Case studies · 1995
The '327' China government bond futures incident
A broker dumped a massive last-minute sell order on a government bond futures contract to avoid a squeeze, and regulators shut the whole market down for 18 years.
What happened
A contract with an unresolved price
In December 1992, the Shanghai Stock Exchange opened proprietary trading in Treasury bond futures. One contract, on a newly issued 3-year Ministry of Finance bond, became known in the market simply by its code: "327". From October 1993, individual investors, not just institutions, could trade it. By the standards of China's young capital markets, it was a huge and liquid product.
The "327" bond was unusual for a bond future because its final redemption value was not fixed at issue. On top of a coupon (reported as 9.5% in most English-language sources, 9.7% in one), the Ministry of Finance would pay an inflation-linked subsidy on redemption, and the size of that subsidy had not yet been announced. With inflation running near 22% in 1994, the market spent the second half of that year arguing about how large the subsidy would be. That single unresolved number was the entire dispute: whoever guessed right on the subsidy would make money on the futures price, and whoever guessed wrong would lose it.
The short, the long, and the announcement
Shanghai Wanguo Securities, chaired by Guan Jinsheng, was short the "327" contract in enormous size, betting the subsidy would stay low. China Economic Development Trust and Investment Corporation, together with allied buyers, was long, betting the government would raise it. On 23 February 1995, the Ministry of Finance announced the final redemption terms, and the subsidy favoured the long side. The contract's price swung sharply higher during the session, from a previous close of 148.21 yuan (per 100 face value) to an afternoon peak of 151.98 yuan, by one detailed secondary account. Wanguo was now facing a loss that reports put anywhere between roughly RMB 1.6 billion and RMB 6 billion; that range, rather than a single number, is what the record supports, because no single regulatory document with an exact figure has surfaced in English.
The dump order
Rather than post more margin or accept the loss, Wanguo used its own trading access in the last few minutes of the session, around 16:22, to push an enormous sell order into the market. Reports of the exact size vary: the widely cited figure is around 10.56 million lots (about RMB 211.2 billion face value), while a separate academic summary gives 7.3 million lots (about RMB 146 billion). Both numbers circulate in secondary sources and neither reconciles with the other, so the honest position is a range rather than a settled figure. Whatever the true size, the order was large enough to force the price down hard, from the 151.98 peak to 147.40 yuan, before the close.
The exchange reverses the trade
The Shanghai Stock Exchange held an emergency meeting that same evening and declared all trades placed in the contract during those final minutes invalid, reinstating an earlier valid price of 151.30 yuan as the official close. In effect, the exchange reversed Wanguo's dump order after the fact, by exchange decree rather than by any pre-agreed rule.
Arrests, a merger, and an 18-year shutdown
The affair was not fixed by a change to position limits or margin rules. On 17 May 1995, the China Securities Regulatory Commission, backed by the State Council, suspended Treasury bond futures trading nationwide, with all positions to be wound down by the end of that month. Guan Jinsheng was arrested two days later, on 19 May 1995, and was sentenced on 3 February 1997 to 17 years in prison, reportedly on charges connected to embezzlement rather than any dedicated market-manipulation statute, since China had no such law at the time. Wanguo itself did not survive: in 1996 it was merged into Shenyin Securities, forming Shenyin Wanguo, and the Wanguo name disappeared. China's government bond futures market then stayed closed for roughly 18 years. Only in September 2013 did the China Financial Futures Exchange (CFFEX) relaunch Treasury bond futures, with a standardised 5-year contract, followed by a 10-year contract in March 2015 and a 2-year contract in August 2018.
