China's 2015 Stock Market Crash: From Margin Boom to Black Monday

Key Takeaways

• The trigger had a date. On 12 June 2015, the same day the CSI 300 closed at its cycle peak, the China Securities Regulatory Commission released draft rules tightening oversight of unregulated shadow-financed margin trading. The market fell on every one of the next four weeks.
• The first leg was fast and large. The CSI 300 Index lost almost a third of its value between 12 June and 8 July 2015, and the Shenzhen ChiNext board, dominated by small technology companies, lost 40 percent over the same three and a half weeks.
• Leverage without a ceiling did more damage than leverage with one. In an account-level study of the crash, shadow-financed margin accounts averaged 6.62 times leverage against 1.43 times for CSRC-regulated brokerage accounts, because only the brokerage system had a market-wide maximum leverage rule, the Pingcang Line.
• The rescue worked, briefly. Rate cuts, a brokerage-funded buying pool, an IPO freeze and a selling ban on large shareholders helped the market rebound 5.8 percent on 9 July. Eighteen days later, on 27 July, the CSI 300 fell 8.5 percent in a single session, its worst day since 2007, once support was scaled back.
• A second, separate shock followed in August. The People's Bank of China's 11 August change to how it sets the yuan's daily reference rate sent the currency down 2.8 percent against the dollar in two days and reopened the equity selloff, culminating in a 24 August global selloff now generally called Black Monday.
• The VIX confirmed how unusual the spillover was. It closed at 40.74 on 24 August 2015, its highest close since 4 October 2011, while the S&P 500 fell 4 percent for the day after an intraday drop of 6 percent.
• The damage stayed mostly inside China. Foreign investors owned only about 1.5 percent of Chinese shares in mid-2015, and equities made up less than 15 percent of Chinese household financial assets, which is the main reason a stock crash of this size did not become a banking crisis.

What Was the 2015 China Stock Market Crash?

Between mid-2014 and 12 June 2015, mainland Chinese equities went through one of the fastest re-ratings any major stock market has ever recorded. By the government's own commission on the relationship, the Shanghai market was up 149 percent year over year and the Shenzhen market up 190 percent year over year as of the start of June 2015, days before the peak. The price-to-earnings ratio on the benchmark CSI 300 Index, which tracks the 300 largest listed companies across both exchanges, rose from about 10 in mid-2014 to about 21 by June 2015. On the Shenzhen ChiNext board, a Nasdaq-style venue for smaller technology companies, the price-to-earnings ratio peaked near 143.

That run-up ended on 12 June 2015, a Friday. The CSI 300 and the Shanghai Composite both closed at their cycle highs that day. The following Monday, 15 June, the market began falling, and it kept falling almost without interruption through 8 July, by which point the CSI 300 had lost close to a third of its value and the ChiNext board had lost 40 percent. Chinese investors, overwhelmingly retail and overwhelmingly using borrowed money, lost roughly $3.5 trillion in the five weeks after the peak, a sum the U.S.-China Economic and Security Review Commission noted was equal to China's entire stock market capitalization in 2012.

What followed was not one crisis but three distinct episodes across ten weeks, each with its own trigger. The first was a margin-driven unwind from 12 June to roughly 8 July, met by an escalating series of state interventions. The second was a renewed 8.5 percent one-day drop on 27 July once some of that support was pulled back. The third, and the one that reached the rest of the world, began on 11 August when the PBoC changed how it sets the yuan's daily trading band, a technical reform that culminated in the global selloff of 24 August now known as Black Monday. Collapsing these into one event obscures that each had a different transmission channel.

What Set Off the Crash on June 12, 2015?

The proximate trigger has a specific date attached to it, which is unusual for a crash this size. On 12 June 2015, the China Securities Regulatory Commission released a set of draft rules that would tighten oversight of shadow-financed margin trading, the large and largely unregulated pool of borrowed money that fintech lending platforms had extended to retail stock investors outside the normal brokerage system. Researchers who later obtained account-level trading data for both regulated and shadow-financed margin accounts found that a month-long crash began on the very next trading day, wiping out almost 40 percent of the market index by the time it was over.

The mechanism is not mysterious once the account data is examined. Investors using shadow-financed leverage were, on average, holding roughly 6.6 times their equity in assets, and many of those accounts had no CSRC-mandated maximum leverage limit, only whatever ceiling their individual lender had set. When the CSRC signaled it intended to restrict new shadow accounts, lenders pulled back credit and investors already close to their negotiated limits started selling to deleverage before anyone forced them to. That selling pushed prices down, which pushed remaining leveraged accounts closer to their limits, which produced more selling, the same feedback loop leverage and fire-sale models describe, confirmed here with real account-level data rather than inference.

