Trade misinvoicing: UNCTAD is still confused

03Aug16

A couple of weeks ago UNCTAD published a study on illicit financial flows through trade misinvoicing which they said “revealed staggering revenue losses to developing countries”. It made the extraordinary claim that “virtually all gold exported by South Africa leaves the country unreported” and suggested this was explained by  massive smuggling. The South African authorities took these claims seriously and responded robustly. [I wrote about the study here and have tracked the responses to it here].

Yesterday they published a further statement ‘UNCTAD welcomes discussion, transparency on commodities and misinvoicing’. It is disappointing, as it simply repeats the basic misunderstanding which led to the wild claims (‘mismatches in the trade data = ‘misinvoicing’), while adding new layers of confusion. It is also a bit odd, as it does not acknowledge the responses to their central claim on South African gold -by the South African Revenue Service, the South African Statistician General or the Chamber of Mines.

The statement is quite long, and so this blog post is too, but it is worth being clear on this stuff. The text in red is UNCTAD’s statement.


The paper used a standard methodology and publicly available (Comtrade) data to look at the mismatches between import and export figures on a range of commodities exported from five natural-resource-rich countries over periods of about 20 years. In doing so, it identified clear and consistent patterns of misinvoicing, a term we use without attributing blame or making any specific accusations.

There are two things to note from this opening paragraph:

‘It identified clear and consistent patterns of misinvoicing’

No. It did not. This is the basic criticism of the report and its findings. It did not examine any invoices or look for any direct evidence of misinvoicing . What it identified was clear and consistent mismatches in parts of the trade data between exports reported by one country and corresponding imports reported by partner countries. This is what it labeled as ‘misinvoicing’.

Trade misinvoicing is a method for moving money illicitly across borders which involves deliberately misreporting the value of a commercial transaction on an invoice submitted to customs, as Ndikumana explains;

trade misinvoicing consists of manipulations of exports and imports invoices by operators seeking to either secure foreign exchange advantages not reported to the relevant authorities, such as a central bank, and/or to avoid taxation or customs duties”. (Leonce Ndikumana)

As the UN Statistics Division explains (in the ‘read me first’ notes to COMTRADE) there are limitations in the trade data such as that not all countries report in every detailed commodity code. They warn that the database should be used with good knowledge of its limitations, and in particular they say;

“Imports reported by one country do not coincide with exports reported by its trading partner. Differences are due to various factors including valuation, differences in inclusions/ exclusions of particular commodities, timing etc.” (UN Statistics Division)

In other words mismatches in the data cannot be directly interpreted as misinvoicing.

The World Customs Organisation is also clear on this. In a paper published last year they give guidance to customs officials on  comparing exports of a country with corresponding imports reported by its partner countries  as a risk analysis tool to help them target investigation of revenue fraud – but they stress that mismatches in the data are not reliable on their own.

The Tax Justice Network also explained yesterday that anomalies in trade data can arise from genuine data errors and different rules of reporting, as well as from deliberate manipulation , and that there is a  risk of misidentifying data errors and different national reporting practices as  ‘proof’ of misinvoicing.

It is simply a mistake to point to mismatches in the trade data and interpret them as direct evidence of  misinvoicing.

‘Without attributing blame or making any specific accusations’

Rather than reconsider their findings in light of the overlooked evidence UNCTAD instead seem to be attempting to recharacterise what they meant by ‘misinvoicing’ in the first place.

‘Misinvoicing is a term we use without attributing blame or making any specific accusations.’ (UNCTAD- yesterday)

This makes no sense. Misinvoicing, is a form of customs fraud. It involves submitting false trade documentation to the authorities (or in the case of pure smuggling, submitting none at all but hiding the cargo completely). If they had found trade misinvoicing on the scale that they claimed, there would be plenty of blame to go around . As lead author Professor Léonce Ndikumana, told the FT;

“The implication is that it is the operators who are explicitly manipulating the invoices.” (Leonce Ndikumana)

[Then they give some context to the study : Trade misinvoicing has been generating more and more attention in the research and policy communities, it fits into UNCTAD’s wider work on commodities etc… nothing to argue with here, but it is filler]

‘This is what our data has shown us’

Challenges to our paper have fallen broadly into two specific camps. First, some argue that UNCTAD has over-interpreted – or misinterpreted – the data with its finding that some commodity dependent developing countries are losing as much as 67 percent of their exports worth billions of dollars to trade misinvoicing. …. To reply to the first argument, this is what our data has shown us. We do not see a convincing challenge to the source of our data (Comtrade), the data itself, or even the methodology used to arrive at the figures for misinvoicing, a term that we try to use without any implicit judgment or bias. Rather, the challenge has been to suggest that, for a range of historical and legacy reasons, South Africa simply does not record gold exports in the normal way, even if its trading partners do. We think this issue merits further research and discussion.

