Country taxes, fees and scoring

The directory includes 250 country and territory references. Directory completeness does not mean complete tax-law coverage. Rates, fees, statistical ratios and authored scores remain distinct.

Sources and scope

PwC Worldwide Tax Summaries supplies professional secondary references for 146 jurisdictions across corporate and personal income taxes, VAT, capital gains, net wealth, inheritance/gifts and withholding. Only unambiguous numerical facts, explicit no-tax statements and individually reviewed national/range patterns are normalized. Conditional descriptions remain linked to their original country chapter; a parser does not assume the first number is the general rate.

National/federal values exclude local or alternative schedules where stated. Canadian consumption-tax ranges remain ranges. Absence of a national US VAT is not interpreted as zero state/local sales tax. A top statutory marginal rate is not an effective tax burden. Publisher review dates are not statutory effective dates; exact applicability still depends on location, residency, income type, thresholds, reliefs and treaties.

World Bank tax-revenue-to-GDP and weighted import-tariff series retain reporting years and denominators. Tax revenue is not an individual income-tax rate. Discontinued Doing Business cost benchmarks describe historical standardized cases, through 2019 in this capture; they are not current fee schedules. Business entry, property registration, permits, courts, enforcement and insolvency have different denominators and may include professional costs.

Selected official registry charges for Canada and the United Kingdom retain service/channel, native CAD/GBP, source page modification date and retrieval time. They are not an inventory of all national/local charges, are not summed across different actions, and are not converted into a comparable percentage without a defined denominator. The Canadian source page's older modification date remains visible.

Fiscal sensitivity model

Version country-fiscal-evidence-v1 uses fixed authored preference weights: personal headline reference 35%, corporate headline reference 30%, consumption headline reference 25%, and a comparable mandatory business-entry-cost reference 10%. Each eligible percentage maps to a bounded 0–100 pressure component; fee ratios above 100 are capped for this model only. The weighted inverse is a dimensionless sensitivity screen, not a sum of tax rates, a tax bill, a crisis probability or an empirically validated freedom measure.

Known regional ranges are propagated. Unknown, conditional, stale, future or conflicting components span the model's full 0–100 range. Weights are never redistributed over the known subset. The lower/upper result shows missing-input sensitivity, not statistical confidence. A scalar is withheld unless every required component is eligible and the interval collapses. Current comprehensive comparable fee inputs remain unresolved, so no complete fiscal point score is claimed at launch.

Admission requires finite values in the declared domain, exact country identity, no duplicate component, a capture within seven days and an eligible source review within 366 days. Future or invalid dates and source errors fail closed. A comparable fee input also requires the correct case, denominator and a period within two years. Native-currency charges and old case benchmarks cannot fill that slot.

Country-model hardening

The Bayes structural model is versioned as ancap-liberty-atlas-v4.3.0-fiscal. Its tax pillar reads the fiscal evidence model rather than stacking tax revenue, a derived pressure score and a policy composite. The repeated tax penalties in property and business pillars are removed. Tax revenue remains separately qualified fiscal context.

Required structural inputs are checked before scoring. Missing penalties do not become favorable zeros. The complete structural point score requires all fixed-weight pillars, replacing the earlier partial-weight renormalization. A sensitivity envelope preserves what is known and what remains unresolved; a withheld score does not receive a numerical confidence label.

News/social penalty inputs require a valid source time within 72 hours. Canonical URL/ID/title-day duplicates are suppressed across the two channels, each report counts once per relevant category, and each category is capped at three reports. Keyword matches remain contextual authored inputs, not verified legal changes or independent confirmations.

Global authored priors, the global pressure/scenario calculation and the 60/40 super-score formula are preserved. Revised fiscal data cannot unlock missing governance, security or monetary inputs. This update makes no calibrated predictive-accuracy claim.

Replay and history

New fiscal model inputs are captured with accepted Bayes replay source state. The new version separates historical comparisons. Exact previous structural functions and validator are retained for captured V4.2 replay; older scores are not relabelled as newly calculated values. Live tax panels clear during replay and account/access transitions.

Coverage gaps

Many jurisdictions lack a qualified current headline reference. Comprehensive social contributions, excises, provincial/municipal schedules, reliefs, filing and residence/exit fees require further specific evidence. A B-READY 2025 bulk download returned HTTP403 during research; that restriction was not bypassed and no data was invented from it. Unknown does not mean tax-free or fee-free.

References

• PwC tax references and country chapters
• World Bank Doing Business source metadata
• Corporations Canada fee schedule
• Companies House fees — Open Government Licence v3.0

General research evidence, not a personalized tax-liability calculation.