It is worth being precise about what the CSRC actually did, because the popular shorthand overstates it. The 12 June rules did not ban existing margin positions or force anyone to sell; they restricted new shadow-financed accounts and tightened paperwork on existing ones. That a rule this narrow could set off a nearly one-third repricing of the benchmark index says less about the rule than about how much leverage had accumulated by the time it was written, a distinction our guide to margin requirements and how a maximum leverage rule works covers more generally.

There had also been an earlier tremor that the market shrugged off. On 26 May 2015, the Shanghai Composite fell 6.5 percent in a single session, a drop attributed at the time to Central Huijin, a subsidiary of China's sovereign wealth fund, selling down stakes in Industrial and Commercial Bank of China and China Construction Bank, combined with the central bank draining liquidity from commercial banks through repurchase agreements and brokerages tightening their own margin credit terms. The market recovered within days, rising 4.7 percent on 1 June, and continued climbing into the 12 June peak. In hindsight the 26 May episode looks like the first sign that the structure was fragile; at the time, it looked like a buying opportunity, and for six more trading days it was one.

How Big Was the Run-Up That Made the Crash Possible?

A crash this size needs a run-up this size, and the numbers behind the run-up explain why the CSRC's narrow rule change had such outsized consequences. Three separate metrics moved together in the year before the peak: valuations, participation and leverage.

Valuations more than doubled by the standard measure. The CSI 300's price-to-earnings ratio rose from about 10 in mid-2014 to about 21 in June 2015. On the ChiNext board, where many companies had little or no earnings to speak of, the price-to-earnings ratio reached roughly 143, a level that leaves almost no room for anything to go wrong in the underlying businesses.

Participation surged alongside price. More than 56 million new brokerage trading accounts were opened in the first half of 2015 alone, the overwhelming majority by individual retail investors rather than institutions. Combined daily turnover on the Shanghai and Shenzhen exchanges averaged roughly 1.8 trillion yuan, about $300 billion, in the month leading up to the 12 June peak, about six times the 2014 average and, for that stretch, larger than turnover on the U.S. stock market. According to the Shanghai Stock Exchange's own 2015 annual statistics, retail investors accounted for about 85 percent of total trading volume on the exchange that year, a share far higher than in developed markets where institutions dominate turnover.

Leverage grew fastest of all. Broker-intermediated margin trading balances reached roughly 2.2 trillion yuan, about $360 billion, in early June 2015, an almost sixfold increase from a year earlier and equal to about 8 percent of the market's tradable capitalization. That figure counts only the regulated brokerage system; the shadow-financed system operating through fintech platforms, structured products and informal lenders sat outside any single reporting regime and is harder to size precisely, which is itself a warning sign discussed further below.

None of this happened by accident. The People's Bank of China had been easing policy through 2014 and into 2015, and state media, including the Communist Party's People's Daily, ran editorials during the run-up describing rising stock prices as a sign of economic strength. The Shanghai-Hong Kong Stock Connect program, launched in late 2014, opened a new channel for cross-border flows and helped narrow, though not close, a persistent premium at which mainland-listed shares traded over identical shares of the same companies listed in Hong Kong. Investors had reason to believe policy wanted stocks to go up, and for eleven months it did.

How Far and How Fast Did Prices Fall?

The decline did not happen in a single move. It happened in three distinct legs across ten weeks, each with a different immediate cause, which is easy to lose in a summary that only reports the total decline.

Reading the table in order matters more than reading the total. The first leg was almost entirely a domestic margin story: leveraged retail accounts deleveraging into a market with too few buyers. The second leg, the 27 July drop, happened after roughly two weeks of apparent stabilization, once investors concluded the extraordinary support measures described below were being wound down rather than sustained. The third leg had a different trigger entirely, a currency mechanism change that was, on its own terms, a modest technical reform tied to China's ambitions for the yuan's inclusion in the International Monetary Fund's Special Drawing Rights basket. Collapsing all three into "the China crash" makes the episode sound like one continuous panic; it was closer to three separate tests of the same underlying fragility.

Why Did Shadow-Financed Margin Accounts Do More Damage Than Regulated Ones?

China ran two parallel margin-lending systems into the 2015 peak, and the difference between them is the clearest lesson the episode has to offer about how leverage actually breaks a market.

