Pre-registration
Germany's Schuldenbremse (constitutional debt brake, 2009) produced lower debt-to-GDP trajectories than comparable-economy fiscal-rule-absent peers over 2009–2019, without output loss relative to the Eurozone mean.
Falsification criterion — what would disprove this
This hypothesis is considered falsified if:
PRIMARY 1 (debt discipline, dispositive): Germany's Δgross-general-govt-debt-to-GDP from 2008 to 2019 must be at least 10pp LOWER than the unweighted donor-pool mean Δdebt-to-GDP over the same window. PRIMARY 2 (no output cost, dispositive): Germany's cumulative log-real-GDP-per-capita growth 2008-2019 must be at least 90% of the donor-pool unweighted mean cumulative log-growth. SUPPORTED if BOTH primaries hold. partial if PRIMARY 1 holds but PRIMARY 2 fails (debt discipline came at non-trivial output cost). REFUTED if PRIMARY 1 fails (Schuldenbremse failed at its stated job). INFORMATIVE: pre-trend log-growth 2000-2008 gap |Germany - donor mean| ≤ 5pp for clean parallel-trend; if violated, verdict text flags caveat. METHOD_VALID: imf:GGXWDG_NGDP and wdi:NY.GDP.PCAP.KD available with at most 3 of 10 donor countries missing.
formal test & threshold
test: deu_vs_donor_panel_debt_and_output_2008_2019 threshold: PRIMARY 1: (deu_delta_debt - donor_mean_delta_debt) ≤ -10pp. PRIMARY 2: deu_log_growth / donor_mean_log_growth ≥ 0.90 (or, when donor mean ≤ 0, deu - donor mean ≥ -5pp absolute).
Method
- Template
synth_did- Clustering
none- Sample
- 11 countries · 2000 – 2019
- Evidence type
- associational
Synthetic difference-in-differences with Germany as the treated unit and a donor pool of fiscal-rule-absent advanced economies (FRA, ITA, ESP, GBR, etc.). Outcomes: debt-to-GDP path and real GDP per capita 2009-2019. Tests Germany's debt path vs synthetic counterfactual without an output penalty.
Data
| Variable | Source | Transform |
|---|---|---|
gross_general_govt_debt_pct_gdp outcome | imf:GGXWDG_NGDPtier 2 | level |
log_real_gdp_pc outcome | world_bank_wdi:NY.GDP.PCAP.KDtier 2 | log |
gen_govt_balance_pct_gdp outcome | imf:GGXCNL_NGDPtier 2 | level |
post_schuldenbremse_indicator treatment | constructed:indicator = 1 for DEU, year >= 2010 (Schuldenbremse binding from 2010)tier 5 | indicator |
log_initial_gdp_pc control | world_bank_wdi:NY.GDP.PCAP.KDtier 2 | log |
trade_openness control | world_bank_wdi:NE.TRD.GNFS.ZStier 2 | level |
log_population control | world_bank_wdi:SP.POP.TOTLtier 2 | log |
● ready · ● pending · ● reconstruct-needed
Detailed result card
Schuldenbremse: debt discipline without output cost (2009-2019)
Verdict: SUPPORTED — Germany's debt-to-GDP rose by -6.5pp 2008-2019 vs donor-pool mean +23.4pp (-29.9pp differential, ≥10pp threshold met). Cumulative log-GDP-pc growth: Germany +12.6%, donor mean +8.2% (Germany 153% of donor mean, ≥90% threshold met). N=10 donor countries.
Summary
- N = 10 of 10 donor countries with complete coverage.
- Δdebt-to-GDP 2008-2019: Germany -6.5pp, donor-pool mean +23.4pp (differential -29.9pp; threshold ≤ -10pp).
- Cumulative log-GDP-per-capita growth 2008-2019: Germany +12.6%, donor mean +8.2% (ratio 153%; threshold ≥ 90%).
- Pre-trend (2000-2008): Germany +10.5% vs donor +11.5% (-0.9pp gap; within 5pp diagnostic band).
- Δfiscal balance (informative): Germany +1.6pp, donor mean +1.6pp.
Method
Before/after panel comparison: Germany (treated) vs unweighted mean of a 10-country donor pool of fiscal-rule-absent advanced Eurozone economies + UK. PRE=2008 (year before the constitutional amendment), POST=2019 (cleanest pre-COVID, pre-debt-brake-suspension year).
Outcomes:
- Δdebt-to-GDP (IMF GGXWDG_NGDP, level)
- Cumulative log-real-GDP-per-capita growth (WDI NY.GDP.PCAP.KD, log diff)
- Pre-trend log-growth 2000-2008 (parallel-trend diagnostic)
- Δgeneral-govt net lending / GDP (IMF GGXCNL_NGDP, informative)
Spec downgrade. Original spec called for synthetic-DID; the donor-
weight optimisation step is non-trivial and synth_did is not in the
project venv. Per HANDOFF_TO_RUN_AGENT.md allowance, downgraded to a
transparent before/after comparison with an explicit pre-trend
diagnostic. A real synth-DID would weight donors to match the pre-
period; this comparison instead reports the unweighted donor mean
and the pre-period gap.
Falsification thresholds
- PRIMARY 1 (debt): Germany Δdebt 2008-2019 must be ≥ 10pp lower than donor-pool mean.
- PRIMARY 2 (output): Germany cumulative log-growth 2008-2019 must be ≥ 90% of donor-pool mean.
- METHOD_VALID: at most 3 of 10 donors missing data.
- INFORMATIVE: pre-trend gap |2000-2008 Δlog-GDP-pc| ≤ 5pp for clean parallel-trend; if violated, verdict text flags caveat without overriding direction.
Data
- imf:GGXWDG_NGDP (gross general-government debt / GDP)
- imf:GGXCNL_NGDP (general-government net lending / GDP)
- world_bank_wdi:NY.GDP.PCAP.KD (real GDP per capita, constant USD)
Caveats
- Attribution to Schuldenbremse alone is contested. Confounds: ECB policy reaction post-2010, intra-Eurozone competitiveness gaps, German export demand boost from peripheral austerity, post-2010 Hartz IV labour-market effects on unit labour costs.
- Donor pool is a fixed Eurozone+UK convenience sample. Including non-Eurozone non-fiscal-rule advanced economies (USA, JPN) would shift the donor mean substantially.
- 2008 pre-period endpoint is the eve of the Great Recession; both Germany and donors entered POST with elevated debt due to crisis interventions, which the simple endpoint comparison absorbs into Δdebt for both sides equally.
- The output ratio at low / negative donor means becomes unstable; code falls back to a 5pp absolute-gap test in that regime.
Notes
Seeded from an Ordoliberal claim about Germany's Schuldenbremse delivering fiscal discipline without an output penalty 2009-2019. Synthetic-DID design; donor pool selection is the key sensitivity that requires human review.