As noted above the UN Statistics division states that there are gaps and mismatches in the data and warns against overinterpreting them. Beyond this in-principle challenge to the data and methodology, there is the in-practice challenge that it generates false positive results. The statement that ‘some commodity dependent developing countries are losing as much as 67 percent of their exports worth billions of dollars to trade misinvoicing’ is based on the South Africa-gold smuggling claim.

UNCTAD says, somewhat torturously “the challenge has been to suggest that, for a range of historical and legacy reasons South Africa simply does not record gold exports in the normal way”. The South African Revenue Service, which made this challenge and is responsible for the data puts it in simpler terms: “the statistics used in the report are not correct”. SARS statement on the issue says that reported gold exports from South Africa over the period 2000 to 2014 were $124.1bn (i.e. a little over what partner countries report as imports) rather than the $34.5 billion figure generated by Ndikumana’s methodology.

UNCTAD says the South African gold statistics merit further research and discussion but say they don’t think it is reason to revisit the findings of their study;

“We do not see it as a convincing challenge to the source of our data (Comtrade), the data itself, or even the methodology used to arrive at the figures for misinvoicing” (UNCTAD)

This is just weird. If your methodology has been shown to yield a major false positive result, as attested to by respected national authorities and publicly available data then it is time to take another look at the methodology and data.

The paper gives a clue as to how they fell into this particular hole and started digging. Ndikumana argues in the paper (and in the FT) that he ruled out interpreting the discrepancies as anything other than misinvoicing because of their year-to-year consistency. He says

“It may be argued that trade misinvoicing is merely a reflection of imperfections in export and import data arising from incorrect recording, delays in reporting and/or differences in pricing mechanisms… While it is possible that recorded import and export data might be affected by statistical errors, these errors would not persist and have a trend over time. The series of the errors would be zero-mean-reverting.” (Leonce Ndikumana)

On the contrary, mismatches in the trade data caused by idiosyncrasies in the way that different countries report exports such as South Africa’s gold and Zambia’s copper would not be expected to be zero-reverting but would yield consistent patterns in the data – this is what Ndikumana and UNCTAD have misinterpreted as evidence of misinvoicing in the case of South Africa-gold (and potentially other commodities).

A second line of argument is that when Zambia says it exports its copper to Switzerland, it actually means that a Swiss-based company is exporting the copper, and therefore we should not take import and export data seriously when the copper does not arrive in Switzerland…. To reply to the second argument, we do not see any practical or moral reason why the destination country should not be recorded. Transparent documentation should make it possible to identify clearly the source country of any cargo, no matter how many times the cargo is sold or traded between source and destination countries. We agree with those traders who say that they should be free to do whatever they want with their cargoes, but we also think the traders should report to whom they sell their cargoes in order to ensure full transparency between the source and destination countries.

This is confused and does not answer the issue with the methodology. The issue over Zambian copper exports is not about incorrectly recording the source country, it is about not identifying the final destination country. The question is not whether Zambia’s export records could be improved but whether it is reasonable to interpret mismatches in the data as ‘misinvoicing’.

‘The importance of transparency’

The importance of this transparency and traceability is that, first, full transparency on the export and trade of commodities is a necessary (if not sufficient) condition for the citizens of resource rich countries to benefit from their countries’ natural resources. Second, it allows an importer to determine that its imports have no connection, for example, to human rights abuses. This is as true for gold exports from Eastern Congo as it is for blood diamonds. Third, if some companies are benefitting from tax incentives based on how much they have exported from a given country, then they should be able to identify the destination country too.

These are a broad set of general arguments for transparency . Certainly countries do need to have reliable information on the quantity and quality of commodity exports (e.g. mineral content of ores), the export price, and whether it is fair or if it has been manipulated  (as the Minerals Value Chain Monitoring project is working on in Zambia for example). However it is not obvious that traceability to the destination is the key gap in critical information in natural resource revenue management.

The study was not making a general case for transparency but claimed to show substantial trade misinvoicing in all the primary commodities which suggested large losses in terms of foreign exchange and government revenue in the form of taxes and customs duties. It makes a specific case for countries to focus their transparency efforts on international traceability and stopping massive and systemic apparent misinvoicing. If this is based on misunderstanding it risks raising impossible expectations and  distorting transparency efforts away from areas of higher priority.

