Introduction
A Government Valuator does not value diamonds. He is like an auditor, he approves whether the assortment of the diamonds is correct, according to the standard assortment. That is all he does. The value really is the price which is in the Price Book [which is drawn up by De Beers]. So the government valuator has got no input into the value of a diamond. (Bertie Lincoln, former De Beers director, speaking under oath to a South African court in 1997)
Mining has been at the centre of the South African economy, and its partial industrialisation, since the second half of the nineteenth century. Unsurprisingly, South Africa exhibits a number of the challenges associated with resource abundance and resource curse characteristics (Auty 1993; Ross 2008; Sachs and Warner 1995): it is one of the most unequal countries in the world, both when measured against income and against broader measures of inequality (ranked 2nd most unequal by Gini coefficient, World Bank 2015, 2.9); its manufacturing sector continues to struggle to be competitive as resources are devoted to the ‘minerals–energy complex’, while it displays characteristics of a ‘rentier state’ in its dependence on foreign multinationals and its privileging of their interests (Mahdavy 1970). How far the development of South Africa has benefited from the minerals economy has been recently questioned, given persistently low wages and contributions to the tax base (Bracking and Sharife 2014) and is not the subject of this article. Nor does this article attempt to review again the vast literature generated by the ‘resource curse’ metaphor (see Bebbington et al. 2008; Williams 2011). Our contribution to this debate is an empirical one: we are arguing that there exists a disparity between import and export values that suggests transfer pricing manipulation of the value of rough stones in South Africa. This is a substantive reason why South Africa, and other similar natural resource exporters, may be performing poorly economically, as described and predicted in the resource curse literature. It can be taken as evidence of a process which contributes to the aggregate correlations between mineral resources and low growth.
The article in particular focuses only on one mineral, diamonds, and does not seek to frame the same conditions governing the valuation of diamonds with that of other minerals, many of which have an international benchmark price and can therefore be considered ‘pure commodities’. In contrast, diamonds have only an internally imputed pricing system that has no international or ‘free market’ measure against which the commodity can be compared by value and volume. For example, value of a kilogram of gold, prior to production costs, can be ascertained from general benchmarked pricing (currently averaging $36,000 per kg), on the basis of a consistency in value from volume. But the mere presence of inclusions and other criteria can drastically reduce the value of gem-quality diamonds on the basis of volume. Nor is value limited by volume itself. These qualities, in fact, form the basis of diamonds as both an exceptional mineral and an entirely opaque one, constructing for the mineral the dubious honour of being called the ‘money launderers’ passport’. Indeed, a confidential contract, dated 2015, and seen by the author, shows high-valued stones sold by De Beers, to a De Beers sightholder, with a value of $20 million at less than 300 Ct. The transaction was negotiated in secrecy through a tax haven, with a sightholder entity incorporated in a tax haven, and thus was exempt from traditional external scrutiny found in non-tax havens.
The article finds that there is a low developmental and fiscal contribution made by the diamond industry in South Africa using data from 2004–2012. Our data suggest that either or both of import overvaluation and export undervaluation are being practised in the South African diamond market, although we are not able to establish (because of lack of public data on domestic beneficiation) which, or in what proportion, each (overvaluation of imports/undervaluation of exports) contributes to discrepancies that are in evidence between the values of diamonds as they leave relative to when they are imported or mined. But we can find that there exist significant discrepancies indicative of possible transfer pricing manipulation of rough diamond values, with ‘lost’ values of US$3.34 billion recorded for domestic production from 2004 to 2012 after controlling for imports at the point of export. Because of these discrepancies it can be plausibly suggested that the industry is not contributing the level of tax that could be reasonable expected by South African citizens, estimated at a deficit of tax paid of about ZAR1 billion per annum (Bracking and Sharife 2014).
However, this article has a purpose beyond empirical evidence: it explores, using more than seven months of research and analysis through interviewing key players in the diamond industry, how this valuation system works, who is involved and how the outcomes of this system at various locations in the import and export chain, expressed in price, determine the fiscal contribution eventually paid. We use the research protocol designed to explore valuation processes of the Leverhulme Centre for the Study of Value (LCSV), in order to theorise how these empirics are generated from within the underlying valuation system. This protocol isolates institutional assemblages, calculative entities and discursive framings as together contributing to the valuation system in place (Bracking et al. 2014; Fredriksen 2014; Fredriksen et al. 2014). Callon (2007, 320) writes that ‘agencements [mostly translated into English as “assemblages”] are arrangements endowed with the capacity of acting in different ways depending on their configuration.’ We describe here how an institutional assemblage is contributed to and conditioned by a single company, wherein a calculative entity, in ‘The [diamond] price book’, records the value of a particular quality of diamond carat; while a discursive framing assists in making diamonds, which have little intrinsic value, into luxury goods. This analytical effort is inspired by the disciplines of economic sociology (MacKenzie 2009, 2011) and work on performativity (cf. Callon 2005, 2007; but see Miller 2002) and more specific work on valuation in marketisation processes (Callon 2007).
The article finds that diamond valuation in South Africa largely escapes public regulation, and works heavily in favour of capital-holders who, by means of their control over marketisation, can then condition prices and taxes more broadly. By controlling valuation, the industry, and predominant within the sector the De Beers corporation, can then choose to make profit in low-tax jurisdictions, rather than in the country of primary economic activity. Value is shifted by the opportunity for setting prices at various points, leading to ‘mispricing’. The article ends by concluding that it is this structural configuration of the market, which confers disproportionate power to mining corporations relative to the citizens of South Africa, which allows trade mispricing to occur. The market is conditioned by historical patterns of race and class accumulation, oligopoly and conditions of enclavity (Ferguson 2005; Hirschman 1973; Singer 1950). However, the agency that allows the permissibility of trade mispricing, though structurally embedded, is also reproduced by the political elite who have aligned their own policy and practice synergistically to the interests of mining companies, and who are seemingly content to allow this market configuration to continue.
