Case studies · 2015
China's index-futures discount
After the 2015 stock-market crash, China's exchange restricted index-futures trading, and futures traded persistently below fair value for years.
What happened
A market at its peak
China's stock market ran hot through late 2014 and into the first half of 2015: the CSI 300 index roughly doubled between November 2014 and its June 2015 peak. On the very day that run-up was at its strongest, 16 April 2015, the China Financial Futures Exchange (CFFEX) launched two new stock index futures contracts, on the CSI 500 (small-cap, contract code IC) and the SSE 50 (large-cap, contract code IH), alongside the CSI 300 contract (code IF) that had traded since 2010. For a few months, all three contracts were actively traded, and by July 2015 CFFEX's index futures market ranked, by the World Federation of Exchanges' measure, as the world's most actively traded stock index futures market.
Black Monday and the clampdown
The market then turned. The sell-off that became known as China's 2015 stock market crash deepened through the summer, and on 24 August 2015, a day the financial press labelled "China's Black Monday", the Shanghai Composite fell by about 8.5%, its steepest one-day fall since 2007. Regulators moved to calm the market, and one of the levers they pulled was the index futures market itself. Between 25 August and 2 September 2015, CFFEX raised the margin on non-hedging index futures positions from 10% to 40% (hedging positions went from 10% to 20%), cut the daily position-opening limit for non-hedging accounts from 500 contracts to just 10 contracts per product per day, and increased the fee on same-day position closing roughly a hundredfold, from about 0.0115% to 0.23% of contract value. The tightened rules took full effect from 7 September 2015. From that date, opening more than 10 contracts a day in a single product was treated as abnormal trading and investigated.
A market "killed" within weeks
The market reacted fast. By 8-9 September 2015, Bloomberg reported that CSI 300 and CSI 500 futures trading volumes had collapsed by about 99% from their June 2015 peak. A market that had been the world's largest stock index futures venue two months earlier had, in the press's own words, effectively been killed.
A four-year unwind
Easing did not begin until 16 February 2017, and it took three further rounds, in September 2017, on 3 December 2018 (margins on CSI 300 and SSE 50 cut from 15% to 10%, CSI 500 margin halved to 15%, the non-hedging intraday position limit raised from 20 to 50 contracts, and the intraday closing fee cut from 0.069% to 0.046%), and again on 22 April 2019, before conditions moved back towards anything like their pre-2015 state. From the first restriction to the last recorded easing round, the process took roughly four years.
Background
A stock index future should, in principle, be one of the easiest instruments to price. An arbitrageur buys the underlying basket of shares, finances the purchase, collects the dividends the shares pay along the way, and the fair futures price follows a cost-of-carry relation. Two arbitrage trades keep the quoted futures price pinned close to that fair value. If futures trade above fair value, the trade is to buy the shares and sell the future (cash-and-carry). If futures trade below fair value, the mirror trade is to sell (short) the shares and buy the future (reverse cash-and-carry).
The structural weakness that made this case possible sat in the second trade. It needs genuine short-selling of the underlying basket, and in mainland China, securities lending for A-shares has long been tightly limited, so this arm of the arbitrage was already weak before 2015 even started. The restrictions that followed the 2015 crash then made the other side of the trade expensive and capped too: hedging or expressing a bearish view through index futures now meant steep margin, a ten-contract-a-day cap, and punitive intraday fees. Demand to hedge or to bet against the market was squeezed almost entirely into a market that could not correct itself, because there was no cheap way to arbitrage a too-low futures price back up when shorting the underlying shares at scale was not realistically available either.
The mechanics, in course language
This case is what happens when the standard no-arbitrage argument has one side of it disabled. Fair value for a stock index future follows the standard cost-of-carry relation, where \(S_0\) is the spot index level, \(r\) the financing rate, \(q\) the index dividend yield, and \(T\) the time to expiry:
$$F_0 = S_0\, e^{(r-q)T}$$
Normally both arbitrage arms operate, and the futures price stays pinned close to this fair value in both directions. After 2015, only the upper arm (buying the future, when it is too cheap relative to a long spot position already held) had any real force, and only for participants who already held the shares and needed no fresh short position. For everyone else, the correcting trade needed a short sale of the index basket that the market would not allow at scale, and it needed unrestricted futures access that CFFEX had just capped. With one arm effectively closed, the no-arbitrage relation stopped being an equation that must hold and became an inequality that could drift, sometimes for years, especially in the CSI 500 contract, whose small-cap constituents were the hardest to borrow and whose hedging demand from quantitative funds was the most concentrated of the three CFFEX products.
