'People buy diamonds out of vanity and they buy gold because they are too stupid to think of any other monetary system which will work - and I think vanity is probably a more attractive motive than stupidity.'
The above statement was made by Harry Oppenheimer, current chairman of De Beers and a man more aware than most of the amazing illusion surrounding diamonds that De Beers has managed to create and sustain. Diamonds are in fact cheap to produce and would be lower in price but for the global cartel operated by De Beers. Its aim is to maintain a strong monopoly position, an objective which it has successfully achieved for several decades, notwithstanding the pundits' predictions to the contrary. In controlling the diamond market, De Beers have exploited a relatively simple idea: put an armlock on production and keep prices high. What makes De Beers so special is its execution of the idea. Over the past 60 years the cartel has done for diamonds something that eluded the oil producers of OPEC and even the cocaine barons of the Medellin cartel. It had the muscle and the nerve to impose its own order on the market and it built a syndicate not for weeks or months but for decades.
This paper will attempt to examine these diamond prices as set by De Beers. To this end both the market structure and control system as operated by the cartel is introduced, before the choice of variables and model is specified. The third section presents the results of the regression analysis, and in the final section these results are evaluated.
Gem diamonds, as distinct from industrial diamonds, are different from most other minerals in that they are non-homogenous. Unlike gold for instance, they do not have a standardised unit price. Each diamond has to be individually classified and valued and different diamonds of the same price are not necessarily tradable substitutes. At the micro-level, the diamond market is not a single market, but embraces many sub-markets, with different prices, supply and demand characteristics. Price differentials exist of course, in most commodity markets but the lack of homogeneity in diamonds is such that market segmentation is much more of an issue than in otherwise comparable situations. Owing to the wide variation in diamond qualities, the volume of a producer's output is not a reliable indicator of the value of his production. As a result the interests of producers obviously vary greatly. Due to the considerable differences in the scale of production, producers are not solely concerned with market share. Marketability is a major consideration and co-operation between producers can be a strategy for maximising revenues. This is because different productions are often complementary and different types of stones are easier to sell at different times. A mix from different sources may be more marketable and realise a higher long-term average price than could individual marketing. What drives the major rough diamond producers to co-operate is that there is no guarantee that independent marketing will realise higher sales revenues than centralised selling over the life of a De Beers' sales contract.
Certain key characteristics of the market as described above render it suitable to cartelisation. Firstly, there are a small number of significant suppliers and rigorous barriers to entry. Secondly, the product is durable, has a high value to volume ratio and is easy to store. Thirdly, the demand for jewellery-quality gems, which make up 90% of the $5 billion rough diamond market is relatively price inelastic. The rewards to a cartel from controlling supply are therefore likely to be considerable, provided that the difficult tasks of establishing such a cartel and managing it can be accomplished.
De Beers and its associate companies buy of rough diamonds from the mines, value them and then sell them to 'sightholders'. The policy of the cartel is to maintain stability in both of price and supply, flattening out short-term fluctuations in the demand-supply equation. The central thrust of its strategy is to exercise control over the market and thus maintain rough prices at the highest sustainable level, in essence aiming for long-run revenue maximisation over the whole of the demand cycle rather than short-run market clearing at spot prices. Unlike other commodity cartels, the diamond cartel it controls both supply and influences demand, combining the roles of major distributor, marketing agency and buffer stock manager. It has developed an expertise in matching supply to demand and the financial strength to operate an extensive buffer stock, capable of holding rough diamonds temporarily off the market.
The cartel has six key characteristics. The first of these is the existence of producer quotas. Most significant producers have a long-term and exclusive contract to supply a certain proportion of De Beers annual sales. This proportionate contract ensures that the burden of weak markets is passed directly back to the mines. This commitment by the cartel to purchase regardless of the state of demand, provides guaranteed cashflow and price stability, which induces the manufacturing trade to continue buying confident that the market has long-term support. Thus, in a rising market the cartel benefits from both higher prices and stock appreciation as goods are sold from the buffer stock. Yet in a weak market, it bears the full brunt of financing the buffer stock and its commitment to purchase from producers.
Secondly the cartel has created a strong antidote to any producer's incentive to cheat, with the ability to release from its own stocks a supply of any type of diamond. This means that the stockpile-supported price can drop dramatically as the market is flooded with a similar type of stone, released from the buffer stock by De Beers. This punishment was enacted against Zaire in 1981 when it chose to market its production independently of the cartel to traders in Antwerp. Market clearing of similar type stones lead to substantial price discounting in that segment which took a considerable length of time to work itself out, even though Zaire had been considered the dominant producer in its market segment. Thirdly, De Beers acts as a swing producer. It can play this role credibly since its own mines are among the cheapest sources of fine diamonds in the world.
