The Indian diamond trade - notes from a field visit

by Ajay Shah.

India and the democratisation of diamonds

Diamonds were first discovered in India, and until the 17th century, India was the only source of diamonds in the world. But the mines at Panna have now died down for all practical purposes. Rough diamonds come out of mines in South Africa, Angola, Sierra Leone, Russia, Australia, etc. De Beers was once 70% of the world's rough diamonds. Now they're down to 50% or so. They do "price stabilisation" and try to tell buyers of diamonds that they'll get a good return on diamonds as a financial investment.

Many years ago, the framework of the gems industry used to be one where the cutting of rough diamonds was done in Israel and Europe. It was once at Amsterdam, but in ~ 1911, the government did some nasty taxation. The (mostly Hassidic) Jewish traders moved shop to Antwerp, based on an informal understanding with the Belgian government that they would get a friendly environment of enforcement of taxation and (say) lack of harrassment about conversations with Geneva. So Antwerp is now the centre of gravity of the global diamond trade.

In the bad old days, before India came along, rough diamonds were classified into 3 groups: gems, industrial, and in-between. For many years, the in-between were a gray area - not quite clear what to do with them. In the 1950s (?) there was a chap in Antwerp who came up with a brilliant idea. He purchased a parcel of the in-between, and brought it to India. Indian labour was quite fine with dealing with these small stones, polishing them, making beauty out of them. Most important: Indian labour was able to do all this at low cost! This was the start of a revolution. Indian labour led to a democratisation of global diamond consumption. Diamonds went from big exotic things sold to the upperclass to little things sold in supermarkets to teenagers.

The transformation has been nothing less than spectacular: 11 out of 12 diamonds of the world are now polished in India, giving a revenue of \$10 billion a year! The remaining 1/12 are the biggest and highest quality value diamonds, where the master craftsmen of Israel still win. These big diamonds are a very high value business, so even though India scores 11/12 on a headcount of diamonds, only 2/3 of global polishing revenue reaches India.

\$10 billion a year is big when compared with software, the most famous Indian export. And, diamond polishing is highly labour intensive, so this gives you atleast one example of Indian domination of global exports in a labour-intensive manufacturing problem. It's interesting to see that once again, this is a complex, skill-intensive area; it isn't really like the Chinese-style low-skill low-wage manufacturing.

I have heard that in Antwerp, there is an S shaped road - totally 200 metres or so - that is the diamond market. When you walk there, the people are mostly (Hassidic) Jews and (Palanpuri Jain) Gujaratis. There are restaurants there serving Gujarati thalis. All the world's non-De-Beers rough diamonds come to Antwerp, get sold to (mostly Indian) buyers, and roughly half of that output goes back to Antwerp for sale of polished diamonds.

Folks in the Indian industry think that the production of artificial diamonds is not yet a realistic threat to De Beers. None of the Indian firms are thinking of setting up shop to make artificial diamonds. Their view is that even if new channels for roughs come about, they'll still need to come to India to get polished. I wondered if one day a machine will grow a perfect polished diamond with 57 faces, thus making the very polishing stage irrelevant.

From rough diamonds to polished diamonds

There are 2 channels through which you can buy rough diamonds: directly from De Beers or from "the open market" (mostly in Antwerp). There is an elaborate "Kimberley process" through which "conflict diamonds" (from African countries with civil wars raging) are blocked from the global trade: this seeks to eliminate diamond-related politics and mercenaries from hitting on those countries.

A key player in the wholesale rough diamond trade is De Beers Diamond Trading Company (DTC). There are 37 Indian firms that buy direct from De Beers. The way it seems to work is that the firm says to De Beers (or some other seller of rough diamonds) that it wants to buy a consignment of rough diamonds with some approximate parameters. Then you get to fly to De Beers in London (or to the market in Antwerp) every five weeks, where you get shown a pile of diamonds sitting on a brightly lit table. The price of the consignment is non-negotiable. The quality of rough diamonds in the pile is heterogeneous. The prospective buyer gets around 15 minutes to make a decision: to take it or leave it. The buyer typically rapidly inspects around 50 diamonds in that much time. Generally when faced with De Beers, the buyer tries to not say no, because if you do that too often, De Beers can cutoff your access to these transactions. The 93 global firms that DTC invites into these soirees go by the sweet name of "sightholders". As of December 2006, there were five listed firms who were DTC sightholders: Asian Star Co., Gitanjali Gems, Classic Diamonds, Shrenuj and Suashish Diamonds. (Source: Article in Business World by Rohit Viswanath).

De Beers (or the open market) merely sell rough diamonds. After buying roughs from one of these two channels, the buyer bears all business risk about uncertainty of revenues. The architecture of the Indian business is: to buy batches of rough diamonds, polish them, and figure out ways to sell the polished diamonds and turn a profit. The thumb rule is that part of the family sits outside the country, hangs around Antwerp and London in buying roughs, and sets up the distribution channel for selling polished diamonds.

