Tax administration & revenue mobilisation — Public Finance Module 6

Tax administration is where good tax design either becomes revenue or remains words on paper. African economies routinely have statutory tax rates in line with OECD peers but collect 30-50% less as a share of GDP. The gap is administration — not policy design. This module is about closing that gap.

The four functions of a tax administration

Registration — getting every taxable person and entity onto a tax register with a unique identifier (KRA PIN, GST number, equivalent). The Kenyan PIN is the foundation of the system; without it you cannot import goods, register a business, open a bank account at most banks, transfer property
Filing — collecting the periodic returns (PAYE monthly, VAT monthly, CIT annually, etc.) that report taxable activity and self-assess the tax due
Audit — checking returns against third-party data (eTIMS records, bank statements, customs declarations) and identifying mis-statement. Risk-based audit (audit the highest-risk filers, not everyone) is the modern standard
Collection — actually receiving the money. This includes voluntary payment, withholding, garnishment, asset seizure, and the dispute-resolution / tribunal system that handles contested assessments

Why withholding dominates in low-capacity contexts

Withholding is the system whereby a third party (employer, bank, contracting agency, importer) deducts the tax at the point of paying the taxable person and remits it directly to the tax authority. The taxable person never holds the gross amount.

The withholding advantage

Self-assessment requires every taxpayer to honestly calculate and remit their own liability. In a context with high informality, low literacy in tax matters, and weak audit capacity, self-assessment collapses — non-compliance becomes the norm. Withholding shifts the collection burden to a smaller number of capable agents (employers, banks, KRA itself at customs) who have the systems to comply and a regulatory relationship that punishes non-compliance. Most African economies collect 50-70% of total tax revenue through withholding mechanisms, vs 15-25% in OECD economies. The difference is administrative capacity, not policy choice.

Kenya withholding examples:

PAYE — employers withhold income tax from monthly salary and remit by the 9th of the following month
Withholding tax on dividends — paying companies deduct at source (5% resident, 15% non-resident)
Withholding tax on interest — banks deduct on interest credited (15% resident, 25% non-resident)
Withholding tax on professional fees — entities making professional-fee payments above KES 24,000/month deduct 5% (resident) or 20% (non-resident)
Withholding VAT — designated 'withholding agents' (mostly public entities and large corporates) deduct 2% of VAT-inclusive payments to suppliers and remit directly. Controversially expanded in 2023 to many private-sector buyers; partial rollback in 2024
Digital service tax — non-resident digital providers self-register, but for many practical cases, payment intermediaries withhold and remit

The eTIMS revolution

Electronic Tax Invoice Management System (eTIMS), rolled out by KRA progressively from 2023, requires every VAT-registered business to issue tax invoices through a KRA-certified electronic system that transmits the invoice data to KRA in real-time. The economic effect is to close the VAT input-output gap that was the largest source of revenue leakage.

Pre-eTIMS: a business could claim input VAT on invoices that the supplier never reported as output VAT. The gap was ~25% of theoretical VAT yield in IMF estimates
Post-eTIMS (target state): every input-VAT claim is matched against a real-time output-VAT report. Fictitious invoices fail validation
Compliance trajectory: by mid-2025, ~340,000 businesses on eTIMS, ~85% of large taxpayer office filings via eTIMS, with ongoing rollout to medium and small filers

Presumptive and turnover taxes for the informal sector

The informal sector — businesses too small, mobile, or unstructured for normal income-tax compliance — represents 80-90% of African employment and a non-trivial share of GDP. Standard income-tax design fails here: no books of account, no clear separation of personal and business cash flow, no fixed premises to audit. The administrative response is a simplified tax regime.

Turnover tax (Kenya) — 3% of gross revenue for businesses with annual turnover between KES 1 million and KES 25 million. No deductions, no separate VAT, no expense-tracking. A flat rate on declared revenue
Presumptive tax (Kenya 2019, replaced) — a flat annual fee tied to the business permit, paid annually. Failed because business-permit administration is weak and the linkage was contentious
Single-business permit + tax (some counties) — combining business-permit registration with a flat annual income-tax amount, collected by the county. Improves compliance because the business cannot operate without the permit

The compliance-trap trade-off

Simplified regimes are designed to be EASIER than full self-assessment, to bring informal businesses into the tax net. But they create a trap: a successful informal business that grows past the turnover-tax threshold faces a sudden jump in compliance cost (full VAT, full income tax, full books). The 'graduation cliff' creates an incentive to stay just below the threshold — to under-report revenue or split the business into multiple entities. Smooth-graduation regimes (gradual rate increases above the threshold, transitional simplified-bookkeeping rules) reduce the trap.

Digital payments as a tax-administration multiplier

M-Pesa, Airtel Money, and bank-driven mobile money are now the dominant transaction infrastructure in most East African economies. Every transaction leaves an electronic trace, which (in principle) makes audit trivially possible. The challenges:

Data access — KRA can compel data from telcos and banks under court order or specific legislative authority. The Kenya Finance Act 2023 expanded data-access powers; the Data Protection Act 2019 sets the privacy framework. Tension between the two is ongoing
Algorithmic risk-scoring — the data is too voluminous for human audit. Risk-based selection (audit the 5% most-anomalous filers, where anomaly is defined by ML-based scoring against peer benchmarks) is the operational model
Personal-business separation — informal businesses transact via personal M-Pesa accounts. The data tells you what flowed through, not which transactions are 'business' for tax purposes
Mobile-money excise — Kenya levies excise on mobile-money transactions (15% as of 2025). This is a tax on the infrastructure itself, creating tension with the 'tax administration benefits from digital traceability' argument. Higher excise discourages digital-payment use and pushes activity back into cash, reducing the audit data and the formalisation premium

International benchmarks

Tax revenue as a share of GDP, selected economies (2023, World Bank/IMF):

text

Country              Tax/GDP
Denmark              45.9%
France               45.4%
United Kingdom       33.5%
South Africa         26.7%
Mauritius            21.3%
Kenya                15.5%
Uganda               13.6%
Nigeria              10.8%
DR Congo              7.9%

The African average (~17%) sits below the OECD average (~33%).
The Kenya target under MTRS (Medium-Term Revenue Strategy) 2024-2027
is to raise tax/GDP from ~15% to ~20% over the period — entirely
through administration improvements, not rate increases.

Exercise

Kenya's tax-to-GDP ratio sat at 15.5% in 2023. Compare this with Mauritius (21.3%) — both small open economies, both Anglophone. (1) Identify three plausible structural reasons for the gap. (2) Identify three administrative reasons. (3) Propose three reforms aligned with the Medium-Term Revenue Strategy that target the administrative gap without raising statutory rates. (4) What's the political-economy obstacle to each reform you propose?