The brokerage-financed system is the one most coverage of the crash focuses on, because it was larger in total assets and more visible to regulators. But researchers who obtained account-level data covering both systems for May through July 2015 found that the shadow-financed system, smaller and largely invisible to the CSRC, contributed disproportionately to the crash's selling pressure. Brokerage accounts, even though they were far more numerous, mostly stayed well below their regulated leverage ceiling throughout the episode. Shadow accounts, holding more than four times as much leverage on average, produced selling that tracked far more closely with the market's actual price declines.

The behavioral finding underneath that comparison is what makes it useful beyond this one episode. Investors did not wait until a broker forcibly liquidated their account to start selling. Selling intensity rose sharply as an account's leverage approached its own maximum, whether that maximum was the CSRC's uniform Pingcang Line or an informal limit set by a shadow lender, and the relationship was two to three times stronger on days the market was already falling than on days it was rising. That is the leverage spiral described in academic models of fire sales, observed directly in trading records rather than inferred from prices: falling prices push leveraged accounts closer to their limits, which produces anticipatory selling, which pushes prices down further.

There is a counterintuitive detail worth keeping. Shadow accounts had higher leverage ceilings than brokerage accounts, not lower ones, which is the opposite of what "unregulated lending is riskier" usually implies about individual loan terms. What made the shadow system more dangerous to the market as a whole was not that each account was closer to a hard stop; it was that there was no uniform, transparent ceiling at all, so neither regulators nor other investors could see how much stress was building until the selling had already started. A visible constraint that everyone can see coming is safer for a market than an invisible one, even when the visible constraint is, on paper, stricter.

What Did Beijing Do to Try to Stop the Rout?

The Chinese government's response, compressed into a little more than two weeks in late June and early July 2015, is one of the largest and fastest peacetime market-support operations any government has attempted. The sequence is worth setting out in order, because the pace itself is part of the story: authorities moved from a conventional monetary easing to direct state purchases of equities in under two weeks.

Two of these measures show how far the response went beyond a normal playbook. China Securities Finance Corporation had existed since 2011 as a conventional lender of last resort to brokerages, plumbing rather than a market participant; directing central bank funding through it to buy equities directly turned it, for a period, into a vehicle for the state to hold stocks on its own balance sheet, widely described at the time as the emergence of a so-called national team of state-directed buyers. The pension measure worked differently: it did not inject new money so much as relax a rule keeping a large, stable pool of capital out of the market.

It is tempting to conclude from the eventual global spillover that Chinese authorities did too little, too late. The record does not support that. Within two weeks Beijing cut rates and reserve requirements simultaneously for the first time since the global financial crisis, opened a new source of demand through pension investment, froze new equity issuance, banned the largest shareholders from selling, and stood up a direct state buying operation. Few governments have moved with more force. What the episode shows is a limit on what even forceful intervention can do once a leverage-driven repricing is under way, a distinction our explanation of how correlations and support mechanisms behave during crises covers more generally.

Did Suspending Half the Market Help or Make Things Worse?

Alongside the official interventions, individual companies began voluntarily suspending trading in their own shares to avoid further declines, a step Chinese exchange rules allow far more liberally than most developed markets. The scale reached by early July 2015 has no real precedent. By 9 July, 1,476 stocks, more than half of every company listed on the Shanghai and Shenzhen exchanges combined, had stopped trading. The ChiNext board was hit hardest: as of 10 July, only 205 of its 484 listed companies were still trading, with the rest suspended.

The stated purpose of a suspension is to give a company breathing room from a disorderly selloff in its own stock. The account-level research into the crash suggests the actual effect, at the scale China reached in July 2015, ran the other way. Chinese trading rules also impose a 10 percent daily price-move limit on individual stocks; once a stock hits that limit or is voluntarily suspended, an investor holding it cannot sell it no matter how urgently leverage elsewhere in the portfolio needs reducing. Researchers found that investors facing this constraint significantly intensified their selling of whatever else in the portfolio was still tradable, to hit deleveraging targets their frozen positions could not reach. A halt on one stock became extra selling pressure on every other stock the same investor held.

This is a mechanical consequence of forced deleveraging, not investor psychology. A margin account facing a leverage limit does not need to sell any particular stock; it needs to raise a particular amount of cash. If half the portfolio cannot be touched, the required selling concentrates on the half that can, precisely the securities least protected from the panic the suspension was meant to contain. Our guide to circuit breakers and trading halts covers the broader tradeoffs of how exchanges use halts.

