypothesis tests, in a panel of OECD countries 1995-2010, whether the change in household credit-to-GDP over a five-year window predicts Laeven-Valencia-classified banking-crisis onset 2007-2010, with a predictive coefficient material in magnitude and surviving conditioning on standard macro controls. A market-liberal "subprime-as-policy- failure" reading attributes the same pattern to a US-specific Community Reinvestment Act / GSE channel; the cross-country panel test discriminates by showing the relationship holds across countries with very different mortgage policy regimes.
  • Falsification rule: Not supported if (a) the coefficient on household-credit-GDP five-year change is statistically zero or negative as a crisis predictor (p > 0.10), OR (b) the predictive power disappears after conditioning on real house-price change (i.e. household credit predicts only because it is collinear with the housing boom and adds no independent information), OR (c) the AUC of household-credit change as a crisis predictor is below 0.65 in the OECD pre-2008 panel, indicating it is not a useful predictor relative to alternatives like current-account deficit.
  • Falsification test: panel_logit_household_credit_change_2007_2010_crisis_with_house_price_conditioning

Estimate

  • Method: linearmodels.PanelOLS
  • Coefficient (treatment): +0.3377
  • Std error: 0.1288
  • p-value: 0.00971
  • Observations: 162, countries: 15
  • Within R²: 0.547
  • Fixed effects: entity=True, time=False
  • Clustering: country

Variables resolved

  • bis:WS_CREDIT_GAP → banking_crisis_indicator (outcome, publisher=bis, n=1914)
  • world_bank_wdi:FS.AST.PRVT.GD.ZS → household_credit_gdp_5y_change (treatment, publisher=world_bank_wdi, n=9562)
  • world_bank_wdi:FS.AST.PRVT.GD.ZS → household_credit_gdp_level (treatment, publisher=world_bank_wdi, n=9562)
  • bis:WS_SPP → real_house_price_5y_change (controls, publisher=bis, n=2272)
  • world_bank_wdi:BN.CAB.XOKA.GD.ZS → current_account_balance_gdp (controls, publisher=world_bank_wdi, n=7621)
  • world_bank_wdi:NY.GDP.MKTP.KD.ZG → real_gdp_growth_5y_avg (controls, publisher=world_bank_wdi, n=13897)
  • world_bank_wdi:SP.POP.TOTL → log_population (controls, publisher=world_bank_wdi, n=14447)

Variables missing data

  • oecd:OECD.SDD.STES,DSD_KEI@DF_KEI,4.0 (controls, name=short_term_interest_rate) — vintage not on disk

Generated by scripts/run_panel_fe.py at 2026-06-29T17:54:16+00:00

Strongest opposing argument

Every hypothesis ships with its charitable opposing argument. The framework earns credibility by handling objections at their strongest, not weakest.

Notes

Candidate, not pre_registered. On promotion, secure a registered fetcher for Laeven-Valencia banking-crises (IMF WP) and JST cross- validation; confirm BIS WS_TC dataflow URN; pin the ROC-curve AUC threshold using a held-out sub-sample.

Authored framework. Read the transparency note.