The cost of this gap fell on a specific party. Anyone short the future to hedge a real equity book was not collecting a free lunch from the discount; they were paying it, every time they sold the future below fair value and later bought it back as the contract converged towards spot at expiry. A persistent discount of this kind is a recurring cost of hedging in a constrained market, not a mispricing that an ordinary trader can safely arbitrage away.
| Date | Event |
|---|---|
| 16 Apr 2010 | CFFEX launches CSI 300 index futures (code IF), China's first onshore stock index futures contract. |
| 16 Apr 2015 | CFFEX launches CSI 500 (code IC) and SSE 50 (code IH) index futures on the same day, giving the market three stock index futures contracts. |
| Jun 2015 | The Chinese equity bull run peaks; the CSI 300 has roughly doubled since November 2014. A sharp sell-off begins the same month. |
| 24 Aug 2015 | "China's Black Monday": the Shanghai Composite falls about 8.5%, its steepest one-day fall since 2007. |
| 25 Aug-2 Sep 2015 | CFFEX raises non-hedging margin from 10% to 40% (hedging margin 10% to 20%), cuts the daily position-opening limit to 10 contracts, and raises the intraday closing fee roughly a hundredfold. |
| 7 Sep 2015 | Tightened rules take full effect; opening more than 10 contracts a day in one product is treated as abnormal trading. |
| 8-9 Sep 2015 | Bloomberg reports CSI 300 and CSI 500 futures volumes have fallen about 99% from their June 2015 peak. |
| 16 Feb 2017 | CFFEX begins the first round of easing: margins and position limits relaxed for the first time since 2015. |
| Sep 2017 | A second round of easing lowers margin on CSI 300 and SSE 50 from 20% to 15%. |
| 3 Dec 2018 | A third round: CSI 300/SSE 50 margin cut 15% to 10%, CSI 500 margin halved to 15%, intraday non-hedging limit raised 20 to 50 contracts, closing fee cut 0.069% to 0.046%. |
| 22 Apr 2019 | CFFEX further cuts the CSI 500 futures margin, moving towards pre-2015 levels. |
The mathematics
The question this section answers is what a persistent discount actually costs a hedger who has no choice but to keep rolling the position. The course's own worked figure for this case is that a 0.5% quarterly discount costs about 2% a year of hedge performance for a short hedger who rolls every quarter.
$$\text{Simple annual cost} = 0.5\% \times 4 = 2.00\%$$
$$\text{Compounded annual cost} = (1 + 0.005)^4 - 1 \approx 2.015\%$$
Compounding barely changes the answer at this size, so "about 2% a year" holds under either convention. The same mechanism, pushed to a more stressed, explicitly illustrative level, four times the base case, shows how the cost scales; this is not a sourced market quote. A 2%-per-quarter discount compounds to:
$$(1 + 0.02)^4 - 1 \approx 8.24\% \text{ a year}$$
Translating the base case into cash uses CFFEX's own CSI 500 (IC) contract multiplier of RMB 200 per index point, and an illustrative index level of 6,000:
$$\text{Discount in points} = 6{,}000 \times 0.5\% = 30 \text{ points}$$
$$\text{Cash cost per contract, per roll} = 30 \times 200 = \text{RMB } 6{,}000$$
quarterly_discount = 0.005 # 0.5% per quarter, base-case figure
simple_annual = quarterly_discount * 4
compounded_annual = (1 + quarterly_discount)**4 - 1
print(f"simple annual cost -> {simple_annual:.2%}")
print(f"compounded annual cost -> {compounded_annual:.3%}")
stressed_quarterly = 0.02 # 2% per quarter, illustrative only
stressed_annual = (1 + stressed_quarterly)**4 - 1
print(f"stressed annual cost -> {stressed_annual:.3%}")
multiplier = 200 # RMB per index point, CFFEX CSI 500 (IC) contract
index_level = 6000 # illustrative round index level
points = index_level * quarterly_discount
cash_cost = points * multiplier
print(f"discount in points -> {points:.1f} index points")
print(f"cash cost per contract -> RMB {cash_cost:,.0f} per quarterly roll")Output
simple annual cost -> 2.00%
compounded annual cost -> 2.015%
stressed annual cost -> 8.243%
discount in points -> 30.0 index points
cash cost per contract -> RMB 6,000 per quarterly rollA note on where these numbers come from. The 0.5% quarterly figure and the "about 2% a year" result are the course's own labelled illustrative arithmetic, not a market quote for any specific date. No source behind this case gives one precise, dated "the discount was X% on date Y" figure: the academic literature confirms the discount was persistent and largest in the CSI 500 contract, but as a statistical finding across many observations, not a single headline percentage. The stressed 2%-per-quarter scenario is explicitly stylised too, included only to show how the same mechanism scales at a more severe discount level.