A fourth element of market control is operated at wholesale level. The cartel's role as buffer stock manager is reinforced through its external buying offices, competing with independent traders for diamonds mined outside its own production network. When markets are weak it mops up excess supply by outbidding the independent traders and conversely allowing them to bid as they wish when markets are strong.
An important element of the system is the participation of rough diamond cutters, the middlemen. On every fifth Monday of the year, a select group of diamond dealers is invited by De Beers to the London Headquarters of the cartel, to collect a sack tied with a ribbon. Inside is a cardboard box containing a selection of uncut diamonds worth on average between $2-$5 million, which more or less correspond to the dealer's prior request, but ultimately the choice of diamond supplied in the box is at the discretion of De Beers. A box must be either accepted or rejected in full and price haggling is not permitted. This ritual is called a 'sight'. The invitation to a sightholder to attend is considered a privilege within the industry, with a corresponding loss of status if permission to purchase is withdrawn.
Finally, the cartel pays careful attention to demand management, operating a, highly successful worldwide advertising campaign. This campaign is aimed at attracting the consumer's attention to the particular type of stone De Beers needs to sell, corresponding to the composition of the buffer stock at a point in time.
Dependent Y-Variable
Given the context of the earlier discussion, my dependent variable in this investigation is the wholesale price of diamonds between 1976 and 1990. More explicitly this corresponds to the price paid by De Beers to the mines when it makes the level of purchases agreed in the contract. During the 1970s commodity prices in general boomed. Gold, silver, platinum, art objets and diamonds all soared in price. Diamond prices rose sixfold, with the investment motive strong and the trade suffered from bouts of trade speculation. Market conditions changed rapidly in the 1980s. The early 1980s brought Reagonomics, a stronger dollar and a preference for financial rather than physical assets. A global recession ensued and the diamond market spent the best part of the next decade recovering and adjusting to a new set of circumstances.
Independent X-Variables
As already detailed diamond prices are artificial in the sense that they are set by a controlling interest and are supported by a stockpile. As such numerous variables may be employed in the price setting decision. Thus, it can be seen that by limiting my investigation to two independent variables, my analysis is going to be imperfect due to the absence of many other possible determining factors. Nevertheless, I have chosen two variables to examine their respective influences on diamond prices.
My first X variable, X1, is the annual volume of diamonds produced at mine level. The variable was chosen as all major producers operate under a long-term, quasi-exclusive contract to De Beers providing a certain proportion of the cartel's annual sales. Thus, weak market conditions directly affect production at mine level. The production figures used here are introduced with a one-year time-lag to diamond prices. World rough supplies more than doubled during the 1980s, so it is particularly interesting to examine how De Beers coped with this huge increase, coupled with a slowdown in demand. One would a priori expect a negative relationship between price and supply. Investigation should detail to what extent the cartel has managed to control this.
My second variable, X2, is the US/Yen exchange rate. This variable was chosen because of the growing importance of Asian demand in the diamond market. In the mid-1980s, high US real interest rates and the overvalued dollar were beginning to depress world demand. In 1985, following the Plaza Accord, the major economies embarked on a policy of economic co-operation and the dollar started to depreciate. The market for larger diamonds benefited and the growing Asian markets, notably Japan, Taiwan, South Korea and Thailand, started buying. Prices started rising as the dollar fell against the yen and other hard currencies. Thus one might posit a negative relationship between the exchange rate and prices.
I am employing the ordinary least squares estimation technique in this investigation, which will yield a line of 'best fit' corresponding to the data. The model takes the following form:
Yi = beta0 + beta1X1i + beta2X2i+ epsiloni
for the multiple regression case, where i represents the error term of the regression or the residual. The investigation will arrive at an estimate of the sign, size and significance of the unknown parameters beta0, beta1 and beta2.
The Data
The estimation of the regression line was achieved using the HUMMER econometric package, using 15 observations of annual data, dating from 1975 to 1990 with a one-year lag for production. The diamond market is highly secretive and prices at wholesale level are virtually impossible for the outsider to acquire. Thus, I have constructed a price index (base year 1976=100), using the price increases for rough diamonds as supplied to me by De Beers. It must be reiterated that these refer to wholesale level where an amazing feature of the system is that De Beers has never announced a decrease in nominal terms. Prices are in dollar terms and the index has been deflated using World Bank deflators. The production figures are measured in millions of carats and were obtained from an industry journal, Mining Annual Review. These should be accurate as they are updated on an annual basis where necessary to ensure precision. The exchange rate series was obtained from the Financial Times.