The suspended stocks did not stay frozen forever. Most of the shares halted in the first week of July had resumed trading by mid-August, though the CSRC's investigation into abuse of the suspension mechanism, including companies that cited reasons for a halt that later proved unrelated to any genuine corporate need, continued for years afterward.

Why Did the PBoC Change How It Prices the Yuan on August 11?

The third leg of the episode had almost nothing to do with the margin dynamics that drove the first two. On 11 August 2015, the People's Bank of China announced that it would continue letting the yuan trade against the dollar within a daily band of plus or minus 2 percent, unchanged, but that the central parity rate around which that band is set would now be determined by the previous day's closing market rate rather than by a rate the central bank chose administratively. Officially, this was described, accurately, as a step toward a more market-determined exchange rate mechanism, and it was explicitly connected to China's push for the yuan's inclusion in the International Monetary Fund's Special Drawing Rights basket, then under review for later in 2015.

The market read it differently. Because the yuan had been trading consistently toward the weak end of its permitted band through the first half of 2015, tying the new central parity to the previous day's market close meant the fixing itself would move weaker as soon as the new mechanism took effect. Confirmed against exchange rate data, the yuan fell 2.82 percent against the dollar over the two trading days following the announcement, from 6.2094 yuan per dollar on 10 August to 6.3845 on 12 August, before the PBoC intervened to slow the move and the rate stabilized in the low 6.30s through the rest of the month. A 2.8 percent move is small by the standards of a currency that floats freely, but the yuan had not moved anywhere near that much in two days at any point in the preceding several years, and investors interpreted the shift as a signal that Chinese policymakers were more worried about growth than they had been letting on.

The reaction moved beyond currency markets almost immediately. Other Asian and emerging-market currencies came under fresh pressure as investors reassessed regional growth prospects and the odds of competitive devaluations; the Malaysian ringgit, for instance, depreciated more than 6 percent against the dollar in the weeks that followed. Offshore yuan funding costs in Hong Kong tightened sharply as speculation against the currency intensified, evidence of both bets on further depreciation and the PBoC's own defense of the currency. Chinese equities, drifting sideways after the 27 July drop, resumed falling as the move fed a broader reassessment of China's growth outlook, a dynamic our guide to how the dollar and currency moves transmit across asset classes covers more generally.

What Happened on China's Global Black Monday, August 24, 2015?

Between 18 and 25 August 2015, Chinese equities fell a further 21 percent, and the selling accelerated into a single, globally synchronized session on Monday 24 August, the day now generally referred to as Black Monday 2015. The CSI 300 fell almost 9 percent that day, its worst single session since the 27 July drop a month earlier. The move did not stay contained to China.

In the United States, the S&P 500 opened sharply lower and fell as much as 6 percent intraday before recovering somewhat to close down 4 percent for the day. The opening minutes were disorderly enough that some of the largest, most liquid blue-chip stocks in the index, including General Electric and JPMorgan Chase, traded down more than 20 percent intraday before recovering, a dislocation more typical of a market-structure breakdown than of ordinary selling. Implied volatility spiked to match: the VIX closed at 40.74 that day, confirmed against the full historical series back to 2011, its highest close since 4 October 2011, nearly four years earlier, and volatility in commodity, bond and foreign exchange markets rose sharply too.

The scale of the global move is easiest to see against the size of China's direct exposure to the rest of the world's investors, which was small. A large domestic shock producing an outsized global reaction suggests markets were not pricing a direct financial-contagion channel so much as revising their view of global growth, given China's size in the world economy, and their confidence in a Chinese policymaking apparatus that had just spent two months demonstrating the limits of its control over its own equity market. Fear moved faster and further than any plausible balance-sheet link did, a pattern that also shows up in our study of the 1997 Asian financial crisis, though the underlying vulnerability there, fixed exchange rates and unhedged dollar debt, was structurally different from anything present in China in 2015.

The PBoC responded the following day. On 25 August 2015, it cut the benchmark one-year lending rate by 25 basis points and cut reserve requirement ratios by 50 basis points for financial institutions generally, with a larger 300 basis point cut for financial-leasing and auto-financing companies specifically, its fifth broad easing move of the year and a second simultaneous rate-and-reserve-requirement cut inside two months.

How Far Did the Shock Spread Beyond China?

The August spillover is worth measuring on its own terms, separate from the June and July margin unwind, because it is the part of the episode most relevant to a reader who holds no Chinese assets at all.

