On 14 July 2016, recently published a report on primary commodity trade misinvoicing, saying that some countries could lose as much as 67% of commodity exports to misinvoicing. The report prompted significant reaction, broadly divided between those who think misinvoicing is a problem and those who think it is not. On 23 December, UNCTAD published an updated version of the report.​

An early version of the UNCTAD report Trade Misinvoicing in Primary Commodities in Developing Countries: The cases of Chile, Côte d’Ivoire, Nigeria, South Africa and Zambia generated substantial interest and contributed to the debate on the broader issues of transparency in international trade statistics and fairness in the distribution of gains from globalization. The reactions to the report also revealed some areas of confusion in the interpretation of the results and inadequate understanding of the key concepts used in the analysis. The revised report provides a more detailed exposition of the methodology and the concepts used while further stressing the main messages from the analysis.

In the revised report, the concept of trade misinvoicing is explained in greater detail, including its origin in the literature and the estimation methodology. Trade misinvoicing consists of either perverse discrepancies or excessive normal discrepancies in partner trade statistics derived from the comparison of the value of exports as reported by the exporter to the value of imports as reported by the importer. Perverse discrepancies refer to situations where the value of imports is significantly less than the value of exports plus the cost of transport, insurance and duties. This reflects either export overinvoicing or import underinvoicing. Excessive normal discrepancies pertain to the situation where the value of imports exceeds that of exports by an amount that is substantially higher than the reasonable value of the costs of transport, insurance and duties. This situation reflects export underinvoicing or import overinvoicing. Trade misinvoicing is due to factors pertaining to the origin or the destination of trade flows or both. It is therefore not possible to assign a priori the respective share of responsibility on the basis of the estimates of trade misinvoicing alone.

Given that the data on the cost of transport, insurance and duties are not readily available for most countries the report follows the tradition of using 10 percent of exports as a proxy for these costs. In the case of South Africa, one of the only two African countries (the other country is Zimbabwe) that publish imports in f.o.b. and c.i.f. values, the average ratio of the two series over the 1980-2014 period, is 11 percent. Obviously this ratio is likely to vary across countries and products. Therefore if the estimates of trade misinvoicing are low, it may be argued that the discrepancies reflect the gap between the proxy and true value of c.i.f. But when the scale of the estimated trade misinvoicing is substantially large as is the case in the sample of countries and products considered in this report, such an explanation is not plausible.

A number of comments on the first version of the report have centered on the issue of quality of bilateral trade statistics. In addition to potential problems relating to measurement of the costs of transport, insurance and duties mentioned above, comments have also been raised about the timing of the recording of imports and exports, classification of goods, and the reporting of the destination of trade. It may be argued that the observed perverse or excessive normal discrepancies could be due to discrepancies between the data in official national statistics and the data in the databases compiled by international institutions such as COMTRADE. Such discrepancies are likely to be minimal given that these international institutions receive the data from national sources.

It has been argued that excessive normal and perverse discrepancies may arise from inconsistent classification of products across partners and over time. The case of gold exports from South Africa has been referred to as an illustration of this phenomenon. In national statistics, gold exports are split between monetary and non-monetary gold. The analysis in the UNCTAD report focused on non-monetary gold which is reported by both South Africa and its trading partners in Comtrade, thus enabling comparison of similar products as reported on both sides. Contrary to some criticisms of the first version of the report, the series of non-monetary gold exports reported in Comtrade are similar to those in national statistics, as expected, at least up to 2010. This is shown in Table 12 of the revised report using data from South Africa’s Department of Trade and Industry (DTI). However, the values reported show large discrepancies between data from South Africa and its trading partners, which may reflect inconsistencies in classification of gold. A comparison of partner data on non-monetary gold exports with the combined values of monetary and non-monetary gold provided by South Africa still exhibits large discrepancies: South African values are higher in most years up to 2010 and the situation is reversed starting from 2011. Curiously, starting from 2011, all gold exports appear under non-monetary gold in DTI statistics. This change in reporting further complicates the comparison of data from South Africa with that of its trading partners.

