Case studies · 2004
China Aviation Oil and the jet-fuel options collapse
A state-owned jet-fuel trader sold speculative options against rising oil, and doubling down on losing positions ended in bankruptcy.
What happened
A hedger by mandate
China Aviation Oil (Singapore) Corporation Ltd, known as CAO, was the overseas trading arm of a Chinese state-owned company. Its job was simple and, on paper, low-risk: buy jet fuel on the international market and supply it to Chinese airlines. CAO had listed on the Singapore Exchange in December 2001 and had built a reputation as a well-run state enterprise. Because its core business needed to buy fuel, the natural hedge was protection against oil prices rising. A hedger with a real, physical exposure is supposed to use derivatives to reduce risk, not add to it.
The desk turns speculator
From the second half of 2003, CAO's trading desk did the opposite. It began writing (selling) call options on crude oil and jet-fuel-linked benchmarks, alongside related swaps, building a position that would only make money if oil prices fell. Selling a call brings in premium immediately, which can look like steady income, but it commits the seller to hand over oil at a fixed strike price if the market rises above it. As long as prices stay low, the premium is pure profit. The moment prices rise past the strike, the losses are open-ended, because there is no ceiling on how high oil can go.
Doubling down instead of closing out
Oil did not fall. Through the first half of 2004, prices climbed, and CAO's positions moved further into loss: about US$5.8 million by the first quarter, growing to around US$30 million by mid-year. CEO Chen Jiulin rolled the losing options into new ones, larger and with later expiry dates, rather than closing them. Rolling a loss does not remove it; it changes its size and pushes the reckoning further into the future, on the bet that the market will eventually turn back.
The record high and the forced close
The market did not turn back. On 25 and 26 October 2004, NYMEX crude oil hit a contemporary record high of about US$55.65 to US$55.67 a barrel. By then CAO's derivatives book had grown to a notional position of about 52 million barrels, more than three times its actual annual jet-fuel import volume of about 15 million barrels. The positions were ordered closed at a large loss.
A share sale and a wall of margin calls
Two other things happened at the same time, and together they turned a large trading loss into a full collapse. First, on 20 October 2004, days after the losses became known internally, CAO's parent company sold 15% of its stake in CAO to institutional investors through a placement arranged by Deutsche Bank, raising about US$108 million, without telling the bank or the buyers about the derivatives losses. Second, once the losses became public, CAO's eight counterparty banks, including Mitsui, Barclays Capital, Standard Bank, Sumitomo Mitsui and Fortis, issued margin and payment demands totalling about US$247.5 million within weeks. Trading in CAO shares was suspended on 29 November 2004, and the next day CAO disclosed total derivatives losses of about US$550 million (roughly US$390 million realised and US$160 million unrealised) and applied to the Singapore High Court for protection while it restructured.
Rescue, charges and sentence
A rescue followed. CAO's parent (CAOHC), the oil major BP, and a Temasek Holdings subsidiary agreed to inject a combined roughly US$130 million into CAO as part of a 2005 to 2006 debt-and-equity restructuring, in exchange for a majority of the enlarged share capital. Chen Jiulin and four other executives were charged in June 2005. On 21 March 2006, a Singapore court sentenced Chen to four years and three months in prison plus a fine of S$335,000 (about US$207,000), after he pleaded guilty to six charges, including failing to disclose the losses, making false statements, and conspiring to deceive Deutsche Bank over the share placement. The Monetary Authority of Singapore confirmed on 16 December 2004 that it was investigating whether any law or regulation had been broken.
This case is not simply a company that lost money because oil prices rose. CAO's underlying business, buying fuel, actually benefits from being protected against rising prices, so a genuine hedge should have softened the blow of higher oil, not multiplied it. The trading desk built a position several times larger than the business it was meant to serve, in the wrong direction, and kept doubling down rather than admitting the trade had failed.
