The Sumitomo Copper Scandal: How Mr 5% Hid a Decade of Losses

The Press Release Nobody Expected

At 5:30 p.m. Tokyo time on Thursday 13 June 1996, Sumitomo Corporation's head office in the Harumi district of Chuo-ku released a two-page statement to the Japanese press club. Its chief trader in the non-ferrous metals division, Yasuo Hamanaka, had been running unauthorised copper trades for roughly ten years, and the company was writing down a $1.8 billion loss. The number, the statement said, was preliminary. An internal audit would refine it. Within four months it had — upward, to $2.6 billion, then the largest loss ever disclosed by a trading company.

The release hit the Reuters terminal in London at 9:30 a.m., one minute before the London Metal Exchange opened its Thursday ring. Copper was trading at $2,165 per tonne. By the close of the ring session it was $2,050. By the following Friday it had broken below $1,900, a fall of more than 25 per cent in two weeks. Cash-three-month spreads, which Hamanaka had kept tight by hoarding warrants against a shortage of deliverable metal, collapsed from $195 backwardation to a contango within days. Traders who had been short of physical copper for most of the preceding year, paying borrowing costs to carry their positions against Hamanaka's squeeze, were suddenly the market's only bid (Gilbert, 1996).

The man whose book had held the global copper price up was a soft-spoken salaryman who had joined Sumitomo in 1970, worked his way up through the non-ferrous metals division, and never been assigned anywhere else.

Sumitomo's Non-Ferrous Desk

Sumitomo Corporation is one of Japan's sogo shosha — general trading companies whose business models span physical commodities, industrial finance, and long-term supply contracts with the zaibatsu-affiliated manufacturing groups. In copper, Sumitomo was a natural participant. Its own affiliate Sumitomo Metal Mining was one of the largest Japanese copper producers, and Japanese wire rod, cable, and construction industries were among the largest copper consumers in the world. A trading book that hedged physical flows on the LME was consistent with the business.

Hamanaka was hired out of high school, not university — unusual for a career trader at a sogo shosha. He spent five years learning the physical side of copper, another five on Sumitomo's Tokyo desk, and by the early 1980s had moved onto the LME book. His first major trading losses, according to his later testimony and the reconstruction by the Ministry of Finance investigators, came during the unwinding of the late-1980s Tokyo asset bubble, when speculative positions he had taken using Sumitomo's credit lines went against him. Rather than book the loss, he doubled the position. When that trade also lost money, he opened an unauthorised account at Merrill Lynch in London and began running a second, hidden book alongside the official one (Kilman and Howe, 1997).

The hidden book was not sophisticated. It was a leveraged long position in copper futures, financed by off-exchange credit arrangements, rolled forward each contract month. The complexity came from how Hamanaka disguised it — forged broker confirmation faxes, fictitious counterparties, signed authorisations he produced on his own typewriter at home.

Why the Physical Market Could Be Cornered

Copper is a small market. In the mid-1990s annual global refined copper consumption was roughly 11 million tonnes; annual LME-deliverable warrant stock rarely exceeded 500,000 tonnes, and often fell below 100,000. The LME warrant system allowed a large holder of warrants to demand physical delivery, forcing shorts either to source metal from other warehouses — often at premium — or to pay the borrowing cost to roll their position forward. A participant who controlled a meaningful share of the free float of warehouse stock could, in effect, set the near-dated price.

This was the market Hamanaka learned to dominate. By 1993 his combined physical and paper position represented, by the CFTC's later estimate, between one-third and one-half of LME warrant stocks in any given month. He purchased copper from mines under long-term contracts and held the metal in LME warehouses. He bought three-month futures to establish forward exposure. And, crucially, he lent metal to physical users while simultaneously borrowing it back through the futures curve, allowing him to control both the spot squeeze and the spread structure (CFTC Sumitomo Order, 1998).

The pattern was detectable. What it looked like, from the outside, was a persistent backwardation in copper — near-dated metal trading at a premium to metal three months out, month after month, for years at a time. Copper producers hedging their output were forced to sell into that backwardation and roll their hedges at a loss. Industrial consumers who tried to take delivery paid storage premiums. The spread itself became a tax, and it was being paid to Hamanaka's book.

The Counterparties

The CFTC's 1998 order, and the parallel actions brought by private plaintiffs in the Southern District of New York from 1997 through 2004, named a set of Wall Street firms that had acted as financing counterparties to Hamanaka. Merrill Lynch extended him off-exchange credit facilities through its London metals subsidiary. Morgan Stanley arranged structured copper-linked notes that allowed him to accumulate additional exposure without showing it on the LME. Global Minerals and Metals Corporation, a New York-based physical trader run by R. David Campbell, executed warehouse swaps on his behalf. Credit Lyonnais Rouse, the French broker's metals subsidiary, and later Chase Manhattan's commodities unit, each cleared portions of his book.

None of the settlements included an admission that these firms had known the trades were unauthorised. The allegation was narrower — that they had extended credit and executed trades for a single customer in sizes that, taken with publicly reported LME positioning, should have raised manipulation concerns. Merrill settled its share in 1999 for $25 million without admission; Morgan Stanley paid $35 million in 2006 in a private civil class action; Chase Manhattan paid $125 million in 2006; Credit Lyonnais Rouse paid $37.5 million. Combined civil recoveries against the financing counterparties eventually exceeded $600 million (Kilman and Howe, 1997; Jorion, 2007).

