Can Physics Catch a Tax Cheat?

Why gold is the economy

A 2024 EITI-commissioned study estimated mining-related illicit financial flows at USD 4.93 billion between 2012 and 2021, with gold accounting for 61%. Strengthening Burkina Faso's gold tax collection is therefore a question of domestic resource mobilization — not simply compliance.

14.8%

of GDP from mining (2023) — EITI

$4.93B

mining-related illicit flows, 2012–2021 — EITI

61%

of those illicit flows attributable to gold

A new logic for fraud detection

This is where the Multi-Dimensional Revenue Intelligence Algorithm (MDRIA) changes the logic of mining tax fraud detection. Instead of asking whether one tax return looks unusual, it asks whether all the reported numbers could be true at the same time.

Can all the reported numbers be true at the same time?

MDRIA tests production, export prices, costs, royalties, revenue and shipments against physical, market, statutory and mass-balance constraints. It is an audit selection algorithm — a tool for prioritizing human investigation, not declaring guilt.

Testing a mine from every angle

Picture an industrial gold mine reporting its ore processed and gold produced. MDRIA could compare those declarations with independent evidence — cross-checking six dimensions at once:

Energy & fuel

Electricity, diesel and water consumption benchmarked against declared output.

Export prices

Checked against quality- and freight-adjusted market references to expose transfer-pricing risk.

Cost benchmarks

Reported costs tested for inflated management fees and other margin erosion.

Royalties

Recalculated from the applicable rate and the correct sales base.

Shipments

Production, inventories and exports reconciled through partner-country trade data.

Mass balance

Physical consistency across the full production-to-export chain.

An approach made more urgent by EITI's own reports of discrepancies between government and company gold figures.

Catching fraud spread thin

MDRIA is most powerful when manipulation is spread thin — a little under-reported production, a shaved export price, slightly inflated costs, and a mathematically correct royalty on an understated base. Each distortion alone may sit below a conventional threshold, but MDRIA combines the weak signals.

Figure 02 · Illustrative

Detectability by fraud type under joint-consistency testing (synthetic experiment)

Multi-channel (thin)High

Export mis-pricingMed–Hi

Production under-reportMed

Cost inflationMed

Royalty base errorLow–Med

Illustrative schematic of the paper's synthetic experiment, where the multi-channel strategy was the most detectable fraud type — a promising simulation result, not yet proof of on-the-ground performance in Burkina Faso.

Physical reality as a line of defence

The opportunity is not to replace auditors with physics, but to connect tax, customs, mining, utility, market-price and beneficial-ownership data so auditors can see contradictions that separate databases hide.

Physical reality cannot be altered — but reports can. MDRIA helps revenue authorities see the difference.

Stronger data. Smarter audits. Stronger nation.

Protect extractive revenue

Turn physical reality into an additional line of defence for the nation's gold revenue.

Target scarce audit capacity

Prioritise the highest-risk, highest-impact cases so limited audit resources go furthest.

Integrate fragmented data

Connect tax, customs, mining, utility, market-price and ownership records into one view.

Govern with strong safeguards

Local validation and strong data governance keep MDRIA an aid to auditors, not a verdict.