Diamond mining in South Africa
South African diamond mining continues to be dominated by two companies, De Beers Consolidated Mines Limited1 and Petra Diamonds, which together account for more than 95% of production (Petra Diamonds 2013). In recent years De Beers sold a number of its more important South African mines to Petra Diamonds, including the Cullinan and Finsch mines (De Beers 2012). Though De Beers is correctly associated with South Africa, it also owns 50% of Debswana Diamond Company and 50% of Namdeb Diamond Corporation, with the other 50% owned by the governments of the Republics of Botswana and Namibia respectively (AngloAmerican 2015). De Beers once had a global cartel in diamond trading run by its Central Selling Organisation and associated Diamond Trading Companies (DTCs), the rough diamond trading and distribution arm of De Beers. More recently, non-DTC production has increased owing to declining reserves in South Africa; Russia no longer using the De Beers marketing platform; the discovery of new diamond fields including BHP Billiton opening mines in northern Canada; former cartel members opting to sell outside the cartel; and the market entry of new players resistant to the De Beers monopoly. The latter category of new entrants includes both formal commercial companies and informal networks including ‘blood diamond’ sellers who share an independence from the De Beers/Kimberley Process (KP) sphere of influence.2 Nonetheless, De Beers maintains a sizable share of world rough diamond trading.
Until 2008, De Beers controlled a minimum of 90% of domestic South African diamond production. The subsequent sale of mines diluted this dominance. In 2011, De Beers, which had begun selling mines from 2010, generated just 60% of domestic production by value (not volume), down from 92–95% from 2004–2009 as listed in the South African Department of Mineral Resources (DMR) report (DMR 2012). Another 37% from 2011 production was generated by Petra, such that the companies together generate a full 97% of domestic production by value. In 2011, De Beers still produced 77% of total domestic production by volume. However, Petra confirmed that the company neither imports nor exports, and that while the company has acquired De Beers mines (thus the expansion of the former, and contraction of the latter in local market share), ‘Petra does not sell any diamonds to De Beers. We mine, sort and sell the entirety of our rough diamond production [locally]’ (email to Sharife, 2014). Thus, De Beers has an overwhelming market share of the cross-border diamond trade. In other words, if Petra does not export, the 60% of domestic production by value contributed by De Beers must constitute at least 97% of South African production exported by value (adjusting for the 3%).
Local beneficiation
In terms of the polished diamond market, South Africa continues to struggle to develop a significant diamond cutting and polishing industry, but it does have companies initiated from the black economic empowerment (BEE) policy, which provides for a move to increasing local black equity ownership. There is a levy to encourage local beneficiation by taxing exports at 5%, which a company can avoid by providing 40% of local production to the local polishing industry. Companies in South Africa, but principally De Beers, also conduct cross-border sales of gems, particularly with Namibia and Botswana, and to a much lesser degree with Zimbabwe. Demand for polished diamonds is determined, as with any other economic good, by changes in income and changes in preferences, and it should not come as a surprise that the global demand for finished diamonds has gone down in the muted economic period of the financial crisis and afterwards.
De Beers DTCs in South Africa comprise the primary buyers for imported diamonds from De Beers, who are the near-monopoly importers. Parcels sold to DTC sightholders (companies authorised as bulk purchasers of rough diamonds) comprise mixed diamonds, and remain non-negotiable. The parcels can be rejected but sightholders thereafter will allegedly not be invited again. DTC sales are confidential, claimed De Beers, though some values for the years 2007, 2008 and 2009 were disclosed in an annual report, disclosing significant discrepancies between diamonds apparently imported for local buyers, and the sales to those local buyers that actually take place (De Beers 2011). De Beers also exported diamonds to its London valuation hub, which moved to Botswana in 2013. The precise data cannot be determined as De Beers claimed that value, volume, percentage of imports and exports, as well as local and export sales, comprised proprietary information.
Table 1 shows the domestic production, imports and exports of diamonds in South Africa from 2004 to 2012 and their values as recorded by the companies concerned, principally De Beers on Kimberley Process certificates. The parcel values for import and exports that feed into the KP data process are derived from the US dollar estimate of value that forms part of the actual ‘certificate’ assigned to an import or export. It is, in other words, the value assigned by the importer or exporter, which is then verified, in principle, by the Importing/Exporting Authority. In South Africa this is the South African Diamonds and Precious Metals Regulator (SADPMR), with the Government Diamond Valuator (GDV) being specifically tasked to ensure that both import and exports diamonds are traded at fair market value (SADPMR 2012, 38).