A second, separate scandal on the same contract
A contemporaneous 1997 US law review article also describes a separate, earlier episode in February 1994, in which a different broker, Shanghai International Securities, was accused of ignoring exchange trading limits and dumping bond futures on the same "327" contract to escape its own losses, at an estimated cost of about 100 million US dollars. The firm name, the loss figure, and the date do not match the 1995 Wanguo affair. These are best treated as two separate scandals on the same contract, both of which fed into the same 1995 nationwide suspension, rather than one source misdating a single event.
| Date | Event |
|---|---|
| Dec 1992 | Shanghai Stock Exchange opens proprietary Treasury bond futures trading, including the contract that becomes known as "327". |
| Oct 1993 | Individual investors are allowed into Treasury bond futures trading, not just institutions. |
| Late 1994 | The "327" bond's price rises as rumours build that the government will raise the inflation-linked redemption subsidy, with inflation near 22%. |
| 23 Feb 1995 | The Ministry of Finance's redemption announcement favours the long side; the price swings from 148.21 to a peak of 151.98. In the final minutes, Wanguo dumps a huge sell order, pushing the price to 147.40 before the close. |
| 23 Feb 1995, evening | The exchange holds an emergency meeting and declares the final minutes' trades invalid, reinstating a 151.30 closing price. |
| 17 May 1995 | The CSRC, with State Council backing, suspends Treasury bond futures trading nationwide, with positions wound down by end of May. |
| 19 May 1995 | Guan Jinsheng, Wanguo's chairman, is arrested. |
| 1996 | Wanguo Securities is merged into Shenyin Securities, forming Shenyin Wanguo; the Wanguo name disappears. |
| 3 Feb 1997 | Guan Jinsheng is sentenced to 17 years in prison by the Shanghai No. 1 Intermediate People's Court. |
| Sep 2013 | CFFEX relaunches Chinese government bond futures with a 5-year contract, ending an 18-year gap. A 10-year contract follows in March 2015, a 2-year contract in August 2018. |
The mechanics, in course language
This case sits alongside Lecture 6's material on bond futures, margin, and leverage, and it touches Lecture 2's material on margin and clearing. Three features make it a useful, if extreme, teaching example.
A contract with no fixed fair value
A modern government bond future, like the CFFEX contracts trading today, is written on a standardised notional bond, so its fair value is a matter of interest-rate forecasting. The "327" contract's underlying bond was not like that: its redemption value depended on a government subsidy that had not yet been set. That made the contract's fair value genuinely undetermined until a single announcement, not just uncertain in the way a normal bond future is uncertain.
Margin as the trigger for insolvency
Margin turned a price move into a solvency event. Treasury bond futures at the time required brokers to hold roughly 500 yuan of margin for every 20,000 yuan of contract face value, about 2.5%. That is high leverage even by futures-market standards, and it is the same mechanism Lecture 6 sets out: margin turns market risk into a liquidity and solvency problem. The losing side must find cash, or in Wanguo's case capital, fast, rather than simply accept a smaller position value.
A rulebook with no answer for size-driven manipulation
What broke was not the price move itself but the response to it. Rather than post more margin or close out in an orderly way, Wanguo used its own trading access to push a huge sell order through the market in the final minutes, trying to force the settlement price down and shrink its own loss. That is an attempt at price manipulation to escape a squeeze, not a hedge or an arbitrage trade. The exchange's response, cancelling trades after the fact, shows a market whose rulebook had not anticipated this kind of last-minute, size-driven manipulation, and whose fix was an emergency intervention rather than a pre-set circuit breaker or position limit.
Removal rather than repair
The policy response afterwards is itself a lesson in market design. Rather than patch the specific rule that had been exploited, such as a position limit, margin scaled to position size, or a circuit breaker, the regulator suspended the entire product nationwide for about 18 years. Sometimes the cost of a badly designed market is not a repair but the removal of the instrument until the infrastructure, clearing, margining, position limits, and circuit breakers, can be rebuilt from scratch. When China reintroduced government bond futures in 2013, it did so on a new exchange (CFFEX), with a standardised notional bond rather than a specific cash bond with a disputed subsidy, price limits, and an initial margin requirement set deliberately higher, at 3%, than the roughly 2.5% used in 1995.
The mathematics
Two figures fix the scale of the mismatch in this case: how much leverage a 2.5% margin requirement actually implied, and how far a routine-looking intraday price move already exceeded what that leverage could absorb. The calculation follows the same convention as lecture6.tex's Treasury futures material: margin expressed as a percentage of contract notional, and leverage as face value divided by margin posted.
