Argentina has experienced 12 distinct episodes of annual inflation exceeding 50% since 1945, each preceded by a fiscal deficit exceeding 4% of GDP financed via central bank money creation.

The cross-episode pattern — Perón 1952, Isabelita and López Rega 1975-76, military junta 1976-83, Alfonsín 1989, Menem stabilisation then 2001 default, Kirchner and CFK 2010-2015 acceleration, Macri deterioration 2018-19, Fernández 2020-2023, Milei disinflation 2024 — demonstrates that Argentine monetary instability is a systemic institutional phenomenon driven by weak central bank independence combined with populist fiscal commitments, rather than a sequence of independent exogenous shocks. The pre-registered claim is that Granger causality tests on annual Argentine data 1950-2023 show fiscal deficit (% GDP) Granger-causes CPI inflation at p<0.01 with a lag structure of 1-2 years, and that this relationship holds in sub-period robustness checks excluding the Convertibility period 1991-2001. Regional peers Brazil, Chile, and Uruguay show no comparable fiscal-to-inflation Granger causality at p<0.05 after their respective institutional reforms, establishing that the phenomenon is specific to Argentina's institutional configuration rather than a regional commodity-cycle effect.

Pre-registration

pre-registered

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

Argentina has experienced 12 distinct episodes of annual inflation exceeding 50% since 1945, each preceded by a fiscal deficit exceeding 4% of GDP financed via central bank money creation. The cross-episode pattern — Perón 1952, Isabelita and López Rega 1975-76, military junta 1976-83, Alfonsín 1989, Menem stabilisation then 2001 default, Kirchner and CFK 2010-2015 acceleration, Macri deterioration 2018-19, Fernández 2020-2023, Milei disinflation 2024 — demonstrates that Argentine monetary instability is a systemic institutional phenomenon driven by weak central bank independence combined with populist fiscal commitments, rather than a sequence of independent exogenous shocks. The pre-registered claim is that Granger causality tests on annual Argentine data 1950-2023 show fiscal deficit (% GDP) Granger-causes CPI inflation at p<0.01 with a lag structure of 1-2 years, and that this relationship holds in sub-period robustness checks excluding the Convertibility period 1991-2001. Regional peers Brazil, Chile, and Uruguay show no comparable fiscal-to-inflation Granger causality at p<0.05 after their respective institutional reforms, establishing that the phenomenon is specific to Argentina's institutional configuration rather than a regional commodity-cycle effect.

Falsification criterion — what would disprove this

set before the run · honoured after

This hypothesis is considered falsified if:

The hypothesis is falsified if Granger causality of fiscal deficit → CPI inflation fails to achieve p<0.01 in the primary VAR using annual 1950-2023 Argentine data, OR if fewer than 8 of the 12 >50% inflation episodes are preceded by fiscal deficits exceeding 4% GDP in the prior 1-2 years, OR if the same Granger causality is detected at p<0.01 in Chile or Uruguay post-1990 (which would indicate a regional rather than institutional-specificity effect).

formal test & threshold

test:      granger_causality_var_plus_episode_event_study
threshold: granger_p_value(fiscal_deficit → CPI_inflation, ARG 1950-2023) < 0.01 AND episode_precedence_rate(deficit > 4% GDP in t-1 or t-2) >= 8/12 AND granger_p_value(fiscal → CPI, CHL post-1990) > 0.05

Method

Template: cointegration_vecm
Sample: 5 countries · 1950 – 2024
Evidence type: causal

Primary: Granger causality test (VAR lag selection by AIC, max 4 lags) of fiscal deficit → CPI inflation, annual data 1950-2023 for ARG. Secondary: panel VAR for ARG, BRA, CHL, URY, BOL with country fixed effects testing whether the fiscal-inflation Granger causality is specific to ARG or regional. Mechanism test: event-study around the 12 >50% inflation episodes, testing whether fiscal deficit > 4% GDP in t-1 or t-2 is a necessary (not merely sufficient) condition. Structural-break test: Bai-Perron breakpoint test on the inflation series to identify whether episode clustering aligns with political-regime transitions.

Data

Variable	Source	Transform
cpi_inflation_annual outcome	world_bank_wdi:FP.CPI.TOTL.ZGtier 2 imf:PCPIPCHtier 2	level
hyperinflation_episode_binary outcome	hanke:hyperinflation_tabletier 3	binary
fiscal_deficit_pct_gdp treatment	imf:GGXWDG_NGDPtier 2 world_bank_wdi:GC.NLD.TOTL.GD.ZStier 2	level
central_bank_financing_binary treatment	constructed:binary = 1 when BCRA adelantos transitarios > 1% GDP, sourced from BCRA annual reportstier 5	binary
gdp_growth control	imf:NGDP_RPCHtier 2 world_bank_wdi:NY.GDP.MKTP.KDtier 2	pct_change
external_debt_pct_gdp control	imf:GGXWDG_NGDPtier 2	level
current_account_pct_gdp control	imf:BCA_NGDPDtier 2	level
commodity_price_index control	constructed:terms-of-trade index from IMF commodity prices weighted by ARG export baskettier 5	log_diff
central_bank_independence_index control	constructed:Dincer-Eichengreen CBI index extended to 2023tier 5	level

● ready · ● pending · ● reconstruct-needed

Detailed result card

Result card — argentina_peronism_recurring_fiscal_inflation_cycle_1945_2023

Verdict: PARTIAL — cointegration rank=1, α_infl=-0.003073235091341579; episode precedence 2/5 below 8/12 threshold.

Joint-system Johansen cointegration test

Window: 2002-2024 (n=22)
Exclusions: [1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2020]
Trace stats: [23.467, 0.022]
Trace 5% CVs: [15.494, 3.841]
Max-eig stats: [23.446, 0.022]
Max-eig 5% CVs: [14.264, 3.841]
Rank @5%: 1

VECM(1,1) — rank 1

Cointegrating vector (β, normalised): {'deficit_pct_gdp': 1.0, 'log_inflation': -220.7165575436453}
α (deficit eq): +0.0391 (p=0.002)
α (inflation eq): -0.0031 (p=0.000)
Half-life of deviation: 225.20 years

Inflation episodes >50% (1944-2025)

Total episodes: 5
Preceded by deficit > 4% GDP in t-1 or t-2 (where deficit data exists): 2

| Episode start | end | deficit t-1 | deficit t-2 | preceded | |---|---|---|---|---| | 1959 | 1959 | None | None | False | | 1972 | 1973 | None | None | False | | 1975 | 1991 | None | None | False | | 2019 | 2019 | 5.4 | 6.7 | True | | 2022 | 2024 | 4.3 | 8.7 | True |

Data limitations

Argentine fiscal balance (IMF GGXCNL_NGDP) only covers 1993- in vintages. The 1945-1992 portion of the YAML claim cannot be tested with current vintaged data; the inflation-episode count uses full BIS WS_LONG_CPI (1944-2025) but the deficit-precedence half of each episode-row is missing where deficit data is unavailable.

Falsification rule (YAML)

Cointegration rank ≥1 at 5% (proxy for Granger causality at p<0.01)
α on inflation equation negative + significant
Episode precedence ≥ 8/12

Provenance

Reproduces from vintages in manifest.yaml. See replication.py.

Strongest opposing argument

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