| Domestic production | Imports | Exports | |||||||
|---|---|---|---|---|---|---|---|---|---|
| KP Volume cts (mill) | KP Value US\((mill) | KP US\) per ct [A] | Volume cts (mill) [G] | Value US\((mill) [C] | US\) per ct | Volume cts (mill) [D] | Value US\((mill) [E] | US\)/cts | |
| 2004 | 14.09 | 1075.76 | 76.34 | 0.93 | 608.64 | 655.59 | 14.82 | 1835.69 | 123.84 |
| 2005 | 15.56 | 1319.09 | 84.78 | 1.10 | 728.56 | 664.75 | 20.39 | 2148.29 | 105.37 |
| 2006 | 14.93 | 1361.82 | 91.18 | 0.74 | 672.18 | 905.99 | 15.78 | 1930.28 | 122.32 |
| 2007 | 15.21 | 1417.33 | 93.18 | 1.24 | 2113.89 | 1705.67 | 13.89 | 1867.33 | 134.44 |
| 2008 | 12.90 | 1236.24 | 95.82 | 0.68 | 582.25 | 850.09 | 10.14 | 1484.83 | 146.39 |
| 2009 | 6.14 | 885.54 | 144.23 | 0.66 | 357.20 | 544.73 | 9.55 | 1018.67 | 106.67 |
| 2010 | 8.86 | 1194.28 | 134.75 | 0.40 | 307.96 | 773.16 | 3.76 | 709.22 | 188.76 |
| 2011 | 7.04 | 1388.68 | 197.13 | 1.35 | 460.17 | 339.79 | 6.65 | 1370.45 | 205.94 |
| 2012 | 7.08 | 1027.13 | 145.13 | 11.47 | 1082.13 | 94.31 | 8.01 | 1355.53 | 169.13 |
The table shows a pattern where domestic production is generally valued at a relatively low US dollar value per carat, while imports are valued at considerably more, while export prices float marginally above production values but generally at much lower values than imports. In South Africa, if production, import and export trends are analysed by value rather than volume, the years 2007 and 2012 stand out as ones in which large values of rough diamonds were imported to South Africa: US$2.1 billion was imported in 2007, and US$1.1 billion in 2012. This, of course, is simply another way of saying that the value of rough diamonds imported, expressed in USpercarat,isfarhigherthanforproductionandexports.US per carat values for imports have significantly exceeded export and production values for all the years covered by the KP data, with the exception of 2012. Indeed, at times the difference is huge: in 2007, for example, the import value was US$1705 per carat, compared to US$93 and US$134 for production and export respectively. The industry generally maintains that the high US\(per carat values reflect imports from countries that produce high-quality rough diamonds, such as Lesotho, although even Lesotho KP values do not attain these peaks.
Thus, it is impossible to really understand the value and constitution of South African and other southern African stones, as value and valuation formulas are confidential. Confidential contracts seen by the author, and conversation with diamond valuators, reveal that, not infrequently, exceptional rough stones from other countries, such as Russia, may also be imported to Namibia, South Africa and Botswana by De Beers sightholders, while African stones may be exported at once to Europe. This generates the first question: Why are South African diamonds apparently worth so much less than imported ones (which mostly come from Namibia and Botswana)? Given the large volumes of imports of rough diamonds, a second question emerges, since export prices are so low: What happens to these high-valued imports? These two questions lead to an interesting analysis of the structural features of the institutional arrangements for the diamond market.
A quality explanation?
The GDV, the regulator, did give a structural explanation for high import and relatively lower export prices, arguing that imports are mainly gem quality (thus high priced) while exports are run of mine (low priced), which explains the difference in import and export prices (letter to Sharife, 2014). The industry also gives this explanation, that the value difference is simply determined by quality. But the industry and government explanation only makes sense if the high-quality imports (which the regulator says can be cut and polished cheaply locally) were sold domestically, and thus not exported at the same price or (if processed) a higher one. Or, to work as an explanation, the amount processed in this way would have to be volumetrically much smaller than the run of mine exports so that their higher price does not push up the aggregate total value of exports. But, ‘South Africa’s exports of rough diamonds have tended to exceed domestic production, due to the fact that significant amounts of imported diamonds … … are re-exported by these dealers and cutters in the rough form’ (SAMI 2006, 26). In other words, if local beneficiation is so small, then the export volumes must include most of the high-value imports (as domestic production on its own is not enough to generate these export volumes). So why is the export price so low?
Production, imports, exports and sales of diamonds
We are exploring the question of why imports seem to be much more valuable than domestic production and exports, since typically for South Africa the import price has tended to be more than five times the export price for rough diamonds going through the KP. It is important to note that the KP data cover only rough diamonds, not polished or finished diamonds which have been subject to beneficiation. Thus, in order to arrive at an accurate figure for potential mispricing, it is necessary to account for domestic sales, albeit that these are often to companies connected to the De Beers group, since these will be diamonds imported as rough, but after being polished disappear from the KP data. However, these data are obscured. The DMR confirmed that De Beers did not authorise disclosure of any information save for production value, and that the terms of Section 14 and 17 of the Statistics Act (Act 6 of 1999) prevented disclosure by the DMR, since these sections give confidentiality to a company which enjoys 75% or more of market share (email to Sharife, 2014). Thus, it is not possible to systematically correlate domestic sales volumes with import and export volumes until 2013, where ownership dilution brings De Beers under 75%.3
However, there is good reason to believe that in South Africa the local beneficiation industry is small, with just 300 polishers, reduced from 3000 in 2008 owing to declines in market demand (Parliament of the Republic of South Africa 2014) such that we can safely assume in our method for calculating mispricing (below) that the vast majority of imported diamonds are re-exported. The government also confirmed that most imported diamonds are exported in rough form, including when purchased by De Beer’s preferred buyers in South Africa, the DTC sightholders (email to Sharife, 2014), a position that can be further validated using UN Comtrade data and SADPMR reports (Bracking and Sharife 2014, 5–6; citing SADPMR 2011, 29, 2012, 36, 2013, 26).