1995 — The Year the Squeeze Broke

For a decade Hamanaka had managed to run the scheme quietly because the losses had been manageable and the squeezes he engineered had been self-financing. Each round of higher prices generated mark-to-market gains on the existing long, which offset the roll costs and sometimes allowed a drawdown of the hidden book. The market had accepted that copper traded at a structural backwardation because Japanese demand was strong. In early 1995 that narrative broke down.

By the summer of 1995 LME warehouse stocks were declining at a pace physical demand could not explain. Copper had broken above $3,000 per tonne in January 1995, fallen back on Chinese demand weakness, then climbed again through the autumn as Hamanaka absorbed available warrants. Cash-three-month spreads blew out to over $200 of backwardation — a level the LME's own surveillance staff later described in internal correspondence as inconsistent with any rational inventory model. Producers, including Chilean state miner Codelco and US-based Phelps Dodge, complained publicly.

In November 1995 the LME's special committee on market conduct opened a confidential review. That December the UK Securities and Investments Board — the predecessor of the Financial Services Authority — formally requested information from Sumitomo on its LME positioning. The CFTC, through its division of enforcement, began a parallel inquiry into whether US firms were aiding the dominant position. Hamanaka's supervisor at Sumitomo, the general manager of the non-ferrous metals division, began asking internal questions that Hamanaka could not answer.

The unravelling was gradual and then sudden. On 5 June 1996 Sumitomo announced internally that Hamanaka was being reassigned from the copper desk pending review. He was escorted from his workstation and his terminal access revoked. A forensic reconstruction of his books, conducted by Sumitomo's own auditors alongside external accountants from Tohmatsu, ran for the following week. On 13 June the company concluded that the unauthorised book had accumulated losses of $1.8 billion. The public disclosure came that afternoon.

The Collapse

What happened next was the largest unwinding of a single position in copper market history. Sumitomo's investment banks — Goldman Sachs was engaged as liquidation adviser — had to close out more than 2 million tonnes of paper copper exposure and manage the release of more than 500,000 tonnes of physical warrants without creating a complete price collapse. They succeeded only partially. Copper fell 25 per cent in ten trading days. Cash-three-month spreads flipped to contango. LME warehouse stocks, which had been 113,000 tonnes on the day of the announcement, rose to more than 450,000 tonnes by year-end as Sumitomo released metal into the market.

By October 1996 Hamanaka had been indicted in Tokyo for forgery of private documents and fraud. He was arrested on 22 October, held without bail, and appeared in Tokyo District Court on 24 February 1997 to enter a guilty plea. His testimony was that he had acted alone, that no supervisor had directed him to conceal losses, and that the forgeries were his responsibility. On 13 March 1998 Judge Hiroshi Kondo sentenced him to eight years in prison with labour — among the harshest sentences ever imposed in Japan for a white-collar offence. He served his term in Fuchu Prison and was released in July 2005.

Settlements and the Regulatory Rewrite

The civil and regulatory aftermath ran on a longer clock than the criminal case. On 11 May 1998 the CFTC announced its order finding that Sumitomo had manipulated copper prices in violation of the Commodity Exchange Act, and imposed a $125 million civil penalty — at the time the largest in the agency's history. In a parallel settlement with the UK SIB, Sumitomo paid £8 million. Both orders were negotiated without an admission of wilful misconduct by the firm, but the factual findings — that Hamanaka's book had engineered a squeeze, and that Sumitomo's internal controls had been materially deficient — were accepted by the company.

The LME's institutional response was more consequential. Through 1996 and 1997 the exchange introduced what became known as the lending guidance, a set of rules requiring large holders of long positions approaching the spot date to lend metal to the market at capped premiums, limiting the squeeze mechanic Hamanaka had exploited. Warehouse stock reporting was tightened, and beneficial-ownership disclosure was required for positions over a percentage threshold. The CFTC used the Sumitomo findings as a template in subsequent manipulation cases, and the jurisdictional theory — that foreign traders whose off-exchange behaviour affected US-regulated markets fell within US enforcement reach — was later written into Section 753 of the Dodd-Frank Act and its enhanced position-limit regime.

The broader lesson landed alongside the one taught the previous year by Nick Leeson's Singapore desk. The 1990s produced two canonical Asian rogue-trader cases — Leeson at Barings in 1995 and Hamanaka at Sumitomo in 1996 — each built on the same structural weakness: a trader with back-office control over his own settlement, operating from a geographically distant office that parent-company management treated as a profit centre rather than an exposure. Two years after Sumitomo, the Long-Term Capital Management collapse demonstrated the same dynamic scaled to a different instrument — a single concentrated book that was too large to unwind without a market-wide dislocation.

Risk management as a discipline, as described in the second edition of the Financial Risk Manager Handbook, absorbed the Sumitomo lesson specifically. The case became the textbook example of operational risk divorced from market risk — a position whose size and direction were not the primary failure mode, but whose concealment and financing structure were (Jorion, 2007). That framing fed into the Basel II operational-risk capital charges introduced from 2004, and into the commodity-specific position-limit expansions enacted after 2010.

A Corridor in Fuchu Prison

On the day in 2005 when Hamanaka completed his sentence and walked out of Fuchu Prison, the spot LME copper price had just crossed $3,800 per tonne — higher than any level he had ever maintained. Global copper demand, driven by Chinese construction, had outgrown his old squeeze many times over. He gave no interviews. He returned, according to one former colleague who spoke to a Japanese newspaper anonymously, to a family home he had kept during his incarceration, and was not seen in the metals business again. The institutions that had enabled him — the LME, the sogo shosha trading culture, the back-office arrangements of the non-ferrous division — all survived him, each modestly more supervised than before, and each aware that a man working quietly for ten years inside a trusted firm could move the global price of an industrial metal more than any central bank ever had.