Pre-registration
Monetary finance of fiscal deficits (central-bank balance-sheet expansion directed at sovereign obligations in the absence of independent policy rate adjustment) produces a three-order causal chain. FIRST-ORDER: the state pays ongoing expenditure without raising statutory taxes or issuing bonds to third parties — a legitimate short-run cash-flow achievement, acknowledged as working on its own terms during the window before second-order effects kick in. SECOND-ORDER: headline inflation rises, real wages fall, nominal savings erode, and relative price distortions accumulate. THIRD-ORDER: spontaneous dollarisation (ARG dolar blue, TUR dolarizasyon, EGY parallel dollar), black-market FX rates diverge from official, capital controls are imposed as a desperate response, and middle-class emigration accelerates. Cases: Argentina Fernández 2019-2023, Turkey Erdogan 2021-2023 unorthodox monetary regime, Venezuela 2013+, Sri Lanka 2020-2022, Ghana 2022, Egypt 2022-2024.
Falsification criterion — what would disprove this
This hypothesis is considered falsified if:
PRIMARY (dispositive, executable). For each of the six EM cases a country-specific monetary-finance intensification event year T is fixed in the script (ARG 2019, TUR 2021, VEN 2013, LKA 2020, GHA 2020, EGY 2022). FIRST-ORDER (cash delivered to government) is treated as confirmed by case construction — these are documented MF episodes. SECOND-ORDER: per case, mean CPI inflation (IMF PCPIPCH) over T+1..T+3 must be >= 2.0x the T-5..T-1 mean. THIRD-ORDER: per case, the LCU/PPP$ private conversion factor (WDI PA.NUS.PRVT.PP) at T+3 must be at least +30% above its T-1 value (cumulative depreciation). Chain is SUPPORTED if SECOND-ORDER passes in >=5/6 cases AND THIRD-ORDER passes in >=4/6 cases. Chain is REFUTED if neither gate is met. Partial otherwise. INFORMATIVE: per-case inflation multiple, depreciation %, and spec context series (gov debt/GDP, primary balance/GDP) are reported but do not gate the verdict. METHOD_VALID: at least 4/6 cases must have non-missing IMF PCPIPCH coverage in both pre and post windows. If >=3/6 cases have an inflation data gap, the verdict is inconclusive rather than refuted.
formal test & threshold
test: event_study_six_case_em_monetary_finance_chain threshold: PRIMARY: post/pre inflation mean >= 2.0 in >=5/6 cases AND cumulative LCU/PPP$ depreciation >= +30% in >=4/6 cases.
Method
- Template
event_study- Fixed effects
country, year- Clustering
country- Sample
- 6 countries · 2000 – 2024
- Evidence type
- causal
Primary: panel event study with country and year FE around monetary-finance intensification events per country. Secondary: local-projections-IV (lp_iv) using independence-of-central-bank legislative changes as instrument where available (TUR 2019- 2021 governor changes, ARG 2020 BCRA adelanto expansion).
Data
| Variable | Source | Transform |
|---|---|---|
central_bank_claims_on_government outcome | national central banks (BCRA, TCMB, BCV, CBSL, BoG, CBE) | log_real_local_currency |
headline_inflation outcome | indec:IPC (ARG)tier 2 tuik:CPI (TUR)tier 2 dcs:CPI (LKA)tier 2 gss:CPI (GHA)tier 3 capmas:CPI (EGY)tier 2 | yoy_pct_change |
real_wage_index outcome | indec:salariostier 2 tuik:wage_indextier 2 ine:salariostier 2 dcs:wages (LKA)tier 2 gss:wages (GHA)tier 3 capmas:wages (EGY)tier 2 | yoy_pct_change_real |
dollarisation_share_of_bank_deposits outcome | bcra:depositos_moneda_extranjera (ARG)tier 1 tcmb:doviz_mevduati (TUR)tier 2 bcv:depositostier 3 cbsl:deposits (LKA)tier 2 bog:deposits (GHA)tier 2 cbe:deposits (EGY)tier 2 | share |
parallel_vs_official_fx_rate_gap outcome | constructed:ARG blue-dollar from Ambito Financiero; TUR kara para; VEN dolartoday; LKA parallel rate; GHA parallel rate; EGY paralletier 5 | ratio |
capital_control_stringency outcome | constructed:capital_controls_event (annual); constructed: event indicator for controls imposition (ARG cepo Sep-2019, TUR KKM + capitier 5 | index |
middle_class_emigration_flow outcome | constructed:country-of-origin emigration stats from destination countries (OECD IMO, Spain INE migration register, Canada IRCC) for tier 5 | log_level |
monetary_finance_intensity treatment | constructed:central-bank claims on government as share of monetary base + direct-monetisation events (BCRA adelantos transitorios, Ttier 5 | share_and_event_indicator |
fiscal_primary_balance_pct_gdp control | imf:WEOtier 2 | level_pct_gdp |
terms_of_trade_shock control | unctad:commodity_price_indicestier 2 | yoy_pct_change |
global_interest_rate_proxy control | fred:DGS10tier 1 ecb:euribortier 1 | level_pct |
● ready · ● pending · ● reconstruct-needed
Detailed result card
Monetary-finance -> currency-collapse chain (six EM cases)
Verdict: partial — Currency depreciation confirmed in 4/6 cases, but inflation-acceleration second-order response missed: 4/6 (need 5/6). 3rd-order holds, 2nd-order weak.