Step 1. Margin as a percentage of notional. With a contract face value of RMB 20,000 per lot and margin of RMB 500 per lot:
$$\text{margin\_pct} = \frac{500}{20{,}000} = 2.5\%$$
Step 2. The implied leverage follows directly.
$$\text{leverage} = \frac{20{,}000}{500} = 40\times$$
Step 3. That leverage sets the break-even move. At 40x leverage, the price move that wipes out 100% of the margin posted is just:
$$\text{break-even move} = \frac{1}{40} = 2.5\%$$
Step 4. The move that actually happened exceeded it. The reported intraday move from the post-announcement peak (151.98) down to Wanguo's own dumped price (147.40) was:
$$\frac{147.40 - 151.98}{151.98} = -3.01\%$$
That is already past the 2.5% break-even point. Taking the widely cited figure for Wanguo's final sell order, around 10.56 million lots, the total face value involved was:
$$10{,}560{,}000 \times 20{,}000 = \text{RMB } 211.2\text{bn}$$
Margin theoretically required on that notional, at 2.5%, is RMB 5.28 billion. Using the lower figure some sources give instead (7.3 million lots, about RMB 146 billion face value), the margin implied is still about RMB 3.65 billion. Either figure is far beyond what a single mid-sized 1990s Chinese brokerage could hold in spare capital, which is why a price move of only a few percentage points against a position of this size was a solvency event, not just a bad trading day.
margin = 500
face_value_per_lot = 20_000
margin_pct = margin / face_value_per_lot
leverage = face_value_per_lot / margin
lots_high = 10_560_000
lots_low = 7_300_000
total_face_high = lots_high * face_value_per_lot
total_face_low = lots_low * face_value_per_lot
margin_required_high = total_face_high * margin_pct
margin_required_low = total_face_low * margin_pct
move_peak_to_dump = (147.40 - 151.98) / 151.98
break_even_move = 1 / leverage
print(f"margin_pct = {margin_pct:.1%}")
print(f"leverage = {leverage:.0f}x")
print(f"total_face_high = RMB {total_face_high:,.0f}")
print(f"total_face_low = RMB {total_face_low:,.0f}")
print(f"margin_required_high = RMB {margin_required_high:,.0f}")
print(f"margin_required_low = RMB {margin_required_low:,.0f}")
print(f"move_peak_to_dump = {move_peak_to_dump:.2%}")
print(f"break_even_move = {break_even_move:.1%}")
Output
margin_pct = 2.5%
leverage = 40x
total_face_high = RMB 211,200,000,000
total_face_low = RMB 146,000,000,000
margin_required_high = RMB 5,280,000,000
margin_required_low = RMB 3,650,000,000
move_peak_to_dump = -3.01%
break_even_move = 2.5%
All outputs above are recomputed and checked: margin_pct = 2.5%, leverage = 40x, total_face_high = RMB 211.2bn, total_face_low = RMB 146.0bn, margin_required_high = RMB 5.28bn, margin_required_low = RMB 3.65bn, move_peak_to_dump = −3.01%, break_even_move = 2.5%. The margin and face-value inputs (500 yuan / 20,000 yuan) come from a contemporaneous 1997 law review citing The Economist, and are treated here as solid. The lot counts for Wanguo's order are a different matter: secondary sources disagree on the number, so this page presents both and lets the margin-required figure move across that range, rather than asserting a single confirmed regulatory number that does not exist.