| Date | Event |
|---|---|
| Dec 2001 | CAO lists on the Singapore Exchange. |
| H2 2003 | CAO's trading desk shifts from hedging into speculative options positions betting on falling oil prices. |
| Q1–Q2 2004 | Oil prices rise instead of falling. Mark-to-market losses reach about US$5.8m in Q1 and about US$30m by mid-year; CEO Chen Jiulin rolls the losing positions into larger, longer-dated ones instead of closing them. |
| 20 Oct 2004 | CAO's parent sells 15% of CAO to institutional investors, raising about US$108m, without disclosing the trading losses to the placement bank or the buyers. |
| 25–26 Oct 2004 | NYMEX crude hits a record high of about US$55.65–55.67/bbl. CAO's positions, by then about 52 million barrels, are ordered closed at a large loss. |
| 29 Nov 2004 | Counterparty banks' margin and payment demands reach about US$247.5m across eight institutions. CAO shares are suspended on the SGX. |
| 30 Nov 2004 | CAO discloses total derivatives losses of about US$550m and applies to the Singapore High Court for protection. |
| 2005–2006 | CAOHC, BP and Temasek subsidiary Aranda Investments agree to inject about US$130m as part of a restructuring; Chen Jiulin and four others are charged (June 2005). |
| 21 Mar 2006 | Chen Jiulin is sentenced to four years three months in prison plus a S$335,000 fine after pleading guilty to six charges. |
The mechanics, in course language
Hedger and writer, pointing the same way
The hedger/writer distinction is the key to this whole case. CAO's real business made it a natural hedger: it had to buy jet fuel, so it wanted protection against oil rising. Selling call options does the opposite. As the option writer, CAO collected premium up front in exchange for an obligation: if the market price rose above the strike, CAO had to deliver oil at the strike price, or cash-settle the difference, no matter how far the market had moved. That is the standard asymmetric payoff of a short option position. The seller's gain is capped at the premium received. The seller's loss, if the underlying keeps moving against the position, has no natural ceiling.
Because CAO's derivatives book pointed the same way as its physical exposure (both effectively "short oil"), the two did not offset each other. They added up. A position meant to be a hedge had turned into a second, larger bet layered on top of the business's own risk, so when oil rose, both the physical cost of fuel and the derivatives book moved against CAO at the same time.
Rolling losses instead of realising them
The rolling behaviour matters just as much as the direction of the bet. Every time a losing option approached expiry, CAO closed it and replaced it with a larger, longer-dated one rather than realising the loss. The same mechanical trap applies to any margined position: an unrealised loss can be closed (crystallised, and usually still survivable) or rolled forward (deferred, but now larger). CAO chose to roll repeatedly, which is why a loss that was around US$30 million by mid-2004 had become roughly US$550 million by the time it was finally disclosed.
Margin turns a paper loss into a cash emergency
Margin is what turned the paper loss into an emergency. CAO's counterparties are all banks that demand collateral against positions that move against a counterparty, the same margining mechanism that applies to futures and OTC derivatives alike. As oil kept rising, those margin and payment demands, about US$247.5 million in total, arrived as immediate cash calls, not as a slow accounting write-down. It was the cash calls, not the eventual disclosed total, that actually forced CAO to stop trading and seek court protection.
A governance failure alongside the market bet
A separate governance failure sat alongside the market bet. CAO's parent company raised about US$108 million by selling shares on 20 October 2004, days after the losses were known internally, without telling the placement bank or the buyers. That is why Chen Jiulin's eventual guilty plea combined market-risk charges with disclosure failures and a charge of conspiring to deceive the bank that arranged the placement. A single bad trade rarely ends a company on its own. It was the combination of an oversized, wrong-way derivatives position, repeated rolling of losses, margin calls that demanded real cash, and a non-disclosure that turned a trading loss into a criminal conviction and a near-total collapse.
The mathematics
Two figures fix the scale of the mismatch: how far CAO's derivatives book had grown relative to the fuel volume it was meant to hedge, and what a single, fully sourced trade lost. The first figure is the leverage ratio:
$$\text{Leverage ratio} = \frac{\text{peak derivatives position}}{\text{annual jet-fuel imports}} = \frac{52{,}000{,}000}{15{,}000{,}000} \approx 3.47$$
CAO's speculative options book was around 3.5 times the size of the physical volume it needed to protect. A position that large is no longer a hedge against a real exposure; it is a directional bet several times bigger than the underlying business.
A single, fully sourced trade shows exactly how a short call loses money. CAO had sold Brent crude call options with a strike of US$34.25 a barrel on a notional 300,000 barrels. The options expired around 29 October 2004, when the settlement price was US$49.34 a barrel:
$$\text{Loss per barrel} = \text{settlement} - \text{strike} = 49.34 - 34.25 = 15.09 \text{ USD/bbl}$$
$$\text{Total loss on this trade} = 15.09 \times 300{,}000 = \$4{,}527{,}000$$
That matches the trade-press figure of "about US$4.5 million" on this single position. As the call seller, CAO was obliged to deliver, or cash-settle, oil at US$34.25 a barrel when the market price was US$49.34, a loss of US$15.09 on every one of the 300,000 barrels in this one trade alone. This is one leg out of many in the full 52-million-barrel book. The full blotter of strikes and maturities was never made public, so this single trade stands as a worked illustration of the mechanism, not as a share of the total loss.