Summary
- Per-case 2nd-order test: post-event 3-yr mean inflation >= 2x pre-event 5-yr mean. Pass: 4/6 (gate: >=5/6).
- Per-case 3rd-order test: cumulative LCU/PPP$ depreciation >=30% over T-1..T+3. Pass: 4/6 (gate: >=4/6).
Per-case detail
| Country | Event | Pre-infl | Post-infl | Mult | FX chg% | Infl pass | FX pass | |---------|-------|----------|-----------|------|---------|-----------|---------| | ARG | 2019 | 30.0 | 54.3 | 1.81 | nan% | N | N | | TUR | 2021 | 12.6 | 61.6 | 4.90 | 395% | Y | Y | | VEN | 2013 | 26.6 | 146.3 | 5.51 | nan% | Y | N | | LKA | 2020 | 4.3 | 22.9 | 5.34 | 68% | Y | Y | | GHA | 2020 | 12.8 | 27.0 | 2.11 | 96% | Y | Y | | EGY | 2022 | 13.7 | 26.0 | 1.90 | 88% | N | Y |
Method
Six EM cases of documented monetary-finance intensification, each with a country-specific event year T (ARG 2019, TUR 2021, VEN 2013, LKA 2020, GHA 2020, EGY 2022). For each case:
- Compute mean CPI inflation (IMF PCPIPCH) in pre-window T-5..T-1 and post-window T+1..T+3. The ratio is the 'inflation multiple'. Gate: ratio >= 2.0 -> 2nd-order chain confirmed for that case.
- Compute LCU/PPP$ ratio (WDI PA.NUS.PRVT.PP) at T-1 and T+3. The change is the cumulative depreciation. Gate: change >= +30% -> 3rd- order currency-collapse channel confirmed for that case.
Chain SUPPORTED if 2nd-order gate passes in >=5/6 cases AND 3rd-order gate passes in >=4/6. Asymmetric thresholds reflect the spec's expectation that inflation transmission is near-mechanical while currency response can be muted in cases with capital controls or managed-float regimes (where parallel-FX would be the cleaner test, but parallel-FX series for ARG/TUR/EGY/LKA/GHA are not on disk in this repo's vintage tree — only Venezuela has dolartoday).
Data
- imf:PCPIPCH (CPI inflation, annual % change)
- world_bank_wdi:PA.NUS.PRVT.PP (private-sector PPP conversion factor)
- imf:GGXWDG_NGDP (gov debt/GDP, context)
- imf:GGXCNL_NGDP (gov primary balance/GDP, context)
Caveats
- Real-wage erosion (spec 2nd-order outcome) is not separately tested: no harmonised real-wage series for the six cases is on disk. CPI inflation acceleration is the primary 2nd-order signal here.
- Dollarisation share of bank deposits, parallel-FX premia, capital- control stringency, and emigration flows (spec 3rd-order outcomes) are not separately tested: bcra/tcmb/cbsl/cbe vintage data not on disk. PPP-currency depreciation is used as a single 3rd-order proxy.
- VEN dolartoday parallel-rate is on disk and could supplement future v2 work; for v1 a single PPP proxy keeps the cross-case test uniform.
Strongest opposing argument
Every hypothesis ships with its charitable opposing argument. The framework earns credibility by handling objections at their strongest, not weakest.
Notes
Chain: monetary finance delivered to gov (1st) -> inflation + real wage decline (2nd) -> dollarisation + parallel FX + capital controls + emigration (3rd). Scoped to EMs with weak central-bank independence; DM QE cases (USA, JPN) are acknowledged separately and do not participate in the test.