Comparative African Revenue Intelligence

Fifteen countries, one question

This is the seventeenth and final paper in the African Revenue Intelligence Research Series. It draws together the 15 country instantiations of ARIM (Paper 1) across Papers 2 through 16 into a single comparative maturity framework across eight descriptive dimensions.

The series instantiated a single six-domain architecture against fifteen genuinely different data environments, legal traditions and economic structures: Kenya, South Africa, Nigeria, Ghana, Rwanda, Tanzania, Uganda, Zambia, Zimbabwe, Botswana, Namibia, Ethiopia, Senegal, Côte d’Ivoire and Morocco. Each study was written to be country-specific, never a copied template.

Only when the whole set is read together do three patterns appear: a full e-invoicing maturity spectrum, consistent evidence on what transfers between countries, and a need to compare administrations without a league table.

Described, never ranked

A single composite score would have to weight e-invoicing coverage against legal capacity against cybersecurity posture. That weighting is a value judgement about which dimension matters most, and it is not one to make on behalf of fifteen sovereign administrations with different priorities, resources and legal traditions.

The framework follows the precedent of the IMF’s TADAT and ATAF’s African Tax Outlook: dimension-by-dimension description, so a Commissioner-General, an adviser or a country economist can assemble the comparison that matters for their own purpose. The synthesis draws on evidence already developed in the country studies rather than collecting new data.

The e-invoicing maturity spectrum

Ordering the fifteen countries on transaction-data maturity alone gives a near-complete spectrum: Rwanda’s EBM system, in place since 2013, then Kenya (live in 2023), Uganda, Côte d’Ivoire (freshly universal in 2026), Tanzania, the large-taxpayer-first phases of Nigeria and Senegal, Ghana, Namibia, Ethiopia, Botswana with a fixed future date, and finally Morocco with no implementing decree.

South Africa and Zambia sit outside the ordering, with no current mandate. South Africa’s multinational and Global Minimum Tax capability shows that overall sophistication does not predict e-invoicing position: the two are independent.

Patterns that cross borders

Five of the fifteen countries — Ghana, Uganda, Zambia, Namibia and Morocco — built a dedicated threshold or cliff-proximity model, each for a structurally different boundary. Namibia’s 43-point sector-rate spread, Morocco’s four-tier VAT, Nigeria’s federal-state split and Tanzania’s dual VAT jurisdiction mark where sector and network analysis carry the most weight.

A live dual currency, a dual administrative calendar and a self-enforcing invoicing design are features no generic “AI for tax fraud” framework from outside the continent would have anticipated.

Eight dimensions, no single score

The African Revenue Intelligence Maturity Framework places each administration on eight independent dimensions, each with descriptive levels rather than numbers. A country can be advanced on one and early-stage on another, and nothing is added up.

Architecture transfers, weights do not

The founding hypothesis (H5) predicted that a model calibrated in one country would lose precision elsewhere without recalibration, and that architecture, not weights, is what travels. All fifteen countries used the same six domains and the same decision chain. No two used the same weights: Botswana and Morocco set transaction behaviour to zero, while Rwanda and Côte d’Ivoire weight it most heavily.

Not one signature model was copied between countries, even where typologies looked alike. Ghana’s and Uganda’s threshold models share a pattern but were derived independently against each statutory threshold. The evidence is strong within the series, yet it rests on synthetic data, and fifteen unvalidated demonstrations do not sum to one validated finding.