This week a new study came out on Trade Misinvoicing in Primary Commodities in Developing Countries by Professor Léonce Ndikumana of the University of Massachusetts at Amherst, published by UNCTAD. It looks at the value of commodities such as copper, oil, gold, cocoa etc…exported from Chile, Cote d’Ivoire, Nigeria, South Africa and Zambia and compares this with the value of the same commodities imported into countries like the Netherlands, Switzerland, the UK and US.

Professor Ndikumana believes that mismatches in these records reveals systematic misinvoicing by mining, oil and commodity trading companies, and that it points to large-scale smuggling and tax evasion by major companies. As the FT reported “commodities from the developing world worth billions of dollars are being exported illicitly every year through misinvoicing”.

“The scale of the problem exposed by the new report should prompt governments in commodity exporting nations to prioritise tackling misinvoicing over chasing new investment, donor funding or issuing debt” a senior UN official is quoted as saying.

“Some countries are losing 67% of export revenues”

The study interprets the trade data as evidence of systematic large-scale misinvoicing (i.e. fraud) across the countries and commodities studied. For example (as summarised by UNCTAD);

• Between 2000 and 2014, underinvoicing of gold exports from South Africa amounted to $78.2 billion, or 67% of total gold exports.
• Between 1996 and 2014, underinvoicing of oil exports from Nigeria to the United States was worth $69.8 billion, or 24.9% of all oil exports to the US.
• Between 1990 and 2014, Chile recorded $16.0 billion of copper exports to the Netherlands, but these exports did not show up in the Netherlands’ books.
• Between 1995 and 2014, Zambia recorded $28.9 billion of copper exports to Switzerland, more than half of all its copper exports, but these exports did not show up in Switzerland’s books.

Ndikumana says “the implication is that it is the operators who are explicitly manipulating the invoices. There has to be complicity on both sides.“

These findings align broadly with other studies such as Global Financial Integrity’s regular estimates of trade misinvoicing, and the UNECA High Level Panel report on Illicit Financial Flows, which use a similar methodology and generate figures in tens of billions in Africa, and hundreds of billions globally.

These are extraordinary claims

What is interesting about Ndikumana’s country-and-commodity level study is that it illuminates the real world actions that are implied by the widely-cited illicit flow estimates. For example he observes that “virtually all gold exported by South Africa leaves the country unreported” and suggest that best explanation is that it is smuggled out.

Imagine the network of people in major corporations, politicians, civil servants, and the media that would have to be complicit in carrying out such a massive annual heist, and paying off or hushing up any potential whistleblowers.

My brain hurts when I try to believe that such a fantastic conspiracy could remain undetected year-on-year, while at the same believing that it was discoverable all along by downloading the monthly trade statistics from the UN Comtrade database (as you may remember, we have been here before).

Are there simpler explanations for the data mismatches?

We know that smuggling and misinvoicing are real problems (politically connected trading companies in Nigeria’s oil trade, and misinvoicing used by tax evaders to hide money for example). And it is certainly true that there are large mismatches in the trade data.

But it is not obvious that these mismatches are all cases of illicit behaviour, or even that they are mysterious or suspicious. Commodities are globally traded including through futures markets. A shipment of copper, gold or cocoa loading up on the docks of a producer country will make its way to Rotterdam or Antwerp, Dubai, Singapore or other hub, where it may be warehoused before being split or combined with others, repacked and eventually delivered to ultimate destinations. The final endpoints may not be known when the cargo sets off, so the destination might be recorded as the home-base of the commodity company (e.g. Switzerland) or the first port of consignment (e.g. the Netherlands), even though it would not be expected to be reported as an import there.

Tim Worstall at Forbes explains for example that copper from Zambia or Chile being shipped to the Netherlands would not be expected to show up in Dutch import statistics, because the mining companies operate under the London Metals Exchange. This means the copper would be kept in a bonded, LME approved warehouse before being transported on. It is simply not accurate to describe this kind of normal trade practice as ‘misinvoicing’, as it does not involve companies manipulating invoices or submitting false customs forms, or anything nefarious at all.

Metal stored in a bonded warehouse. Or evidence of massive misinvoicing?

The South African gold case is a bit more unusual. As Ndikumana notes there is a very large discrepancy between the exports recorded in that country’s COMTRADE files and those reported in its trading partners’ import records. South Africa reports just $35 billion of gold exports from 2000 to 2014, whereas partners such as India, Germany, Italy and the UK say they imported over $116 billion of South African gold. Ndikumana thinks this suggests that most of South Africa’s gold exports take place through ‘pure smuggling’ (this is the basis for the ‘some countries loses 67% of their exports’ headline).

However a less mind-boggling explanation is not hard to find. Tralac, the South African trade law centre says simply that South Africa does not report gold trade data. The South African Revenue Service trade statistics website explains that South Africa does include gold exports in its domestic statistics, but due to ‘legacy rules’ it is treated ‘both as a good and as a country’ and reported under the special code ZN “Origin of Goods Unknown”. The South African Reserve Bank and SARS do not report details of where the nations gold exports go to (so therefore this data would not be retrievable from COMTRADE). SARS is aiming to move to the more general system of reporting in the future.

Why are such extraordinary claims so readily believed?

The figure that ‘some countries lose 67% of exports to misinvoicing’ has now gone around the world, adding to the body of common knowledge about illicit financial flows.

It is based on no more than an oversight – Nobody checked to see if there was a straight-forward explanation of South Africa’s strange gold export statistics, and instead they reached for the explanation which best fit the strong narrative about a massive hemorrhaging of commodity revenues through misinvoicing. This single misunderstand of South Africa gold statistics was then exaggerated into a broad finding about ‘some countries’ and ‘[overall] commodity exports’.

If your calculations tell that these companies are involved in a massive smuggling operation do you a) check your methodology and data b) look for additional evidence or c) just run with it ?

It was readily believed because of its source perhaps – the assurance of credibility of an eminent professor and a UN agency. Perhaps also because it fits into people’s existing perceptions; suggesting massive corruption by African governments and big corporations.

It also appears to corroborate the findings of other studies, such as Global Financial Integrity’s works and the Mbeki Panel report. However in practice, the reverse is true. These are not independent studies triangulating from different data. Rather they all use variations on a similar methodology and dataset; comparing Country A’s reported exports with country B’s corresponding imports, whether on a product-by-product basis, or indirectly through aggregate Direction of Trade Statistics (DOTS). All therefore are likely to treat legitimate mismatches in the data such as those related to Zambia-Copper and South Africa-gold trade as illicit flows.

Open knowledge

Enormous hopes have been placed on transparency, and in open data. But the real prize is open knowledge; better common and shared understanding across society.

Yes, trade data can and should be improved.

But so too must understanding and analysis of whatever data we have.

What Ndikumana’s study highlights is that the big numbers about trillions of dollars of illicit flows tell stories; about things that must have been done by real people and companies involving commodities, contracts, warehouses and vessels for them to be true. You can’t keep believing in a number if you can’t believe the stories it tells. And the way to test the stories is against other evidence, including the knowledge and experience of people who know these commodity markets and organisations.

That requires processes for learning, creating common language and testing assumptions across the silos, echo-chambers and opposing sides that have built up around this issue. It requires that organisations actively discard assumptions and methodologies if they prove to be unreliable, rather than building them up as unassailable beliefs to be protected against haters .

Challenging misunderstandings and wild claims one-by-one after the fact does not work. A myth can travel half way around the world before the truth has got out of bed. But if we are serious about reforming international trade, tax and finance towards sustainable development this has to be built on a body of knowledge, not a house of cards.