Clearance Mandates Compared

Turnover / size descent

The dominant modern design: start with the largest taxpayers and walk the threshold down. India went from ₹500 crore to ₹5 crore in six waves over 34 months;^[25] Saudi Arabia ran 24 integration waves from SAR 3bn down to the SAR 375k registration floor over 3.5 years;^[31] South Korea descended thresholds for 13 years; Malaysia phased RM100m → RM1m in four waves.^[44]

Every announced go-live date in this design family that we could document was met (after India's single pre-launch COVID deferral) — slippage, where it occurred, was absorbed by trimming scope at the bottom rather than moving dates.

Invoice-value descent

Israel's unique variant stages by the size of the invoice, not the business: allocation numbers were first required only on invoices above NIS 25,000 (2024), stepping down to NIS 5,000 by June 2026 — keeping the smallest firms out entirely until the final step.^[34] Hungary's RTIR began the same way, reporting only invoices above HUF 100,000 in VAT before going universal in 2020.^[37]

Counterparty staging (B2G → B2B)

Serbia's 20-month ladder — government-to-government, then business-to-government, then full B2B on 1 Jan 2023 — is the cleanest example; Romania, Greece, Portugal and most EU rollouts also led with the public sector, where government buyers guarantee a receiving counterparty from day one.^[41]

Regional pilot → national cutover

Vietnam rehearsed in 6 provinces from November 2021, extended to the remaining 57, then made paper invoices invalid nationwide on 1 July 2022 — the closest thing to a successful modern big-bang, with 92% of obligated taxpayers registered during phase 2.^[28] China ran the same play more cautiously: a 3-year provincial pilot, nationwide availability in Dec 2024, and still no hard mandate date for incumbent taxpayers.^[47]

Segment big-bang

Spain's SII put ~62,000 large taxpayers — about 80% of national transaction value — onto 4-day real-time ledger reporting on a single date (1 July 2017), and stopped there: the once-floated extension to all 3 million businesses never happened.^[39]

Report first, clear later

The emerging European two-step: Romania required all B2B invoice data to be reported from Jan 2024, then made the platform invoice the only legally valid one from Jul 2024; Greece ran myDATA reporting for five years before its 2026 clearance-style e-invoicing mandate.^[38] This de-risks the platform before invoice validity depends on it.

Peru — the causal benchmark

The IMF's firm-level evaluation of Peru's staged rollout is the cleanest causal evidence anywhere: e-invoicing increased reported firm sales, purchases and value-added by over 5% in the first year after adoption, with effects concentrated among smaller firms and low-compliance sectors.^[22] A follow-up study found powerful network effects: smaller firms whose trading partners were mandated into e-invoicing reported 11% more sales and paid 17% more VAT — compliance propagates through supply chains before coverage is even complete.^[53]

The same study is candid about limits: cash revenue gains were muted by weaknesses in Peru's VAT refund system. Digitalisation needs complementary reforms to convert reported liabilities into collected cash.

Italy — the big-bang dividend

Italy's VAT compliance gap fell from 23.5% of theoretical liability in 2017 to 10.8% in 2021 — a €33bn gap cut to €14.6bn — with the 2021 drop alone (−10.7 points) the largest in the EU-27.^[7] The finance ministry's own early estimate attributed roughly €2bn of extra VAT to e-invoicing in 2019, revised toward €4bn as the year progressed, plus ~€700m of fraudulent VAT credit offsets blocked at the gate.^[8]

Caveat: the EC's re-benchmarked December 2025 report revises Italy's 2023 gap to 15.0% — a real partial rebound plus a methodology change. The improvement is large either way; its exact size depends on the edition cited.^[4]

Hungary — reporting alone worked

Hungary never mandated clearance — just real-time invoice reporting — and cut its VAT gap from 14.3% (2017) to 4.4% (2021), among the EU's best, a decline the EC country chapter calls "steady and steep."^[6] Domestic analysis attributes over HUF 400bn of additional budget revenue to the combination of online cash registers, the EKÁER transport system and RTIR.^[54]

Poland — gains without (then before) clearance

Poland's VAT gap fell from 23.9% (2015) to 9.9% (2018) on the strength of SAF-T files, split payment and real-time bank-flow analysis — before KSeF existed.^[14] By 2023 the gap had rebounded to 16.0% while KSeF sat delayed — the dataset's clearest suggestion that enforcement gains can erode without continuous transaction-level controls, though other factors (inflation, methodology revision) share the blame.^[4]

Latin America — the founding evidence

The IDB/CIAT programme evaluations found positive revenue effects in all five countries studied (Argentina, Brazil, Ecuador, Mexico, Uruguay).^[50] Peer-reviewed point estimates: Ecuador's declared VAT rose +19.4% (2015) and +28.1% (2016) among affected taxpayers (~US$133m); Uruguay saw +3.7% on corporate VAT paid; Mexico's SAT-commissioned study measured VAT evasion falling from 31.9% to 19.4% and income-tax evasion from 39.8% to 25.6% between 2011 and 2015 — confounded, honestly, with the 2014 fiscal reform it accompanied.^[55],[23]