The Estimation
The results of the multiple regression are detailed below. The line of Best Fit has been estimated as:
Y = 2.19391 - 0.005276X1 - 0.004593X2
The correlation coefficient, R^2, indicates that 75 percent of the variation in Y can be explained by the linear influence of all the independent variables.
Independent Parameter Estimate t-Statistic
Variable Ho: i = 0
constant 2.919391 9.13874
X1 0.005276 -2.40822
X2 -0.004593 -5.21805
R2 = 0.7598
This is quite a satisfactory result but as with any econometric result it cannot be taken in isolation. By regressing Y on X1 alone, an R^2 of 0.21537 was yielded, showing that 21 percent of the variations in diamond prices were explained by variations in production figures.
Independent Parameter Estimate t-Statistic
Variable Ho: i = 0
constant 1.322931 8.28737
X1 0.004107 1.88899
R2 = 0.21537
By the same procedure on X2, an R^2 of 0.64398 was yielded, showing that 64% of variations in Y are explained by variations in X^2.
Independent Parameter Estimate t-Statistic
Variable
constant 2.194303 17.56708
X2 -0.002853 -4.84918
R^2 = 0.64398
This evaluation shall deal first with the results of the multiple regression, followed by an examination of the individual regression results and ending with a statistical evaluation of these results. In evaluating the results yielded by the regression, I intend to compare the estimates with the a priori assumptions of the author. As expected, in all, the two X variables, have a considerable impact on diamond prices (75 percent). The relationship with X1 (production) is negative as expected, as is that with the exchange rate. Examine the results more closely and the picture is not as clear-cut. A separate regression of the independent variable X1 on the independent variable X2 reveals an R^2 = 0.67365, indicating a high degree of muticollinearity, the problems engendered by the presence of which are a priori recognised by the author though they shall not be elucidated on here. Indeed, this is evident to such an extent that an individual analysis of Y on X1 yields a differing nature of relationship between the two than was revealed in the multiple regression case - this time indicating a positive relationship.
In order to make a relatively simple statistical evaluation of my results, I am going to examine the accompanying t-statistics in the previous tables. Considering the hypothesis that there is no relationship between the X and Y variables (Ho: i:= 0), one can examine the t-statistic which measures the ratio of the estimate to the standard error. An estimate of a parameter is statistically significant if the t-statistic associated with it, causes us to reject, at a particular significance level, the hypothesis that i is equal to zero. From the multiple regression, the estimates for both i and i, at both 5 percent and 10 percent significance levels are statistically significant. The estimate provided by regressing Y on X1 is statistically insignificant at a 5 percent level, but significant at a 10 percent level with a t-statistic of 1.88899. Examining the regression of Y on X2, the t-statistic is -4.84919, showing a high degree of statistical significance. This is a strong indication of a causal relationship. It suggests that the dollar/yen exchange rate plays an important role of diamond prices.
In the markets of today, anti-competitive rules are outlawing commercial agreements like cartels. But the legal competence of the nation state can only extend so far. Cartels, historically speaking, given the stringent conditions necessary for success, do not have a high survival rate. De Beers success in this environment is testimony to its ingenious domination of the trade. Unlike other commodity cartels, it both controls supply and influences demand, combining the roles of major distributor, marketing agency and buffer-stock manager. It has developed an expertise in matching supply to demand and the financial strength to hold diamonds temporarily off the market. But the cartel can never be infallible. Several factors in the present environment could pose a potential threat to its commanding position Economist Intelligence Unit (1994). De Beers sway over customers for rough diamonds is not matched by its influence on customers for polished diamonds, namely, jewellers and their suppliers. They are increasingly well organised, good at hard bargaining and make it difficult for De Beers set prices to pass smoothly through the system to the end buyer. Russia always adds uncertainty to the diamond market. Rumours of a Russian stockpile persist and increasing numbers of Russian stones have found their way to the market outside of the 'single channel' as operated by De Beers. Assuaging this threat will require all the political and diplomatic clout of the corporation. Yet, there is a degree of consensus within the industry that centralised selling and buffer stock management are in its collective and long-term interest. The diamond is the strongest material known and it is therefore fitting that the diamond cartel has been so resilient slightly ironic that the diamond cartel is not only extremely strong but also extremely old. Only time will tell whether De Beers will continue to control this market and whether diamonds really are forever.
[1] The author wishes to gratefully acknowledge the contribution of Professor Dermot McAleese to an earlier research project on the same subject.
Economist Intelligence Unit (1994) 'Diamonds: A Cartel and Its Future'