Commodity markets, which had briefly stabilized earlier in 2015, resumed a downtrend led by oil, as investors weighed weaker Chinese demand alongside already ample supply. Currencies of commodity-exporting economies, from Australia and Canada to Russia and Brazil, came under renewed pressure at the same time the dollar kept strengthening on U.S. rate expectations, a double squeeze hitting producers from both the demand and financing side at once. A number of central banks outside China, including Hungary, India, Korea, Russia and Thailand, eased policy around the same period.

Government bond markets moved the opposite direction from equities, as is typical in a broad risk-off episode: yields on safe developed-market debt eased lower on safety-seeking demand, though the moves were modest next to the swings in equities and currencies, since yields in the United States, Germany and Japan were already historically low heading into August 2015.

One detail is easy to miss: the Federal Reserve's own rate-hike expectations moved with the turbulence. Pricing derived from federal funds futures showed the implied probability of a rate increase by the Fed's September 2015 meeting falling from around 80 percent at the start of the year to roughly 32 percent by early September, an example of how a shock originating entirely inside China's domestic equity market fed back into expectations for the world's largest central bank within weeks.

Why Didn't a $3.5 Trillion Stock Crash Become a Chinese Banking Crisis?

A loss on the scale reported for the first leg alone, roughly $3.5 trillion of market value in five weeks, is the kind of number that in other contexts has triggered systemic banking crises. It did not do so here, and the reasons are structural rather than a matter of the intervention working better than it looks.

The first reason is who owned the shares. Foreign investors held only about 1.5 percent of Chinese equities in mid-2015, a consequence of capital controls and quota restrictions that had, until that point, mostly been criticized as an obstacle to China's integration into global capital markets. In this episode, that same isolation functioned as a firewall: there was no large pool of foreign-bank balance sheets directly exposed to Chinese share prices the way, for instance, European and American banks were exposed to U.S. mortgage securities in 2007 and 2008.

The second reason is how much of Chinese household wealth actually sat in stocks. Equities made up less than 15 percent of Chinese household financial assets at the time, with the much larger share held in bank deposits and real estate. A decline of even 30 percent in a category that represents a minority of household wealth produces a real but bounded hit to consumption, not the kind of broad wealth-effect shock that follows a comparable decline in a market where households are more heavily invested in equities.

The third reason is that the crash, while enormous relative to the stock market itself, was small relative to the Chinese banking system. Total margin debt at its peak, brokerage and shadow-financed systems combined, was a modest fraction of the assets held by China's major banks, so even a near-total loss on leveraged equity positions did not threaten institutions large enough to matter for the broader financial system. Some analysts at the time also argued the financial sector's own contribution to headline GDP growth had been inflated by the trading boom itself, since brokerage activity counts directly in output figures, meaning part of the crash's economic effect was simply the reversal of a boost the rally had manufactured.

None of this means the crash was costless. It means the specific channels that turn an asset-price collapse into a systemic financial crisis, cross-border bank exposure, a heavily-invested household sector, and margin debt large relative to the banking system, were each individually limited here. Readers evaluating exposure to any market should ask the same three questions before assuming a shock will spread: who owns the asset, how large is it relative to total wealth, and how large is the leverage behind it relative to the institutions that would absorb a default. Our guide to international investing covers how ownership structure and capital controls shape which shocks cross borders.

Who Actually Lost Money?

The losses concentrated overwhelmingly among the group that had driven the rally in the first place: individual retail investors, many of them new to the market and many of them trading on borrowed money. With roughly 85 percent of Shanghai Stock Exchange trading volume coming from retail accounts through 2015, and 56 million new trading accounts opened in the first half of the year alone, the losses were not spread across a diversified institutional base the way a comparable decline in a developed market often is. A large share of the people who lost money in June and July 2015 had opened their first brokerage account within the preceding several months.

Within that retail base, leverage decided who lost the most. Stocks disproportionately held by margin accounts close to their maximum leverage limits, whether the regulated Pingcang Line or an informal shadow-lending ceiling, underperformed stocks with little such exposure by about 5 percentage points within 10 to 15 trading days of the worst selling, a gap that narrowed back toward zero over the following 30 to 40 trading days as forced selling subsided. An unlevered investor holding the same portfolio through the same crash therefore lost meaningfully less, purely from the difference in forced selling, independent of any skill in stock selection.

A smaller, specific group of losers emerged outside China entirely: U.S.-listed Chinese technology companies that had begun delisting from American exchanges to relist on mainland markets, hoping to capture the far higher valuations paid there. When the CSRC suspended all new initial public offerings on 4 July 2015, these companies found themselves stranded mid-transition, no longer fully valued on the exchange they were leaving and unable to complete the listing they were headed toward, an overlooked casualty of a policy aimed at an entirely different problem.