Moreover, we do have some data on the volume of locally beneficiated diamonds which indicate that volumes are very small: 150,175 Ct (2012), 167,482 Ct (2011) and 129,417 Ct (2010) (SADPMR, of equivalent years).4 Unfortunately no data are available prior to 2010. These figures represent 32.25%, 12.41% and 1.31% of the imported volumes of rough gems in the respective years of 2010, 2011 and 2012, and the volumes polished locally would also include domestically mined gems (Sharife 2014a, 2014b). Also, there is reason to think that the data on exported polished gems are larger than actual beneficiation given the role of consultation as a service industry. For example, in SADPMR (2013, 26), in reference to polished exports, the ‘Annual report’ claims that ‘[A]n approximate of 46,548 carats (30.99%) of these exports were consignments which were imported for consultation and [which are] being returned to the owners.’ This means that only 103,627 Ct (not 150,175 Ct) were domestically polished output in 2012, which represents only 0.9% of the imported rough high-value gems in that year. In other words, we cannot exactly determine what proportion of imports are beneficiated, and thus removed from the KP valuation system (which doesn’t include polished diamonds) when subsequently re-exported. However, we can suggest that it is very low, which is supported by interview data and other related sources.5
(Mis)pricing of rough export diamonds controlling for import (volume and value), 2005–2012
In this next section we test further the explanation that the valuation regime reflects only quality, by analysing Kimberley Process Certification Scheme (KPCS) data. The KPCS requires that all importing and exporting companies place on each certificate of diamonds moved a value and volume, which enables us to establish an average price per carat. The KP data used here are available from 2004 to 2012 and covers production, import and export figures, subdivided into categories of volume, value and price per carat (in US dollars). These data are largely generated by the firms, and importing and exporting entities enter figures they know will be randomly checked on a light audit basis by the GDV. Despite De Beers having originally contributed these accessible data, the integrity of the values was placed in question by a representative of the company during the course of this research: ‘The primary purpose of the KP process (or the issuing of the certificates at least) is for Governments to certify the origin of diamonds, not to keep track of the volume and value of diamonds imported or exported’ (interview, De Beers Head of Media Relations, 2004). However, in addition to regulating against conflict diamonds, the avowed intent of the Kimberley Process was to increase transparency in the industry, so it would be most unfortunate if De Beers was not adhering to this objective, in a process in which it has been so influential. Here we assume that the avowed intent of the Kimberley Process and De Beers is to provide accurate data, and thus we assume figures to be accurate.
Using the KP data set, we compared the values and volumes of diamonds as they enter South Africa, as they are produced at mines and as they exit South Africa as exports (Table 2, columns 2–12). This gives us a measure of how much less diamonds are apparently worth as they exit the country than they are recorded as being worth as they are imported or mined. Because we cannot follow diamonds in the total global market we do not have a measure of transfer pricing for diamonds imported, that is, how much more or less they are worth when they cross a border from where they are mined. Instead, in order to arrive at a figure of transfer pricing of South Africa’s domestic production, we make the initial assumption that the values recorded for imported diamonds reflect an arm’s-length market price that De Beers is able to report correctly. We then assume that, whatever else happens in transit, they would be exported out of South Africa at least with the same price or more. We also assume that 100% of imports are re-exported, although this does potentially create a margin of error (see last section).6 Thus, using KP data for all three, we are exploring how the value of exports compares to the value of domestic production after allowing for imports – that is, removing imports from the volumetric figures at export at the values De Beers declared them to be worth on import.
| Domestic production | Imports | Exports | E–C Value of dom. prod. US\) (mill) [M] | D–G [I] Vol. of dom. exports | US$ per ct Dom. value of dom. prod. (adjusted for import) [M/I], [H] [H] | Difference actual export price and KP listed price US\((mill) [H–A], [J] | Value of mispricing US\) (mill) [I x J], [K] | Estimated De Beers % of exports by value [F] | Value of mispricing attributable to De Beers US\((mill) [FK] | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| KP Volume cts (mill) | KP Value US\) (mill) | KP US\(per ct [A] | Local beneficiation Volume cts (mill) | Local beneficiation as % imports (by vol.) | Volume cts (mill) [G] | Value US\) (mill) [C] | US\(per ct | Volume cts (mill) [D] | Value US\) (mill) [E] | US\(per ct | ||||||||
| 2004 | 14.09 | 1075.76 | 76.34 | 0.93 | 608.64 | 655.59 | 14.82 | 1835.69 | 123.84 | 1227.05 | 13.89 | 88.34 | 12.00 | 166.68 | 95 | 158.346 | ||
| 2005 | 15.56 | 1319.09 | 84.78 | 1.10 | 728.56 | 664.75 | 20.39 | 2148.29 | 105.37 | 1419.73 | 19.29 | 73.60 | −11.18 | −215.66 | 96 | 207.03 | ||
| 2006 | 14.93 | 1361.82 | 91.18 | 0.74 | 672.18 | 905.99 | 15.78 | 1930.28 | 122.32 | 1258.1 | 15.04 | 83.65 | −7.53 | −113.25 | 96 | 108.72 | ||
| 2007 | 15.21 | 1417.33 | 93.18 | 1.24 | 2113.89 | 1705.67 | 13.89 | 1867.33 | 134.44 | −246.56 | 12.65 | −19.49 | −112.67 | −1425.28 | 97 | 1382.52 | ||
| 2008 | 12.90 | 1236.24 | 95.82 | 0.68 | 582.25 | 850.09 | 10.14 | 1484.83 | 146.39 | 902.58 | 9.46 | 95.41 | −0.41 | −3.88 | 93 | 3.61 | ||
| 2009 | 6.14 | 885.54 | 144.23 | 0.66 | 357.20 | 544.73 | 9.55 | 1018.67 | 106.67 | 661.47 | 8.89 | 74.41 | −69.82 | −620.70 | 85 | 527.60 | ||
| 2010 | 8.86 | 1194.28 | 134.75 | 0.129 | 32.25 | 0.40 | 307.96 | 773.16 | 3.76 | 709.22 | 188.76 | 401.26 | 3.36 | 119.42 | −15.33 | −51.51 | 70 | 36.06 |
| 2011 | 7.04 | 1388.68 | 197.13 | 0.167 | 12.41 | 1.35 | 460.17 | 339.79 | 6.65 | 1370.45 | 205.94 | 910.28 | 5.3 | 171.75 | −25.38 | −134.51 | 99 | 133.16 |
| 2012 | 7.08 | 1027.13 | 145.13 | 0.150 | 1.31 | 11.47 | 1082.13 | 94.31 | 8.01 | 1355.53 | 169.13 | 273.4 | −3.46 | −79.02 | −224.15 | −775.56 | 55 | 426.56 |
| Total | 3340.34 | 2825.26 |
Notes: This table combines data from Kimberley Process records submitted by De Beers and Department of Minerals annual report data on average value per carat and volume of domestic production (De Beers 2012a; SAMI various years).
