Tax Lab · Compliance, risk & behaviour
Transfer Pricing Risk Scoring Model
A real reference table of headline corporate tax rates and Uganda's double-taxation treaty status, plus an illustrative mirror-trade gap calculator standing in for the real UN Comtrade model.
The jurisdiction table's rates and treaty status are real and sourced. The calculator below is illustrative, driven by numbers you enter.
What this shows
The Tax Lab catalogue describes this prototype as a supervised risk-scoring pipeline: it builds bilateral mirror-trade features from UN Comtrade (Uganda's declared trade values against a partner country's own reported figures) and feeds them into a trained model that ranks CIT filers for audit prioritisation. Its own entry already says the bundled Comtrade file is "a structurally realistic but illustrative file", not real trade data, and this repo doesn't have access to a live Comtrade pull either. So this page does two honest things instead: it ships a real, sourced reference table of headline corporate tax rates and Uganda's actual double-taxation treaty partners, the kind of jurisdiction-risk input the real model would use; and it gives you a calculator to compute the same mirror-trade gap and composite score the real pipeline would, on numbers you supply, clearly marked as illustrative rather than a live finding.
The data
Real · jurisdiction reference table
| Jurisdiction | Headline CIT rate | Uganda DTA | Risk factor |
|---|---|---|---|
| Cayman Islands(No general corporate income tax.) | 0.0% | No | 100 |
| Bermuda(No general corporate income tax; introduced a 15% tax on large MNE groups from 2025 under Pillar Two.) | 0.0% | No | 100 |
| United Arab Emirates(No federal CIT until 2023; new 9% rate applies above a profit threshold.) | 9.0% | No | 85 |
| Ireland(Standard trading rate; a 15% rate applies to the largest multinational groups under Pillar Two.) | 12.5% | No | 79 |
| Hong Kong | 16.5% | No | 73 |
| Singapore | 17.0% | No | 72 |
| Switzerland(Combined federal + cantonal rate varies by canton, roughly 12-21%.) | 19.0% | No | 68 |
| United States(Federal rate only; state taxes add more.) | 21.0% | No | 65 |
| Luxembourg | 24.9% | No | 59 |
| China | 25.0% | No | 58 |
| Kenya(Uganda's largest single trading partner; no DTA in force.) | 30.0% | No | 50 |
| Mauritius(Widely used as a holding-company jurisdiction for African investment.) | 15.0% | Yes | 25 |
| Norway | 22.0% | Yes | 13 |
| Denmark | 22.0% | Yes | 13 |
| United Kingdom(Main rate rose from 19% to 25% in April 2023.) | 25.0% | Yes | 8 |
| India | 25.2% | Yes | 8 |
| Netherlands | 25.8% | Yes | 7 |
| South Africa | 27.0% | Yes | 5 |
| Zambia | 30.0% | Yes | 0 |
Illustrative, not a trained model
The catalogue's real model builds this gap from UN Comtrade's own bilateral mirror-trade statistics, across every CIT filer, and feeds it into a trained risk pipeline. Without that Comtrade pull, enter two numbers yourself: what a filer declared on one side of a trade with a given jurisdiction, and what that jurisdiction's own trade statistics report for the same flow. The composite score is a hand-set weighting (70% gap size, 30% jurisdiction risk factor), not a fitted model.
Mirror-trade gap
+25.0%
Partner reports more than Uganda declared.
Jurisdiction risk factor
8 / 100
From headline CIT rate and Uganda DTA status.
Composite score
46 (moderate)
Illustrative audit-priority tier, not a finding.
Mirror-trade gap = (partner-reported − Uganda-declared) ÷ partner-reported. Composite = 0.7 × min(100, |gap| × 2.5) + 0.3 × jurisdiction risk factor. A nonzero gap has legitimate explanations too (freight, insurance, valuation timing, re-exports); this score flags a pattern worth a closer look, not mispricing.
Where the real data comes from
Headline statutory corporate income tax rates: publicly reported national rates as of 2023 (Deloitte and PwC international corporate tax rate surveys). Uganda double-taxation agreement (DTA) status: confirmed Uganda treaty partners per PwC Uganda tax summaries and Uganda Revenue Authority public guidance, cross-checked 2026-10-07 -- limited to partners consistently confirmed across sources, not every jurisdiction sometimes mentioned.
Limitations
- The jurisdiction list is a small, hand-picked set of major trading and financial-centre partners, not a comprehensive ranked index of every jurisdiction Uganda trades with.
- The "risk factor" column is a transparent formula over two real inputs (headline rate, treaty status), not a trained model's coefficient; it is a plausible proxy, not a validated risk weight.
- The mirror-trade calculator runs on whatever numbers you type in; without a real UN Comtrade pull, it cannot screen Uganda's actual CIT filers.
- A real mirror-trade gap has legitimate, non-evasive explanations (freight, insurance, valuation basis, timing, re-exports through a third country); this score flags a pattern worth an audit look, not proof of mispricing.
Catalogue method reference: supervised risk scoring pipeline over UN Comtrade bilateral mirror-trade data and Uganda CIT filings, run as four sequential R scripts (ingestion, feature engineering, model training, scoring and reporting).