Foreign portfolio investors with direct exposure to Chinese A-shares, a small group given the 1.5 percent foreign ownership share, took losses proportional to their holdings but represented a small fraction of the total damage. The larger foreign-investor loss came indirectly, through the 24 August global selloff, where an investor holding no Chinese assets at all could still have taken a meaningful one-day loss from the S&P 500's 4 percent decline, since the channel of harm for most non-Chinese investors was sentiment and volatility contagion rather than any direct balance-sheet link.

Which Warning Signs Were Visible Before June 2015, and Which Only in Hindsight?

This distinction matters because the 2015 China crash is often described, after the fact, as an obviously unsustainable bubble that any careful observer should have called. The record supports part of that claim and not the rest.

Visible in advance to anyone reading the published data

• Valuations had detached from earnings on any reasonable measure. A price-to-earnings ratio near 143 on the ChiNext board, and a doubling of the CSI 300's own ratio in a single year, were public, daily-updated numbers available to every market participant, not information that required special access.
• Leverage had grown far faster than the market itself. A near-sixfold year-over-year increase in broker-financed margin balances, reaching about 8 percent of tradable market capitalization, was reported and visible well before the peak, even if the shadow-financed system sitting alongside it was harder to size.
• Participation had a first-time-investor character. More than 56 million new trading accounts opened in six months is a level of retail entry that, on its own, has historically coincided with late-stage rallies rather than early ones.
• An independent index provider had just registered a formal objection. On 9 June 2015, three trading days before the peak, MSCI announced it would not yet include mainland China A-shares in its Emerging Markets Index, citing unresolved concerns about foreign investor quota allocation, restrictions on capital mobility, and unclear ownership rules for shares purchased through the Stock Connect program, a public, dated judgment from a major index provider that the market's structure had unresolved problems.

Only clear in hindsight

• That a narrow, technical CSRC rule on shadow accounts would be the specific trigger. Nothing about the 12 June draft rules signaled they would be more consequential than the many prior regulatory statements the market had absorbed without much reaction.
• That the second leg would come from withdrawing support, not from a fresh shock. The 27 July drop followed roughly two weeks in which the state-backed rescue appeared to be working; there was no new piece of adverse information that day, only a market discovering that the buying support it had been relying on was being scaled back.
• That a currency-mechanism reform aimed at an IMF technicality would become the larger global story. The 11 August change was explicitly framed around the yuan's Special Drawing Rights candidacy, a subject of interest mainly to central bankers and reserve managers, not equity markets; its equity-market and global-volatility consequences were not the stated purpose of the reform.
• That the VIX would reach its highest level since 2011 over an event with almost no direct foreign financial exposure. The size of the global reaction, given how contained China's ownership structure made the direct financial contagion channel, was itself a surprise, and points to sentiment and growth-expectation channels mattering more than balance-sheet linkages in this specific episode.

The practical lesson is the same one that recurs across leverage-driven episodes: identifying that a structure is fragile and predicting the exact date and shape of its unwind are different skills, and an investor can be right about the first while being unable to act usefully on the second, a distinction our guide to stress testing and scenario analysis treats as a starting principle rather than a footnote.

What Happened When Beijing Tried the Same Playbook Again in January 2016?

The summer of 2015 was not the end of the story, and what happened next is a useful check on how much the intervention had actually fixed. On 4 January 2016, the first trading day of the year, Chinese regulators introduced a new market-wide circuit breaker, built around the CSI 300 Index: a 5 percent move would pause trading for 15 minutes, and a 7 percent move would halt trading entirely for the rest of the day. It was designed explicitly to prevent a repeat of the disorderly declines of the previous summer.

It failed within four trading days. On its first day in effect the market fell fast enough to trigger both thresholds, closing trading about 80 minutes early. On 7 January, the second trading day it was live, the Shanghai Stock Exchange closed for the day only about 30 minutes after the opening bell, the shortest trading day in the exchange's history, after the PBoC set the yuan's daily reference rate 0.5 percent lower, an echo of the August move. Rather than calming the market, the breaker appears to have accelerated the selling that triggered it: as prices approached the 5 or 7 percent thresholds, investors rushed to sell before trading could be halted and their positions locked in, a dynamic sometimes called a magnet effect. The CSRC suspended the mechanism on 7 January 2016, four days after introducing it, concluding it was doing more harm than good.