The process of converting from a rough diamond to a polished diamond goes in many stages. Rough diamonds have a lot of heterogeneity! Sometimes the material has blemishes and dark blotches, and you can go all the way down to \$300 per carat. I saw one big fat diamond of ~ 20 carats going at \$300 per carat - it did not look nice and I wouldn't want it at all.

On the other hand, the material can be perfect and the result can go all the way to \$12000 per carat. Carats are a linear mapping to mass (200 mg/carat). The size of a shipment is measured in carats. The price per carat varies with: mean mass of diamond (obviously, small diamonds are cheaper), colour (whiter is better), clarity (less blemishes is better) and perfection of polishing or "cut". To use the phrase that everyone in the trade rattles off, "cut, carat, clarity, color".

The first stage of processing is to look carefully at the rough diamond, identify bits of dirty material visible from the exterior, and cut it off, so as to leave you a relatively unblemished rough diamond. A skilled person looks at the diamond and marks on it with a pen where he wants the cut to be made.

In the bad old days, a man then sat with a saw and cut the diamond so as to get rid of the bad stuff. This was labour intensive. And, more importantly, a man could only cut a saw on a single plane. The Israelis (who seem to drive technological change in this industry) came up with a new laser-driven machine which cuts the diamond, and it has an amazing ability to cut in complex ways. So a human carefully mounts the diamond (using a camera + screen which magnifies the rough diamond and his fingers by 50x for him to watch what he is doing), uses a standard PC-gaming joystick (!) to align things right on a screen, and presses a button. Then he settles down and watches while the laser cuts.

These machines first came from Israel. Then some small company in Gandhinagar (Gujarat) figured out how to make them. Now there are roughly 1000 of these laser cutting machines littered all over the country. Since these machines are expensive, they are run 3 shifts. India is great at extracting value out of scarce capital.

At this point, the rough diamond has been stripped of the bad stuff. Now the question is : what cut? There is a large number of "styles" (or geometries). A good cut is one which yields maximal "fire" or visual drama.

The standard solitaire diamond is a way of cutting that induces 57 faces (!). While this is an international standard cut there are ancient Indian cuts which have more "fire" than this 57-face cut when faced with relatively small diamonds. These ancient ideas had faded away but are now being revived.

There are two aspects to profit maximisation in the choice of a cut given a rough diamond. First, the choice of cut induces the extent of wastage of material. Second, different cuts induce a different price per milligram (since the visual drama of "fire" varies by cut). One wants to choose a cut with care, so as to minimise the wastage, but in the bottom line, maximise revenue.

For a small rough, a skilled human looks at the rough for 1-2 minutes, makes a decision on the cut, writes it down on a piece of paper, and moves on to the next. This decision might be suboptimal because humans are probably not that great at visualising 3-d structures that fit within the envelope of the surface of the rough diamond. In the history books, we know that the process of cutting some famous diamonds in history involved gruesome wastage owing to mistakes on these decisions.

The Israelis invented a machine, costing \$4000, which is sheer genius. The way it works is like this. You mount the rough and a 3d image of the rough appears on the screen. The computer figures out the geometry of the rough. Then the computer tries out all possible cuts. You can click on any of the geometries and the computer shows you how to fit that cut within the volume of the rough. You put in standing instructions for the price per milligram associated with various cuts - as of that day. The computer calculates the milligrams that you will get for each cut, and recommends the value maximising cut.

Customers are stupid and hanker after round numbers like 1 carat. So often, you can get a 0.8 carat well-cut diamond which has more fire than what is ostensibly a 1 carat diamond where suboptimal decisions were made on the cut. Remember this when you go to buy your first diamond.

A human takes one last look at the result of the computer and occasionally disagrees (!). I met many of these technicians and it is simply amazing, the kind of specialised skill and judgment that is in the heads of these very ordinary looking people. For the rest, this machine is a revolution. This machine hasn't yet been cloned in India. The first of these machines were only imported into India in 2000. Since these machines are also expensive, they are also run for 3 shifts.

This is the way in which the cut of a rough diamond is chosen, and the diamond is sent off to a factory for polishing. There is another amazing machine which takes a finished diamond and examines it closely. E.g. there is a 57-facet diamond. The computer closely inspects each of the 57 faces and scores each face on its level of perfection.

After all this - with the best thinking about a rough, with the best choice of cut, with the best polishing - each diamond is different. It may have flaws - little dirt within. It can have odd colours. In the end, when you look at it, it should be perfectly beautiful and white. These command the highest value.

Sometimes, it is bluish (very rare and prized). Often, there is blackness - not accepted at all in India but quite accepted in the US where they love the slogan "diamond" and don't worry that much about how it looks. To me, visually, black diamonds were interesting and surprising, but not out and out pretty.