‘Answering specific comments and questions’

UNCTAD welcomes discussion around these issues, which shine a light on the lack of transparency in the trade of natural resources. We answer some of the specific comments and questions below:

They then go on to quote Me, Dewald van Rensburg, of City Press and Tim Worstall  and give a response to our specific comments on South Africa-gold and Zambia-copper.

On South Africa gold:

These comments raise further questions. Why are some data on gold exports captured in the Comtrade database, but not others? Why has South Africa not recorded the destination countries since 2011, even when their trading partners do? Why are the 2011-2014 gold exports all “unallocated”? And why is this a historic practice of South Africa’s tax / customs authority that appears to affect only gold, not other exports? These idiosyncrasies only make South Africa’s gold exports less transparent. We do not understand the need for, or the benefits from, reducing transparency on South Africa’s gold exports.

These are not answers to specific questions and comments. They are a whole new set of questions! SARS say they recognize that there are classification anomalies and say ‘this matter is already under discussion with the UN Statistics department’.

On Zambia copper:

Full transparency of trade flows is necessary before a country’s citizens can be confident of benefitting from their country’s precious natural resources. Bonded warehouses are used either as part of an export processing zone (EPZ) or they are used as a transit point for commodities travelling elsewhere. If they are used as part of an EPZ, then the commodities are registered as a duty-free import before being processed, and the destination country should be known. If the warehouses are being used as a transit point, then at some point the final destination will be clear. A clear trail of documents should allow the identification of the source and final countries.

We understand that the final endpoints may not be known when the cargo sets off. But if this is the case, why register Switzerland as the destination country?

Again these are not answers to the problem with their methodology, but another tangent about improving trade statistics.

In summary, these challenges do little to reassure about the lack of transparency in the trade of commodities from developing countries.

The motivation to challenge this methodology and its findings (certainly on my part – I don’t speak for others) is not to ‘reassure about lack of transparency’, but to get to the bottom of some specific claims about massive misinvoicing which have been broadcast around the world, and to contribute to a clearer debate on the issues. Because it matters.

Characterising this argument as being a battle between the forces of transparency and the forces of untransparency is inaccurate, and is causing UNCTAD to dig itself ever deeper into the hole it has fallen into, rather than to accept directions towards the way out.

In conclusion

As we have highlighted, this misinvoicing – a word that we use in its most technical sense, without attaching value or even accusation – means that some countries may be losing as much as 67 percent of their commodity exports.

Words are losing all meaning at this point. Misinvoicing means customs fraud. There is not an alternative technical sense. The 67% figure derives from a specific claim about South African gold smuggling not to ‘some countries’.

As with related discussions on beneficial ownership and revenue flows in the extractive industries, more transparency in the commodities trade can only be a good thing for the citizens of developing countries. Indeed, this transparency may be the next major frontier to ensure that the citizens of resource-rich countries can benefit from their countries’ natural resources.

Transparency is indeed a general good thing. But citizens generally don’t consult raw data. They rely on the media who rely on academics and research organisations, official bodies and UN organisations to help them make sense of it. Misunderstandings happen. Misinformation and myths are easily spread. UNCTAD have a responsibility to put this one right.

We see this paper as a first step in a process to increase our collective understanding on transparency issues in the trade of natural resources. UNCTAD looks forward to further cooperation with partners in government, business, and the non-profit sector to better understand the lack of transparency in this sector.

I hope that UNCTAD are serious when they say they see this paper as a first step in building better understanding. Enabling collective learning is a critical part of what makes the bureaucracy of organisations worthwhile. But those processes need to be designed to overcome the human impulse to protect cherished beliefs and professional pride by avoiding new information, rationalizing cognitive dissonance away (‘we meant ‘misinvoicing’ as a technical term without judgement’…) and attributing opposing motives to those who disagree (‘we are arguing for transparency, challengers are against it’).

This is a conversation that is not working, but a conversation that works is needed. Facing up to and correcting the error over the interpretation of the South African gold statistics will be painful, but continuing to stand-by the claim does not seem tenable. Opening up on the need to reassess assumptions and ‘zombie facts’ that have become barriers to understanding would create space to convene a more constructive dialogue between those concerned with illicit flows on all sides – revenue and customs agencies, policy makers, researchers and campaigners and companies.



One Response to “Trade misinvoicing: UNCTAD is still confused”


  1. 1 The Great Gold Heist That Never Was | Hiya Maya

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


%d bloggers like this: