Methodology

The index answers one question per country and power: does this tax authority hold this enforcement power — and what is the evidence? There is deliberately no composite “strictness score”. Scores hide judgment calls; per-power evidence exposes them.

The evidence gate

Every affirmative claim (Law / Practice / Partial) must carry a verbatim quote from its source and the source URL. A verification job re-fetches each source and checks the quote still appears, character for character (after whitespace/entity normalisation). Claims whose quotes cannot be machine-verified do not render — they fall back to “not yet assessed”. A blocked or paywalled source is treated as unverified: a page we cannot read is not evidence.

Statuses

• Law — the power exists in statute or binding regulation.
• Practice — documented or self-reported use, whether or not a dedicated statute exists.
• Partial — limited scope, or committed but not yet in force (e.g. CARF signatories pre-2027).
• No — verified evidence the power is absent (e.g. residence-based taxation, no exit tax).
• No evidence found — we searched and found nothing citable. Recorded as a finding, not silence.
• Not yet assessed — no claim either way. Most non-flagship countries start here.

Source tiers

• Official — legislation, government guidance, regulator publications.
• OECD / IMF survey — ISORA and ITTI answers, self-reported by each administration; cells derived from these are marked as such.
• Professional — Big-4 / law-firm analyses and legislative research bodies (e.g. PwC Worldwide Tax Summaries, PRS India).
• Press — reputable press reports; used sparingly and labeled.

Quantitative data

• Tax-to-GDP: OECD Revenue Statistics (Global) — general government total tax revenue, ~130 jurisdictions. World Bank/IMF GFS central-government figures are shown only for countries the OECD set lacks, and the two definitions are never mixed in one ranking.
• ISORA (IMF/OECD/CIAT/IOTA International Survey on Revenue Administration) — audit yields, automated compliance checks, third-party data systems, e-filing mandates, audit staffing; ~170 jurisdictions.
• ITTI (OECD Inventory of Tax Technology Initiatives) — AI and big-data usage questions answered by 100+ administrations in the 2024 Global Survey on Digitalisation.

Vintage

Cross-country rankings compare only the cohort of jurisdictions whose most recent observation falls in the same year — the year most of them reported, not a claim that it is the current one. Every other jurisdiction is listed separately, unranked, against its own year; nothing is dropped for reporting a different year than the cohort.

Regime tables (controlled foreign company rules, interest limitation, country-by-country reporting) instead take each jurisdiction’s country-level answer where one exists, and otherwise its most recently reported numbered regime, shown against its own year.

Stated rates

A statutory rate that PwC states as a single number is ranked. A statement given in words is reproduced verbatim and not ranked. Where a statement scopes its leading rate to one group of taxpayers and gives a different rate to another, it is likewise reproduced as written rather than ranked, because there is no single headline rate to rank. A statement recorded as “NA” is reported as that notation — PwC’s own key defines NA as the territory not having the indicated tax or requirement — and is never converted into a 0% rate: the source gave a notation, not a measurement, and some NA statements are explicitly qualified (for example by naming a province that does levy the tax, or by noting that general tax rules apply).

Honest limits

Survey answers are self-reported and may flatter. Legal texts quoted in their original language are summarised in English — the quote governs, not our summary. Laws change: every claim shows when its quote was last verified, and stale claims are re-verified on a schedule. If you spot an error, the fastest fix is to send the primary source.