De Beers used to have a problem with coloured diamonds - beautiful colours like yellow. Somehow, nobody accepted them. Indian cutters were the first to step forward and take interest. Out of gratitude, De Beers - till this day - has given India a monopoly on coloured diamonds: they won't sell coloured roughs to anyone else.

Roughs are surely heterogenous. On top of that, there are surprises in the processing: how you throw away dirty parts, how you choose a cut, how the polishing actually goes through, the colour and clarity of the output, human mistakes along the way. So an enormous effort is then required in starting with the output of the factory and reducing to little piles of homogeneous groups of diamonds. E.g. imagine a pile of 100 yellow diamonds of 1 carat which have the classic 57-facet cut, all of which are perfect yellow without any dirt. That's a homogeneous group. This is the parcel that gets sold back into the international market.

Alternatively, sometimes a jewellery or watch maker comes to India and gives locals a specific spec: "I want 50,000 diamonds of X kind and 10,000 diamonds of the Y kind". Then a consignment is made which carefully fits his requirement and sent back to him. This requires starting from a very large number of plausible sounding roughs, polishing all of them, and then isolating the 50,000 diamonds of X kind and 10,000 diamonds of Y kind out of this superset. Interestingly enough, the shipment of consignments is done by the likes of Brinks Arya - even though a small pouch is worth a million dollars, it works fine.

An amazing feature of Indian polishing is that our people are willing to do very small diamonds! I saw a pile of tiny diamonds which were 400 to the carat - i.e. 0.5 milligrams each. Each diamond on the pile was 57 facets! Young Indian labour sits with a magnifying glass polishing these TINY diamonds with 57 facets.

A pioneering Indian firm is Mahendra & Company. They have a business office at Panchratna (Opera House), a sorting and thinking centre at Kwality (Worli) and factories in Navsari: adding up to 6,000 workers. Their revenues (100% export) is roughly \$0.75 billion, out of total Indian exports of ~ \$10 billion.

Many other firms in India are alumni of Mahendra & Company, so there has been a Bangladeshi-garment-export type story of human capital building at this company and going on to seed the country. Almost all of Indian diamond polishing is made up of entrepreneurs from Palanpur (in Gujarat) and a few from Kathiawad.

We visited their Kwality office. There were rooms and rooms filled with millions of dollars of diamonds strewn on open tables, being sorted into homogeneous piles. Bright light on a 100 diamonds, each of 1 carat -- it is a truly beautiful sight to behold!

Looking forward for the Indian diamond trade

In terms of value a polished diamond is only a small part of the value of jewellery. The real high value stuff is where a jewellery store in New York sells a diamond ring for \$1000. In this, the wholesale price of the polished diamond is probably no more than \$200. To really capture value from the jewellery business, you have to be the brand and own the customer - the firms that do this are presently buying polished diamonds from India.

With India processing 11 out of 12 of the world's diamonds, there isn't that much upside left in growth of revenues from diamond polishing. Putting these two problems together, the upside for India consists of going up the value chain from diamonds to jewellery. Indian companies now dream of being the brand and selling directly to global customers.

Theft

How does a firm like Mahendra & Company deal with theft? They have a strongroom with steel walls and biometric access - only 3 senior people can get in. At any point in time, there are ~ 150,000 diamonds in the firm. Every day at night, all the diamonds are collected together, reconcilation is done, and the goods are put back into the strongroom. Every morning, the diamonds step out of the strong room for handling for the day. At an overt level, it looks like there is little incidence of theft.

However, there is a slow pace of theft taking place constantly at various points in the value chain, which is reputed to add up to 1.5% of the value of a consignment. On a base of \$10 billion a year, this rate of theft maps to \$150 million a year.

Two essential facts shape this environment:

1. Diamonds are not detected by metal detectors; it is impossible to find out if a person is carrying around 0.2 grams (1 carat) of diamond.
2. No two diamonds are alike.

Suppose I get 10 minutes alone with a consignment of 1 carat diamonds. I am smart and experienced. I open the consignment and look closely. I identify the 2 best diamonds in that batch (the right edge of the distribution) and take them OUT. I toss in 2 1-carat diamonds which are not of good quality. So the shipment is still the correct number of 1-carat diamonds.

Bottom line: I have stolen the gap between the price of 2 1-carat diamonds of high quality versus 2 1-carat diamonds of low quality. This style of theft does NOT get trapped by a reconcilation system that merely counts all diamonds and verifies they are intact.

The size of revenues from this theft is driven by the (max-min) of the value of diamonds in a given consignment. Maybe, the introduction of computers which analyse all 57 faces and give out numerical scores for perfection (as described above) will reduce the dispersion of the quality of gems in one consignment, in which case the profit from theft will go down, but the problem will not go away easily.