The bracketed letters in the headings give a label for the column and are used to track the calculations being made.
The method is illustrated in Table 2, as each step generates a new column moving progressively towards the right. This means we take the view that import prices are accurate and (since the industry regulators and De Beers told us as much) that imports of diamonds are mostly re-exported. We assume that their value on import is at least as much as they would export for. We then simply take that volume, at that price, away from the export revenues, find out how much volumetrically is left of exports and assume that this must come from domestic production. We then find the per carat value of this volume, again from KP data, and generate an adjusted price that the domestic production was exported at (once the imports at the declared value have been removed from the total) (Table 2, columns 13–16). Then we find the difference between the per carat price of domestic production at mine gate, and its price when exported, and multiply this by how much was exported. This gives a figure for mispricing on the far right of Table 2. The last year is an anomaly since exports were less than domestic production, such that we cannot say that they were exported in that year, although the difference in value of exports and production must still relate to production in store.
This method for all the years generates quite significant under-pricing of exports, which from 2005 to 2011 adds up to US\)2564.78 million, or US$3340.34 million if the year 2012 is added. While there may be some error related to time lag, such that the production recorded in one year may not be exported until the next, this error reduces in that we are looking at a number of years. Finally in Table 2, in columns 18 and 19 we estimate how much of this under-pricing of exports can be attributed to De Beers by weighting the overall mispricing figure by their market share, which generates a figure of US$2825.26 for the years 2004 to 2012. However, sales to DTC sightholders that are re-exported where the DTC is related to the De Beers group of companies could inflate this figure, attributable to the De Beers group of companies as a whole, if included.
Lost value
There is no ready explanation available for the value apparently ‘lost’ in the South African jurisdiction related to import values not being reflected in export values, or at least domestic production values not reflecting in export values. The discrepancies in Table 2 above suggest that either or both of two things are occurring: import parcel values (as required on a parcel for KP certification purposes) are being inflated and/or export values deflated. This would have the result of increasing the value of import credits and/or reducing income recorded in South Africa, and thus would reduce the tax burden of importing and exporting mine companies. Since in most instances it can be assumed that these transactions occur between entities under common control, it is correct to refer here to likely transfer pricing abuse. It would also appear as though the regulator (the GDV) has struggled to date, with the possible exception of 2012, to address this. In 2012, the GDV noted an import volume 8.5 times larger than any previous year’s import volume, which seems to have warranted further scrutiny given its particularly anomalous scale.
In 2007, import price manipulation appears evident: over 1.2 million Ct were imported at over US$1700 per carat or more than US$2.1 billion in import value. Former De Beers insiders and GDVs asserted to the authors that this price was highly improbable. Total export sales for the 2007 period, including imports, were US$1.86 billion for 13.8 million Ct. If controlled for import data (volume and value), 12.6 million Ct of rough diamonds produced for export in 2007 had a sales/transfer value of a negative –US$25.9 per carat. Using the KP production price (US$93.1/Ct) multiplied by the actual-hypothetical export volume (less import volume) generates a potential loss in value of US$1.1 billion of the total diamonds being exported by De Beers, against their respective KP prices on entry to South Africa (remembering that the control assumes imports are exported at the recorded value on entry).
In short, in 2007, if the KP import certificates are correct, the South African production of diamonds would have been exported at a loss. In this instance, including imports (re-exported) in the data for total exports serves to generate the impression that domestic production is being exported at a fair price, that is, above the price recorded by SAMI at the mines. However, controlling for imports shows that it is being exported at an under-invoiced price. It is probable that mispricing is spread over a number of years by including high-value imported diamonds (imported not for the purpose of sale but on a ‘round-tripping’ basis)7 to generate the impression that the arm’s-length price assigned to domestic exports is fair even though, when controlling for imports, it turns out to be less than the value of production. This has the potential effect of externalising profits while discouraging higher wage claims, as higher sales values are realised further up the value chain in other sovereignties.