The government's response otherwise repeated the summer 2015 playbook in miniature: state-linked buyers resumed direct share purchases, the PBoC injected roughly $20 billion (130 billion yuan) of short-term funding, and the CSRC extended restrictions on large shareholders' sales rather than let the original summer 2015 ban simply expire as scheduled on 8 January. The combined rout across the two January sessions erased more than $1 trillion of Chinese equity value, and worldwide equity markets lost more than $2 trillion in the first trading week of 2016 in sympathy, with the Dow Jones Industrial Average falling nearly 2.3 percent and the S&P 500 falling 2.4 percent on 7 January alone.

The currency side of this shows up in China's foreign exchange reserves. The PBoC's reserves fell by a record $107.9 billion in December 2015 alone, and by $512.7 billion over the full year, the largest annual drop on record at the time, as the central bank sold dollars to slow the yuan's decline rather than let it float freely, an expensive way of managing a currency move gradually rather than all at once.

The clearest lesson from the January 2016 episode is that a mechanism introduced to prevent a specific past failure can introduce a new failure mode of its own, especially when it interacts with the same underlying behavior, in this case a tendency for leveraged and newly-arrived retail investors to sell ahead of an anticipated constraint. Fixing the exposure that caused a crisis is not the same task as fixing the market structure that a crisis exposed, and China's own regulators discovered the difference between those two tasks in real time, four trading days apart.

Which Rules From This Episode Have Since Changed?

Every specific rule described on this page is a snapshot of August 2015, not a description of how Chinese markets work today, and several of the mechanisms at the center of this episode are exactly the kind of rule that changes on a regulator's schedule rather than a reader's. Anyone using this page to reason about current conditions should treat every number below as historical, not current.

The CSI 300 circuit breaker introduced on 4 January 2016 was suspended four days later and, as this page was written, has not been reintroduced in that form; whether a replacement mechanism has since taken its place is a separate question this page does not attempt to answer, and a reader should check current exchange rules directly rather than assume either the original breaker or its absence still holds. The Pingcang Line maximum-leverage system for brokerage margin accounts, and whatever oversight framework now applies to fintech-based lending after the shadow-financed sector's role in this crash drew regulatory attention, are both subject to revision by the CSRC and should not be assumed to match their 2015 description. The PBoC's benchmark lending rate and reserve requirement ratio, cited here at their 27 June and 25 August 2015 levels, are among the most frequently adjusted tools in Chinese monetary policy and have been changed many times since; citing an August 2015 level is a historical data point, not a claim about today's policy setting. The yuan's daily trading band, plus or minus 2 percent around a market-referenced central parity as of the 11 August 2015 reform, is itself a policy parameter the PBoC has adjusted before and could adjust again. The stamp duty rate on Chinese securities transactions, referenced in contemporaneous reporting during the run-up, is likewise a rate authorities have changed multiple times outside this episode's window. None of these figures should be read as current without independent verification against a live, dated source.

Common Myths About the 2015 China Stock Market Crash

"The government's rescue stopped the crash for good." It stopped the first leg. The CSI 300 rebounded 5.8 percent on 9 July 2015 as the rescue measures took hold, but the market fell 8.5 percent in a single session on 27 July, its worst day since 2007, once investors concluded the support was being scaled back. A second and larger shock followed in August, driven by an entirely separate cause.

"The August selloff was just the June crash continuing." The mechanisms were different. The June and July declines were driven by margin-account deleveraging following a CSRC rule change on shadow-financed lending. The August selloff was set off by the PBoC's 11 August change to how it prices the yuan, a currency-mechanism reform with no direct connection to margin trading. They were two separate shocks that compounded rather than one continuous panic.

"A $3.5 trillion loss of this size must have meant a Chinese banking crisis." It didn't produce one, for reasons that are structural rather than a matter of luck. Foreign investors owned only about 1.5 percent of Chinese shares, equities made up less than 15 percent of Chinese household financial assets, and total margin debt was small relative to the balance sheets of China's major banks. The channels that typically turn a stock crash into a banking crisis were each individually limited here.

"Shadow financing was a minor factor next to the much larger regulated margin system." Shadow-financed accounts held less than the brokerage system in total assets, but account-level trading data shows they drove more of the selling pressure behind the crash, because they carried more than four times the average leverage of brokerage accounts and had no CSRC-imposed leverage ceiling.