position_barrels = 52_000_000 # peak notional derivatives position, barrels
annual_import_barrels = 15_000_000 # CAO's actual annual jet-fuel import volume
leverage_ratio = position_barrels / annual_import_barrels
print(f"leverage_ratio = {leverage_ratio:.2f}")
strike, settle, contract_barrels = 34.25, 49.34, 300_000 # one sourced Brent call trade
loss_per_barrel = settle - strike
total_loss = loss_per_barrel * contract_barrels
print(f"loss_per_barrel = {loss_per_barrel:.2f} USD/bbl")
print(f"total_loss = ${total_loss:,.0f}")Output
leverage_ratio = 3.47
loss_per_barrel = 15.09 USD/bbl
total_loss = $4,527,000Both figures verify directly from the cited numbers: the 3.47 leverage ratio and the single sourced trade above. Reconstructing the full 52-million-barrel book strike by strike is not possible, because that data was never made public. Neither calculation here reproduces the full US$550 million loss trade by trade. That is not a gap in this page; it is a gap in what was ever disclosed.
Data and facts
| Quantity | Value | Source |
|---|---|---|
| Total disclosed derivatives loss | ≈US$550m (≈US$390m realised + ≈US$160m unrealised) | CAO disclosure, 30 Nov 2004, via Al Jazeera / China Daily |
| Peak derivatives position | ≈52 million barrels | Risk.net, "A wrong-way bet" |
| CAO's actual annual jet-fuel imports | ≈15 million barrels | Risk.net, "A wrong-way bet" |
| NYMEX crude peak, 25–26 Oct 2004 | ≈US$55.65–55.67/bbl | China Daily, 23 Dec 2004 (US$55.67); cross-checked at US$55.65 |
| Counterparty margin/payment demands (8 banks), by 29 Nov 2004 | ≈US$247.5m | Risk.net, "A wrong-way bet" |
| Parent-company share placement, 20 Oct 2004 | 15% stake sold, ≈US$108m raised | China Daily, 23 Dec 2004 (a US$111m figure appears in one secondary source and is unreconciled) |
| Rescue investment (CAOHC, BP, Aranda/Temasek) | ≈US$130m combined | Temasek Holdings press release, 2005 |
| Chen Jiulin's sentence | 4 years 3 months + S$335,000 fine (≈US$207,000) | China Daily, 22 Mar 2006 |
The lesson
- A hedger's own trading desk can become the biggest risk to the business it was meant to protect. CAO needed to buy fuel, but its options book ended up making that same exposure worse, because the "hedge" was actually a bet running in the same direction as the company's real risk.
- Rolling a losing position into a bigger one does not fix the problem, it changes its size. Every time CAO replaced a losing near-term option with a larger, longer-dated one, it converted a survivable loss into a potentially fatal one.
- Margin calls turn a mark-to-market loss into an immediate cash emergency. The US$550 million figure was an accounting total, but it was the roughly US$247.5 million of margin and payment demands landing within weeks that actually forced CAO to stop trading and seek court protection.
- A short option position has a payoff that is capped on the good side and open-ended on the bad side. Selling calls to collect premium looks like steady income until the underlying moves hard against the position, and then the loss has no natural ceiling.
- Concentrated leverage against a limited balance sheet turns an ordinary price move into a company-ending one. Oil rising from the high thirties to the mid-fifties dollars a barrel was a large but not extraordinary commodity move. Against a position over three times the size of the real underlying business, it was enough to wipe out the company's market value.
Related course topics
Think about it
- CAO's real business needed protection against rising oil prices, yet its trading desk built a position that only paid off if oil fell. At what point would you say that stopped being a hedge and became a speculative bet, and how would you design internal controls to catch that shift early?
- Chen Jiulin rolled losing positions into larger ones instead of closing them. If you were a risk manager at CAO in early 2004, what rule could you have put in place that would have forced a decision to close, rather than roll, once losses reached a certain level?
- CAO's parent company raised about US$108 million from investors on 20 October 2004 without disclosing known trading losses. Why does this kind of non-disclosure typically do more lasting damage to a company's credibility than the trading loss itself?
Sources
- Monetary Authority of Singapore, media release, "China Aviation Oil (Singapore)", 16 December 2004. mas.gov.sg
- China Daily, "Costly lessons from the CAO scandal", 23 December 2004. chinadaily.com.cn
- China Daily, "Ex-CEO of China Aviation Oil sentenced", 22 March 2006. chinadaily.com.cn
- Al Jazeera, "China oil firm runs up huge loss", 1 December 2004. aljazeera.com
- Risk.net, "A wrong-way bet" (trade-press reconstruction of CAO's positions, counterparties and margin demands). risk.net
- Temasek Holdings, press release, "CAOHC, BP and Aranda, a subsidiary of Temasek, To Invest US$130 Million In CAO", 2005. temasek.com.sg
- FinanceTrainingCourse.com, "China Aviation Oil (Singapore) Corporation Limited's Jet Fuel Scandal (2005) - Casestudy", 2014. financetrainingcourse.com
