Maitras.ai logo
MAITRAS.ai
Tax Data Analytics & Policy Intelligence
← Back to Insights
The Numbers That Can't All Be True
By Maitras.ai Research Desk August 2026
  • Ghana Mining Tax
  • Gold Revenue Leakage
  • MDRIA
  • Mineral Royalty Compliance
  • Transfer Pricing Gold Africa
  • Extractive Audit Targeting

Ghana's gold industry sits at the centre of the country's extractive economy — and mining and quarrying have become its largest source of direct domestic tax revenue. With so much public value tied to minerals, tax administration must do more than collect what is declared; it must also find where gold revenue leakage may be hidden in production, pricing, cost and export records.

01

Gold at the centre of the economy

EITI describes Ghana as Africa's top gold producer, with output of about 4 million ounces in 2023. That year, the extractive sector accounted for 12.8% of GDP, 67% of exports and 10% of government revenue. Mining and quarrying have also become the country's largest source of direct domestic tax revenue.

12.8%
of GDP from extractives (2023) — EITI
67%
of national exports from the sector
10%
of government revenue
02

A different approach to audit targeting

A recently published framework, the Multi-Dimensional Revenue Intelligence Algorithm (MDRIA), proposes a different approach to extractive audit targeting. Conventional risk systems often ask which taxpayer looks statistically unusual. This revenue intelligence algorithm asks something more structural:

Can a company's reported production, costs, export prices, royalties and shipments all be true at the same time?

The paper treats audit selection as a joint-feasibility test across engineering identities, market references, statutory rules and conservation of mass. It is designed to rank cases for human audit — not to determine guilt.

03

Mining leaves a physical footprint

Consider a large gold mine in Ghana. Mining leaves a physical footprint that accounting entries cannot erase — crushing, grinding, hauling and processing consume electricity, diesel and water. Under a constraint-based fraud detection approach, declared production could be compared with independently recorded energy, fuel and water use.

Energy & fuel

Electricity and diesel consumption compared against declared output.

Water use

Process water as an independent check on production volume.

Customs & shipments

Trade records that matter more where mines control their own meters.

Remote sensing

Satellite proxies for activity when on-site signals weaken.

If reported output is unusually low relative to physical inputs, the mismatch becomes a production-consistency signal. Where mines self-generate electricity or control their own meters, the paper cautions that these signals weaken, so customs data, fuel records, shipment information or remote-sensing proxies become more important.

04

Pricing, costs and royalties under scrutiny

The test would not stop at production. Declared gold prices could be checked against quality- and freight-adjusted market references, strengthening scrutiny of transfer-pricing risk. Ghana already has a specialized Transfer Pricing Unit within the Ghana Revenue Authority to address related-party pricing and base erosion. Reported operating costs could also be compared with suitable mine-level benchmarks to identify possible inflated management fees, procurement mark-ups or related-party charges.

Mineral royalty compliance provides another cross-check. Ghana's statutory mineral royalty is generally 5% of total revenue from mining operations, subject to applicable fiscal stability arrangements. The 2023 GHEITI Mining Report recorded mineral royalty receipts of about GHS 2.69 billion, up 49% from 2022.

A royalty can be mathematically correct and still sit on an understated sales base.

Royalty data therefore becomes stronger when combined with production, price, inventory and export reconciliation.

05

Catching the spread-thin evader

The framework is especially relevant to the "spread-thin" evader. A company might slightly understate output, trim its declared export price, modestly inflate costs and understate the royalty base. Each signal could remain below a traditional audit threshold — but MDRIA fuses them.

Figure 01 · Illustrative
Detectability by fraud type when evidence is fused across constraints (synthetic experiment)
Spread-thin (multi-channel)High
Export mis-pricingMed–Hi
Production under-reportMed
Cost inflationMed
Royalty base understatementLow–Med
Illustrative of MDRIA's synthetic experiment, where small multi-channel distortions became the most detectable fraud typology once evidence was fused. A promising simulation result — not field evidence from Ghana.
06

Smarter prioritization, not automated verdicts

For the Ghana Revenue Authority mining function, the opportunity is not an automated tax verdict. It is smarter prioritization — connecting tax filings with Minerals Commission production data, customs records, market prices, utility or fuel information and royalty payments to create an explainable picture of which companies' reports cannot all be reconciled.

MDRIA directs scarce audit capacity toward the cases where physical reality and financial reporting disagree most.

Stronger data. Smarter audits. Stronger revenue.

Integrate the evidence

Link tax filings, Minerals Commission data, customs, prices, utilities and royalties into one view.

Rank, don't rule

Explainable prioritization of cases for human audit — MDRIA never declares guilt.

Calibrate locally

Local calibration and strong data governance keep signals accurate to Ghana's mines.

Mobilise revenue

Strengthen domestic resource mobilization by targeting the highest-risk reports.

Collaboration

Collaborate with Maitras.ai

Revenue authorities, mining regulators and development institutions are seeking data-driven approaches to protect extractive revenue and strengthen fiscal transparency. Let's build constraint-based audit intelligence together.

Start a conversation