Similarly, profits can be externalised by over-inflating import values and transferring money between De Beers entities attached to these sales, but this method does not test for that. Instead we are highlighting differences in values recorded for domestic diamonds at the point of production, controlling for imports and then suggesting that domestic real value is not accurately represented in the export values recorded. That being said, there is also some suggestion in the figures that import overpricing is present. For example, although DTC data were classified as confidential, De Beers’ ‘Report to society’ lists sales to the DTCs (preferred invited buyers) as worth only US$670 million (De Beers 2011, 2012b), generating a discrepancy of more than US$1.3 billion in ‘unsold’ diamonds imported in 2007 and possibly re-exported, plausibly back to De Beers’ head office in Botswana.8
In 2012, import volume manipulation seems likely, with volumes increased by an average 10.5 million Ct in comparison to the preceding eight years. From 2011 to 2012, there was a 747% increase in import volume – in excess of 11 million Ct. This imported volume (11.47 million Ct) is indicative of large quantities of low-quality boart (poorly crystallised diamonds) mixed in with higher value imports. The value of imported diamonds per carat was US$94.31, differing significantly from previous average values per carat of US$400–900 per carat. Import volumes far exceeded export volumes too. It is unlikely that the ‘inventory’ of millions in low-quality diamonds is designated for export in one single year. As with 2007, and given De Beers’ contraction in equity ownership in South Africa as a consequence of first the Diamonds Amendment Act 2006, and in the second instance the more assertive BEE equity share proportions provided for by the extension to BEE legislation in 2012, the Broad-Based Black Economic Empowerment Amendment Act, 2013, it is possible that the company has an incentive to frame the mines and the production from them as less valuable than they might in fact be in order to reduce the value of the ownership share that they must divest. This would be serviced by the slow re-export of boart mixed with home production.
From 2009–2011, undervaluing of exports estimated with this methodology appears to have generated US$806 million in ‘lost’ value for diamond exports compared with KP registered values. From 2005–2009, excluding the years 2007 and 2012, this increases to US$1.143 billion in lost value, value which, if it had appeared on the balance sheet, would have been subject to tax and royalties. Therefore, both 2007 and 2012 appear to represent circumstances of trade mispricing, where diamonds were moved for re-export and inventory in volumes and at values that could enable round-tripping, profit externalisation and tax avoidance, in the absence of another plausible explanation.
Tax effects of lost value
What then can one conclude about the ‘valuation’ difference, for South Africa, for these selected years, and the tax revenue implications of this? In other words, what is the difference between the intra-firm value assigned to export and import parcels and their fair, or arm’s-length, market value? It is not clear what a benchmark ‘fair market value’ might be, since it is not clear which prices might be entirely immune from deflation or inflation, given the market-setting qualities of the market oligopoly. To generate an estimate, we take either the export and production price in each year, whichever is the higher, as a proxy for the ‘fair market value’, and then calculate the overvaluation gap by comparing this with the import value given on the KP certificate. We are assuming that there is no underlying quality difference between gems of overseas and South African origin, which the industry would refute, given that the industry explanation of price differences is an assertion of underlying quality differences. However, most mines produce a range of different qualities irrespective of place.
Using this approach, the estimated inflation of import value over the period 2004–2012 is nominally US$3,935,638,201, or US$3.9 billion. At a conservative exchange rate of ZAR7 to US$1, then, this amounts to ZAR27.2 billion. However, as has been noted, 2012 was an anomalous year, characterised by a huge volume of imports (8.5 times more than the next-biggest volume of imports, 2011), at what seems an oddly low price, US$94.31 per carat, which is not only lower than all export prices but also lower than all production prices since 2008. As a result, the 2012 score reduces the estimates of inflated value considerably. For the years 2004–2011, then, the estimated inflated value of imports is US$4,794,104,875, or roughly a nominal US$4.8 billion/ZAR33.6 billion (Bracking and Sharife 2014).
Diamond pricing and valuation
We suggested above that it looks probable that the initial price paid to the De Beers Company in Namibia or Botswana for rough high-quality imports to South Africa was more than the price registered as the diamond left again. This difference would represent a non-taxable rent that is being externalised to another De Beers subsidiary in a different tax jurisdiction, which in the case of Botswana would be an extremely low tax jurisdiction. These discrepancies in the values of diamonds reported suggest that price can be determined by extra-market factors. Explaining the agency within these processes involves an interpretation of the nature of the market and De Beer’s dominant position in it, the diamond’s intrinsic nature as a commodity and the valuation system in place.
Starting with the product, the determination of the price of rough diamonds is complex. Indeed, the mining and selling of gemstone-quality diamonds has been defined by a basic paradox since 1867 when large deposits were discovered outside Kimberley, whereby the ‘specialness’ of diamonds was constructed by associating the stone, by means of its rarity, indestructibility and perceived beauty, with particular lifestyles of opulence, romance, consumption and privilege. The discovery of large reserves in South Africa promised great wealth to those who mined and sold it, if supply could be controlled in a long-term revenue-maximising manner, and if demand could be increased by making diamonds appealing and semi-affordable for the middle as well as upper classes. Getting both components right has been the spectacularly successful strategy of De Beers, which until recently produced a large share of the world’s rough diamonds, exercised control over virtually all the rest through its Central Selling Organisation and played the leading role in shaping consumer attitudes towards the stone.
In effect, through clever marketing to entrench the association of diamonds with glamour, romance and marriage, this ‘rarest’ of stones has become commonly owned by middle- and upper-class women in Europe and North America, and significant inroads have been made in other, especially Chinese and Japanese, markets. A second component of the limitation of supply has been preventing a secondary or resale market for diamonds from developing: after all, given the large-scale mining of diamonds for over a century, a vibrant secondary market would risk undermining annual new production and push prices down. The undermining of secondary markets emanates seamlessly from the ‘A diamond is forever’ concept, where selling your diamond becomes associated with acknowledging a broken personal relationship. On the other hand, diamonds are consistently marketed as a good ‘investment’, implying capital gains at some point should the owner indeed want to sell. Though the investment value of exclusive high-end diamond jewellery may indeed be positive (though unlikely to be spectacular), there is little doubt that most diamonds owned by most men and women would struggle to be sold, given the absence of a secondary market, and would not achieve even close to the purchase price.