Data and facts
| Quantity | Value | Source |
|---|---|---|
| Underlying bond | 1992-vintage 3-year Treasury bond, maturing Jun 1995 | FT via SCMP; Euromoney (2013) |
| Bond coupon | 9.5% (one source: 9.7%) | Hou (1997), N.Y.U. L. Rev., citing Japan Economic Newswire |
| Margin required per lot | RMB 500 on RMB 20,000 face value (2.5%) | Hou (1997), citing The Economist, 27 May 1995 |
| Contract face value per lot | RMB 20,000 | Derived from reported lot counts and total face value |
| Wanguo's final sell order (disputed range) | 7.3m–10.56m lots (≈RMB 146bn–211bn) | ebrary.net vs other retrospectives; range not resolved |
| Reported loss to Wanguo (disputed range) | ≈RMB 1.6bn–6bn | Multiple secondary retrospectives; no single primary figure found |
| Intraday price path (per 100 face value) | 148.21 → 151.98 → 147.40 → reinstated 151.30 | ebrary.net, "Treasury Bond Incident, Lessons" |
| Nationwide suspension announced | 17 May 1995 | SCMP; Euromoney (2013) |
| Guan Jinsheng's sentence | 17 years in prison | SCMP; Yuan Trends |
| Length of market suspension | ≈18 years (May 1995 → Sep 2013) | Euromoney (2013); FIA MarketVoice |
| CFFEX relaunch margin (2013) | 3% of contract value at launch | Euromoney (2013); Kapronasia |
The lesson
- High leverage on a thin margin turns a normal-sized price move into a full loss of capital. At 40x leverage (2.5% margin), it takes only a 2.5% adverse move to wipe out the margin posted, and the reported move here was already past that threshold.
- When a position cannot survive a price move honestly, the temptation is to attack the price itself. Wanguo did not hedge or raise capital; it dumped a huge sell order to force the settlement price down, which is a control failure and a market-design failure at once, not just a bad trade.
- A market's rulebook is tested by its worst day, not its average day. The Shanghai Stock Exchange had no pre-agreed circuit breaker or position limit that stopped a single broker moving the market in the last few minutes of a session. The fix arrived only after the damage was done, as an emergency, retroactive cancellation of trades.
- Sometimes regulators do not repair a broken product, they remove it. China did not patch the 1995 Treasury bond futures contract; it suspended the entire market for about 18 years and reintroduced bond futures only once a new exchange, new contract design, and new margin and price-limit rules were in place.
- Concentrated positions built on thin margin do not just risk the position holder. Wanguo's failure helped freeze an entire product category for a generation, the same "one firm's leverage becomes everyone's problem" pattern that recurs in other margin-driven cases in this course.
Where it appears in the course
Think about it
- Wanguo's margin was about 2.5% of contract face value, giving roughly 40x leverage. If the exchange had instead required 10% margin, how would the arithmetic in the worked calculation above have changed, and would Wanguo's dump order still have made sense as a way out?
- The exchange cancelled the final minutes' trades by an emergency, same-night decision rather than by any pre-agreed rule. What are the costs of running a market this way, even if it stops the immediate crisis?
- China waited about 18 years to reintroduce government bond futures, and did so with a new exchange, a standardised contract, and higher margin. Was suspending the whole market for that long a proportionate response, or could a narrower fix, such as position limits or a circuit breaker, have achieved the same safety at lower cost?
Sources
- South China Morning Post, "The '327 incident' that prompted a 10-year trading ban" (historical retrospective, citing the Financial Times). scmp.com
- Euromoney, "China: Government bond futures return after 18 years," 16 September 2013. euromoney.com
- FIA (Futures Industry Association) MarketVoice, Gary DeWaal, "China's futures markets: the door opens another crack more." fia.org
- Victor L. Hou, "Derivatives and Dialectics: The Evolution of the Chinese Futures Market," 72 New York University Law Review 175 (April 1997). nyulawreview.org
- ebrary.net, "Treasury Bond Incident, Lessons" (case summary with a detailed intraday price path). ebrary.net
- Yuan Trends, "Guan Jinsheng: The Sole Survivor of China's 327 Treasury Bond Scandal and His Enduring Legacy." yuantrends.com
- Kapronasia, "The Future of Treasury Bond Futures in China" (CFFEX relaunch timeline: 2013, 2015, 2018 contract additions). kapronasia.com
- IMF eLibrary, "Chapter 6: Treasury Futures," in The Future of China's Bond Market (confirms CFFEX launch dates: September 2013, March 2015, August 2018). elibrary.imf.org