The intellectual foundation is Chilean: Pomeranz's randomised evaluation across 400,000+ firms showed the VAT invoice paper trail itself deters evasion and propagates enforcement up supply chains — the result CTC regimes industrialise.^[56]

EU-wide — the aggregate picture

The EU VAT gap fell from €99bn (2020) to €61bn (2021) — "an unprecedented improvement" — and the Commission explicitly credited "new digital reporting tools, the real-time tracking of transactions, and e-invoicing regimes which are particularly effective against criminal VAT fraud."^[5] The ViDA impact assessment quantified the cross-country effect: member states that introduced digital reporting saw VAT revenue increases of 2.6–3.5% per year, worth €19–28bn over 2014–2019.^[57]

The 2025 edition shows the gap widening again to €128bn / 9.5% by 2023 under revised methodology — digital controls are necessary but clearly not sufficient in a high-inflation, post-pandemic environment.^[4]

Projections on the record

France's tax administration expects €2–3bn/year of additional VAT in steady state from its 2026–27 mandate, against an estimated €6–10bn national VAT gap;^[58] its official impact study also projects ≥€4.5bn/year of business savings (≈€10 per paper invoice vs under €1 electronic).^[59] The EU's ViDA package projects up to €11bn/year less VAT fraud and €4.1bn/year lower compliance costs.^[57]

Where no number exists

Honesty requires naming the gaps: no audited revenue-attribution figure was found for Saudi Arabia, Egypt, Jordan, Israel, Turkey, Serbia, India, Vietnam, Malaysia or China. Egypt's +22.5% VAT growth and India's record GST collections are real but officially attributed to broad reform packages, not e-invoicing specifically; Korea's World Bank evaluation explicitly declined to estimate a revenue effect, documenting instead an ~80–85% collapse in bogus-invoice fraud cases from their 2005 peak.^[27],[33]

1 · Central platforms that weren't ready

Poland's KSeF was halted five months before go-live when an external audit found "critical errors... affecting overall system functionality and performance"; the state rebuilt the platform from scratch.^[16] Indonesia's Coretax — a big-bang replacement of the entire core tax system — broke invoice issuance on day one (of 845,514 invoices attempted in the first nine days, only 236,221 validated), forcing a formal apology from the tax directorate and a penalty moratorium.^[60] France, watching, abandoned its public platform's exchange role entirely in October 2024 and pushed exchange onto ~158 accredited private platforms.^[12]

2 · The SME wall

Every extension documented in this research was, at root, about small businesses: the Dominican Republic's two 6-month wave extensions (digital-certificate procurement), Malaysia's decision to exempt sub-RM1m businesses outright, Peru's +13-month wave 5, Chile's deliberate 4-year micro-enterprise tail, Ecuador's 88%-uncovered final sprint.^{[30],[45],[22],[52]} Digital certificates, software costs and connectivity — not willingness — are the binding constraints cited.

3 · Mid-stream redesigns

Colombia switched from post-validation to prior-validation clearance mid-rollout (2020); Mexico's CFDI 4.0 migration took ~9 months longer than planned; Costa Rica's v4.4 migration produced case-by-case extensions for hundreds of firms; Brazil's 2026 CBS/IBS reform is re-opening layouts across every live system; and ViDA's 2035 harmonisation deadline guarantees that even finished European systems (SDI, KSeF) face a second compliance wave.^{[51],[23],[61],[68],[3]}

4 · Operational integration burden

Saudi Arabia's Phase 2 documents the day-two problems of a mature clearance regime: cryptographic hash-chains that break on network drops, clearance bottlenecks at month-end peaks, opaque error diagnostics, and per-transaction fees that scale faster than business volume.^[62] Poland's first weeks surfaced certificate-generation queues and laggard software vendors pushing firms into offline fallback modes.^[63]

5 · Privacy — the European constraint

Italy's data protection authority intervened before the 2019 launch, warning of "disproportionate collection of information"; the outcome barred the Revenue Agency from building a full invoice-content database and excluded healthcare invoices from SDI — a constraint with no equivalent in the Latin American systems Europe borrowed from.^[64]

6 · Credibility decay

Where deadlines slip repeatedly — Pakistan's three schedule rewrites in 18 months, Greece's 2020–2026 sequence, the Philippines' 2018 law still pre-mandate — taxpayer investment stalls, because the rational response to a moving deadline is to wait.^{[48],[43],[49]} The contrast is Saudi Arabia's 24 waves announced ~6 months ahead and never moved: predictability itself became the compliance tool.^[32]