The LCSV research protocol for understanding valuation systems terms this set of ideas a ‘discursive framing’, or overall conceptualisation for market participants of what is being sold. Notions of diamonds as ‘priceless’ and ‘forever’ require no single or standardised price, since each is notionally unique. Thus for rough and finished diamonds an array of prices can exist, apparently depending on quality. This is not to say that price determination is completely ad hoc; it is still partially dependent on a range of conventional economic factors (competitive supply and demand) and also more strategic variables associated with market power enjoyed by larger diamond companies who supply to cutting and polishing firms, jewellery firms and the retail market. However, within the marketisation process, De Beers has been able to deploy, given its market power, its own calculative entity to condition prices. This is embodied in its ‘Price book’, where the value to be assigned to a particular rough diamond or parcel of rough diamonds can vary considerably depending on variations in the ‘four Cs’ – clarity, colour, carats and ‘cuttability’ – which are employed as criteria for rough diamond valuation. The vast majority of South Africa’s domestic production generates diamonds in the low-value gem, ‘rough’ diamond market, where this somewhat imprecise pricing system prevails. Here, price is varied by the role of the expert valuator and his or her authority, necessarily involving differing judgements in assessing the four Cs, none of which have a clear scientific calibration.
One would expect price to vary marginally among qualified experts, but this possibility of flexible pricing potentially allows diamond companies to assign values to diamonds using additional criteria, such as for tax planning purposes, or the requirements of managing profit flows within the global interests of the conglomerate. This arbitrary price can be maintained up to eventual sale with the overall value being realised once a final ‘arm’s-length’ purchaser, such as a retail customer, is persuaded that the price is fair. The alteration of value while the diamond is moved from mine to purchaser is derivative of global strategy, while there is a sense in which even the final price is ‘performed’, given the lack of an intrinsic value for stones. The independent verification of the value of a diamond assigned by a diamond company in any particular place in this market structure is then obscured further by the complex valuation of the four Cs.
Apart from the nature of the product, there are also the structural features of oligopolistic markets that condition value, and the nature of institutional regulation and confidentiality. In the LCSV protocol, these are features of the overall institutional arrangement that conditions and structures the market. In this case, assessing the integrity of price is made near impossible by the confidentiality enjoyed by De Beers in valuing the quality of imports and exports, given that no arm’s-length sale actually takes place. This process is even a secret from the GDV, which is nonetheless expected to independently analyse all imports and exports parcels, totalling millions annually, in more than 12,000 different categories, without access to the De Beers valuation criteria.
Resource curse or marketisation process?
To summarise, the market for diamonds in South Africa can be viewed as a particular institutional assemblage in which De Beers is the dominant actor and is able to influence the behaviour of buyers, sellers and, critically, the regulator. This power is contributed to by the particular nature of diamonds as a commodity with little intrinsic value, bought by the eventual consumer in relation to a discursive framing of romance and ‘foreverness’. Because diamonds are then rarely bought or sold again in a secondary market, there is a weak sense of an arm’s-length price. Instead, arbitrariness and non-materiality contribute to the pricing system through the discursive way in which diamonds are culturally embedded (see Miller 2002). De Beers is also assigned a powerful position from which to determine diamond values by its control of the calculative entity in the diamond market, ‘The price book’, in which the prices of carats are recorded, which acts as a black box in that the expertise of the corporate producer is left as a technical and unquestioned expertism. In sum, and in the conceptual language of the LCSV research protocol (see Bracking et al. 2014), we have a valuation system in which the institutional arrangements are heavily weighted in favour of the interests of the producer and corporate sector; where the calculative entity framing price is also largely under their influence, centrally contained in the ‘The price book’, and where the cultural framing of the commodity deters questioning of the pricing system.
These empirical observations lead to ‘real world’ consequences in terms of, first, an interesting theoretical conundrum, and second, a policy challenge. Empirically, the findings are significant because they contribute to the evidence of tax and economic injustice in the political economy of mining. The figures above suggest significant discrepancies in US\(per carat rates for imports and exports, discrepancies which suggest a possible (mis)valuation of rough diamond parcels which serves to inflate import credits for tax purposes and devalue domestic production for export while also reducing royalties due. Values assigned by companies thus condition low revenues for the South African Revenue Service and the people of South Africa. These are ‘real world’ effects in terms of values assigned which have consequences for the distribution of reward and profit to various stakeholders. The possibility of spending on schools, hospitals and so forth is undermined by this behaviour, which contributes to a wider pattern of illicit financial flows exported unjustly from South Africa and developing countries more generally (Baker 2005; Ndikumana and Boyce 2008), particularly from the minerals sector (Bracking 2015b; Mottiar 2015; Sharife 2015a).