The interest in the issue of trade misinvoicing is, indeed, driven by the important consequences, both direct and indirect, that such a phenomenon has on national economies, especially those of developing countries. Trade misinvoicing carries direct costs in the form of foreign exchange that is not repatriated and surrendered to exporting countries’ authorities, lost government revenues from the taxes and other levies not paid on the associated exports and imports, or from export tax credit issued on inflated values of exports. An important dimension of the indirect costs of trade misinvoicing is the associated unfair distribution of the gains from trade.

---

Q&A with Janvier Nkurunziza, Chief of UNCTAD’s Commodity Research and Analysis Section

Q: Why did you feel it was necessary to publish a new version of the report?

A: When the report was published in July 2016, it attracted considerable attention, and many issues were raised and new information surfaced, including from partners and counterparts in South Africa. We felt it was necessary, as far as possible, to consider the issues raised, and therefore publish a new version.

Q: What is different in this report from the last one?

A: Before talking about the differences, I want to say that the key similarity with the first version – not difference – is that after analysis of export and import data on a range of commodities from developing countries, we still find discrepancies worth tens of billions of dollars.

Our paper covered exports of gold, silver, platinum, and iron ore from South Africa, oil from Nigeria, copper from Zambia and Chile, and cocoa from Cote d’Ivoire, over periods of 14 to 20 years.

We reached our conclusions after analyzing statistics from the UN’s COMTRADE data, one of two major sources of global trade data and the only one disaggregated by both partner and commodity. The other data base is the International Monetary Fund’s Direction of Trade Statistics, which is disaggregated by partner only.

One key difference is that in our original draft we identified discrepancies in South Africa’s gold exports of $78.2 billion, equal to 67% of their total gold exports. This result does not change if we use COMTRADE data. However, since July, South Africa has changed both its key export statistics and its methodologies. This means that, if we use the publicly available data published by South Africa’s Department of Trade and Industry (DTI), as some commentators suggest, then we can no longer calculate with certainty the discrepancies in South African gold exports.

A second key difference is that with the first version, we suggested that the core reason for the discrepancies was intent to deceive for tax evasion or other reasons. In the second version, we make it clear that it’s impossible to know with certainty why these discrepancies exist. However, we do maintain that the discrepancies are too large to be caused by simple human error or methodological differences. Finally, this version of the report looks more closely at the role of transit hubs such as Switzerland and the Netherlands.

Q: What are the key findings in this report now?

A: First, as I said, we find significant discrepancies in import-export data, worth tens of billions of dollars. Whatever the reason for these discrepancies, policy makers and civil society alike should see this as a serious problem. The discrepancies mean that billions of dollars cannot be accounted for. At best, the data is not sufficiently transparent. At worst, some of these discrepancies can represent a loss of tax receipts, foreign exchange, and opportunity.

One other finding which caused some surprise is that South Africa’s DTI made two significant revisions to its gold export data in September 2016, after we had published the first version of our report.

First, South Africa’s DTI statistics on total gold exports over the period from 2000 to 2014 changed from about $34 billion to about $62 billion. Second, exports of non-monetary gold for the period 2011 to 2014 were combined with those of monetary gold. This makes it impossible to compute the estimate of trade misinvoicing over those years, which looks only at non-monetary gold.

Q: What can policymakers learn from this report?

A: First, policymakers may wish to recognize that trade misinvoicing is a sizeable issue in commodity-dependent developing countries. These countries and their development partners may wish to establish programmes that detail the magnitude of this phenomenon, how it occurs, and who the actors are.

Second, exporting countries and their trading partners should improve the transparency and quality of trade statistics. The consistent use of information by all trading partners would allow us all to assess whether exporting countries are getting the right amounts for their commodities in terms of foreign exchange and fiscal receipts.

Third, governments and business should work together to ensure that export records match the import records of receiving countries. At present, especially for transit trade, commodities are often recorded as being exported to a country when in reality they are physically exported elsewhere. This creates confusion in bilateral data statistics, hinders efforts towards increasing transparency in international trade statistics, and complicates traceability throughout the supply chain. While we agree that companies are well within their rights to do whatever they like with the commodities that they have bought and that these commodities may be bought and sold several times while still at sea, we think that, technically, it should be straightforward to record clearly the countries which import and export commodities.