Episodes of such theft could repeat many times in the life of a packet. Each time, the two best diamonds are taken out and two worst diamonds are put in.

It's a reason for having as few situations as possible for NOT having a situation where non-family get a chance to spend ten minutes alone with the goods.

The crucial infrastructure that enables such theft is: a market for me to buy 2 low-quality 1-carat diamonds and to sell 2 high-quality 1-carat diamonds. This market is found on the pavement behind Panchratna building! The place is teeming with eateries. We were taken for a drive around the region. Apparently the bulk of people walking around there are participants in the open outcry spot market for diamonds. There are three kinds of participants: speculators, jewellers and thieves. People walk around here and negotiate trades for small lots of diamonds. Most people know each other - it is literally an open outcry market but being acted out on the road. Two people stand in front of each other and talk. One man pulls out a bunch of diamonds and shows them to the other in his palm. The other person agrees. The DVP happens away from the market.

Principal-agent problems; role of family business

Sometimes the domination of family run businesses in this field is viewed as an anachronistic relic which will inevitably give way to the rise of modern professional corporations. But there are two powerful reasons in favour of family businesses in the diamond polishing field. It is not accidental that the two peculiar subsets of humanity that have dominated this trade are the Hassidic Jews and Palanpuri Jain Gujaratis.

First, the issue of theft as described above. Second, the need for globally dispersed elements of the family. One brother sits in London, visits De Beers, buys a batch of rough diamonds (which requires making a judgment about whether the batch is a good deal). One brother sits in India and does the polishing (which requires making judgments about optimality of cuts, and handling issues of theft). One brother sits in Antwerp and sells the polished diamonds (which is pure bargaining).

There are huge principal-agent problems in the problem of diamonds because diamonds are NOT a commodity - each diamond is different and subtle judgments + negotiations need to be made. And machines like metal detectors cannot detect diamonds: this reduces verifiability. The principal-agent problems can be overcome by bonds of blood.

There are hundreds of families in this trade. I think there is a selectivity process where some families experience a decay of family ties, and individuals start maximising for themselves and not for the family. Those families are then less able to efficiently organise production and lose market share. Through this selection process, the core families who dominate the trade are very traditional families who have powerful family ties, and an entire cultural framework that goes with that.

Such reasoning may also help explain why firms like De Beers didn't make the next step up from rough diamonds to polishing and trading in polished diamonds. De Beers is a giant mining company, which has figured out how to be a corporation (and not just a family) based on acute mistrust and invasive security measures against low-grade mine workers. Such security measures would not work when dealing with the high skill staff that have to figure in the life of a diamond after it is plucked out of the earth.

How the trade would change if equipment could attack the problem of measurement and verifiability

I can't help fantasise about how this could play out in the future. I think one can build a machine where a diamond is mounted, then a pencil of light plays on it in various ways, and the diamond is watched by a CCD array. (It's easier to have a static pencil of light and a static CCD array, and have the diamond move). I think the "fire" of the diamond is related to the contrast seen in the frames grabbed by the CCD array. We can look at the heterogeneity in intensity (i.e. the interquartile range of the histogram in digital camera displays) and the heterogeneity in colour. This allows us to quantify the fire of the diamond. So one can then have a highly homogeneous basket of 100 diamonds, each of which is between 0.99 and 1.01 carats, and all of which are tightly homogeneous in fire.

Once this is done, two big things will change. First, this problem of theft will go down greatly. In this case one could visualise a shift away from family dominated business. Second, one could then start a futures market on diamonds!

Shift to China

One of the people told me a story where he had ~ 500 people in a factory on the outskirts of Bombay (Chinchpokli). The average wage was Rs.8,500 per month. The staff went on strike and he had a huge disaster because he was holding inventory of roughs and was obligated to deliver polished diamonds to a foreign customer.

He gave up, shut down the factory, and setup shop in a remote location in China. The whole world knows about the importance of Indian firms in diamond polishing, so a red carpet is laid out for Indian firms setting up shop. This is happening in China, and in places like Angola / Botswana which are trying to go up the value chain from mining to polishing : they are trying to say that we'll only sell you roughs if you will process some roughs locally.

He said: the 500-man facility in China works great for three reasons. First, their tax system is better. Second, there is good labour law (i.e. hire and fire, no unions, no strikes). Third, the factory is in a remote location, and there is no Panchratna market which is the crucial infrastructure required for theft! Simple reconcilation based on counting diamonds suffices to remove the ~ 1.5% incidence of theft, which adds up to big money when compared with the profit rate of the business.

Update (10 Dec 2006): I just noticed a great article on SSRN titled Ethnic Networks, Extralegal Certainty, and Globalisation: Peering into the Diamond Industry by Barak D. Richman.