The theoretical conundrum is that in this valuation system, an ontologically fixed definition and measurement of value cannot be established because it is largely performed by the valuation system in place, or the ‘social processes behind the constitution of value’ (Beckert 2009, 254), a valuation system which privileges the power of the De Beers corporation. This is a similar finding to that of MacKenzie in terms of the valuation of ABS CDOs9 which contributed to the financial crash of 2008 (MacKenzie 2011), where a calculative mechanism for accounting for correlation in underlying portfolio assets was less a mathematical assessment of risk, than a device generated by sellers in order to meet the requirements of ratings agency classifications, that is, was informed by how ratings agencies would evaluate the eventual product. MacKenzie (2011, 1784) sees this as a clear example of ‘reactivity’ (citing Espeland and Sauder 2007), where the effects of evaluation or ranking change what is being evaluated or ranked. The value assigned to diamonds, similarly to ABS CDOs, is only realised as price at the retail sale, and is somewhat arbitrary before. This generates the policy challenge: fairness in tax is compounded by an a priori arbitrariness in the determination of value. Without recourse to arm’s-length or fair price, how can fair tax be articulated?
Other work on pricing systems for minerals has shown the difficulties of readjusting valuation systems to meet economic justice concerns such as ‘fair trade’ gold (Childs 2014a, 2014b), rubies (Hilson 2014) or efforts to turn capital ‘green’ (Bracking 2015a, forthcoming). However, in the case here, we are far from a situation where anyone but the mining industry is able to condition prices, given the relative weakness of South African trade unions and economic justice groups, and the derogation of government, by omission, commission or design, to reform the regulatory framework in the direction of a more equitable distribution of reward to all stakeholders. Why the government of South Africa has this policy stance is a matter beyond the scope of this article. In a potentially functionalist methodology it could be addressed by systematically analysing the cross-holdings, ownership and shared ownership interests of members of the South African elite and mining houses, which are many.
But this would be entirely reductive of the institutional forces underscoring the political economy, wherein the presence of one player is inextricably linked to a host of other parties, all of which operate as collective entities. According to a former senior employee instrumental in developing the mining bill, De Beers used the full force of its corporate power to bully the government into a position of vulnerability by politically and economically disabling and infantilising the government, claiming lack of expertise, experience and more. Elsewhere, the political history of how De Beers managed to ensure zero taxation and royalties through an arguably fraudulent letter as well as the subsequent hijacking of the state-created entity – the State Diamond Trader – coerced into utilising ‘donated’ De Beers staff to value De Beers diamonds, that the company would then receive exemptions on, has been explored (Sharife 2014a) as well as the institutional undermining of South Africa’s political economy, using Anglo American’s dominance to hold a Parliamentary inquiry hostage (see Sharife 2015b on the Lettergate scandal). Beyond this influencing, we can also explain policy alignment with corporate interests in a more ideological sense; as commensurate to the underlying logics of a neoliberal economic policy stance in South Africa (Bond 2014) that assumes foreign capital to be a scarce and attractive commodity whose constantly threatened withdrawal would have adverse economic consequences, although neither premise is probably true (Bond 2015; see Dean 2003; Hay 2004).
Conclusion
There has been much written about the ‘resource curse’; how resource abundance does not necessarily lead to growth or good development outcomes in poorer countries. In response, campaigners have argued that those countries using resources to fund development should require that the fiscal contribution of mining be significant, in the absence of other growth multipliers and in the context of the permanent loss to the sovereign natural capital base. However, relying on increasing taxes when the whole value system is conditioned by capital-holders does not appear to hold out much promise of success. By using the LCSV research protocol (Bracking et al. 2014) we can identify that an actor’s power over a particular configuration of the three elements – institutional assemblage, discursive framing and calculative entity – in this case leads to a marketisation process wherein value is largely performed, or has been historically, by the key power-holder, De Beers. In this oligopolistic market it is hard to ascertain how much value is being passed through any particular jurisdiction, or indeed how much the price of a natural resource that is exported has an assigned ‘fair’ price in relation to any empirical or materially independent measurement of value. It is an epistemological challenge to establish a fair tax contribution in such a context, and a new basis for ‘fairness’ needs to be developed, probably drawing on contributory and distributional principles.
The theoretical problem is how to understand the real in relation to the performed, in that the implications of our argument for the notion of ‘mis’ pricing requires further theorisation. Given the analysis above, it should be clear that ‘mis’ cannot be understood as a ‘wrong’ price in relation to a ‘correct’ price that actually exists somewhere, which would conform to a positivist epistemology, even though the tax justice discourse relies on this being a possibility. In our methodology we used the proxies that the industry itself provides: KP listed prices to connote value in the measurement of possible import overvaluation and/or export undervaluation, and the higher of the sales or production price, again from KP, as a proxy of ‘real’ price in order to estimate ‘lost’ externalised value (and profits, and therefore tax) which we believe is a consequence of import overvaluation.
However, none of these proxies of value can be privileged ontologically as ‘real’ value. Instead, ‘mis’ valuation can only be understood in relation to the economic injustice that the price consequentially delivers to the parties involved: to the workers, the peoples of South Africa, the consumers, and the shareholders and workers of the De Beers company. This reminds us that the LCSV protocol, and concepts such as assemblage, calculative device and so forth, are principally a means to explore empirical realities, but do not foundationally explain the relative power of different actors and classes in capitalism as a social order, and how these influence the configuration of valuation systems from the outside and from within. The outstanding question is to explore whether these iniquitous marketisation processes are particularly privileged by the nature of resources, which would restore some explanatory power to the resource curse thesis; or whether it is more generically the historical trajectories of imperialism, capitalist accumulation and post-coloniality which allow such unjust marketisation structures to occur. The latter is certainly the context in which they were built and in which they are being reproduced. In this sense, a reform of the valuation system and marketisation process in which it is embedded, inter alia, requires wider reform of South Africa’s highly unequal class and power structures.