Data and facts
| Quantity | Value | Source |
|---|---|---|
| CSI 300 futures launch | 16 April 2010 | CFFEX, official Milestones page |
| CSI 500 & SSE 50 futures launch | 16 April 2015 (same day) | CFFEX, official Milestones page |
| Shanghai Composite fall, 24 Aug 2015 | ≈8.5% (largest since 2007) | Cross-checked secondary summaries of the 2015-16 turbulence |
| Non-hedging margin, Aug-Sep 2015 | 10% → 40% | Review of International Political Economy (2020) |
| Hedging margin, Aug-Sep 2015 | 10% → 20% | Review of International Political Economy (2020) |
| Daily position-opening limit, non-hedging | 500 → 10 contracts | Review of International Political Economy (2020) |
| Intraday closing fee, Aug-Sep 2015 | ≈0.0115% → 0.23% | Review of International Political Economy (2020) |
| CSI 300 & CSI 500 futures volume fall from Jun 2015 peak | ≈99%, by 8-9 Sep 2015 | Bloomberg News, 8 Sep 2015 |
| CSI 500 (IC) contract multiplier | RMB 200 per index point | CFFEX contract specification; Barchart |
| First easing round | 16 February 2017 | FIA, MarketVoice |
| Third easing round (margins, limits, fees) | 3 December 2018 | Yicai Global |
| Contract with the largest documented discount | CSI 500 (small-cap) | China Finance Review International working paper |
The lesson
- A one-sided arbitrage leaves a priced gap, not free money. When only one of the two arbitrage arms that should enforce a no-arbitrage price is available, a persistent mispricing can survive. It is not an anomaly waiting to be exploited, because the trade that would close it is exactly the one that is restricted.
- Margin, position limits and fees are themselves a market design choice, and they show up in prices. CFFEX's 2015 rules aimed to curb destabilising speculation, but a side effect was to make the futures market's own price-discovery mechanism, cash-and-carry arbitrage, partly unworkable for years.
- A discount is a recurring cost paid at every roll, not a one-off loss. A short hedger using index futures to protect a real equity book pays the discount each time the contract is rolled, and this compounds into a real, ongoing drag on hedge performance.
- The contract that is hardest to short, or hardest to replicate, carries the largest gap. The CSI 500's small-cap, harder-to-borrow constituents made its futures discount the largest of the three CFFEX contracts, showing that basis anomalies concentrate where the correcting arbitrage is weakest.
- Restrictions imposed in a crisis can outlast the crisis by years. The 2015 rules were emergency measures, but full easing took until 2019, roughly four years, a reminder that market-design interventions are far easier to impose quickly than to unwind.
Where it appears in the course
Think about it
- If mainland A-share short-selling had been unrestricted throughout, would the CSI 500 discount have appeared at all? Which of the two arbitrage arms would have done the correcting?
- A fund manager needs to hedge a real equity book and cannot avoid using index futures. Is paying a known, persistent discount at every roll a bad hedge, or simply the honest price of hedging in a constrained market?
- CFFEX took about four years to fully unwind its 2015 restrictions. What would you want to know, as a regulator, before deciding whether a crisis-era trading restriction should be lifted?
Sources
- CFFEX (China Financial Futures Exchange), official "Milestones" page (launch dates for CSI 300, CSI 500, and SSE 50 index futures). cffex.com.cn
- CFFEX, CSI 500 Index Futures contract page and specifications (contract code IC). cffex.com.cn
- Bloomberg News, "China Just Killed the World's Biggest Stock-Index Futures Market", 8 September 2015. bloomberg.com
- Review of International Political Economy, "Financialization with Chinese characteristics? Exchanges, control and capital markets in authoritarian capitalism", 2020 (details the August-September 2015 CFFEX margin, position-limit and fee changes). tandfonline.com
- Yicai Global, "China Eases Restrictions on Stock Index Futures Trading", 3 December 2018. yicaiglobal.com
- FIA (Futures Industry Association), "2019 shaping up to be a big year for Chinese equity index futures", MarketVoice. fia.org
- Barchart, "CFFEX CSI 500 Index Futures Contract Specifications" (contract multiplier, RMB 200 per index point). barchart.com
- "Chinese investor sentiment and equity index futures basis", China Finance Review International (working paper hosted by Shanghai Jiao Tong University's Antai College of Economics and Management). acem.sjtu.edu.cn
- ISDA (International Swaps and Derivatives Association), "Developing Safe, Robust, and Efficient Derivatives Markets in China". isda.org