"Suspending half the market's stocks protected investors from further losses." The evidence from account-level trading data points the other way. Investors who needed to reduce leverage but held stocks that were suspended or locked at their daily price-move limit shifted their selling onto whatever stocks in their portfolio remained tradable, intensifying pressure on the less-protected half of the market rather than reducing pressure overall.

"The January 2016 circuit breaker fixed the flaws the summer had exposed." It was suspended four trading days after it launched. Rather than calming trading, it appears to have encouraged investors to sell earlier, to get ahead of the trading halt it would trigger, an unintended consequence known as a magnet effect that the CSRC itself cited when withdrawing the mechanism.

What a Reader Can Actually Carry Forward

A visible leverage ceiling is safer than an invisible one, even when the invisible one is nominally larger. Shadow-financed accounts averaged 6.62 times leverage against 1.43 times for brokerage accounts, yet it was the shadow system, without a uniform, publicly known maximum, that drove more of the crash's fire-sale selling. A market-wide rule everyone can see coming lets both regulators and other participants anticipate stress before it becomes forced selling; an informally negotiated limit hides that stress until it is already unwinding.

A trading halt on one asset is not lost selling pressure, it is redirected selling pressure. An investor deleveraging a portfolio needs to raise a specific amount of cash, not sell a specific stock. When half the market froze in July 2015, the selling that would have hit those stocks landed instead on whatever remained tradable, which is the opposite of what a suspension is meant to achieve and worth remembering the next time a circuit breaker or halt is proposed as a stabilizing tool.

Support that gets withdrawn produces a second leg down, not a floor. The market did not fall in a straight line through the summer of 2015; it fell, stabilized under heavy state support, and fell again once that support was scaled back on 27 July. A rescue that has not addressed the underlying leverage has only postponed the repricing, not prevented it, and the postponed version can arrive with less warning than the original.

Two shocks landing in the same country in the same season are not automatically one event. The margin unwind of June and July and the currency-mechanism shock of August had different triggers, different transmission channels, and arguably different lessons. Treating "the 2015 China crash" as a single undifferentiated episode makes it harder to identify which specific vulnerability, leverage without a visible ceiling, or a currency regime perceived as opaque, is the one worth watching for in a different market.

Low foreign ownership is a real firewall against global contagion, but it does not make a shock small for the people who actually hold the asset. China's international isolation from its own equity market limited the direct financial-contagion channel to the rest of the world. It did nothing to soften the loss for the tens of millions of Chinese retail investors, many of them new to the market and many of them levered, who absorbed nearly all of it.

A tool that worked once can fail completely the second time it is used. The summer 2015 interventions, however expensive, did eventually stabilize prices. The January 2016 circuit breaker, designed specifically in response to that summer's experience, made its own crisis worse and was withdrawn within four trading days. Mechanism design, not stated intent, determines whether an intervention calms a market or accelerates it.

For other episodes where policy struggled to distinguish a funding shock from a solvency shock, see our studies of the 1997 Asian financial crisis and the 2013 taper tantrum, and for a broader look at how sudden volatility spikes propagate across asset classes, see the 2020 COVID crash. The full set of case studies is indexed on our market history hub.

References

Every figure on this page was verified against the following sources, each retrieved on 26 August 2026:

• Bank for International Settlements: BIS Quarterly Review, September 2015, "Markets roiled as China jolts investors": the CSI 300 and ChiNext declines between 12 June and 8 July, the P/E ratio moves, daily turnover and margin trading figures, the 27 July 8.5% drop, the 11 August yuan mechanism change and its two-day move, the 18 to 25 August declines, the 24 August S&P 500 and CSI 300 moves, the VIX level, and the Malaysian ringgit move.
• U.S.-China Economic and Security Review Commission: China's Stock Market Collapse and Government's Response, July 13, 2015: the $3.5 trillion loss figure, the year-over-year and year-to-date index moves, the 26 May and 1 June moves and their causes, the full chronology and scale of government intervention measures from 24 June through 9 July, the trading suspension counts, the foreign ownership and household asset share figures, and the MSCI index-inclusion decision.
• U.S.-China Economic and Security Review Commission: China's Stock Market Meltdown Shakes the World, Again, January 14, 2016: the January 2016 circuit breaker mechanism, its introduction and suspension dates, the PBoC's yuan fixing move on 7 January 2016, the government's renewed intervention measures, the erased-value figures, the U.S. equity market reaction, and the 2015 foreign exchange reserve decline.
• Bian, He, Shue and Zhou: Leverage-Induced Fire Sales and Stock Market Crashes, working paper: the 12 June 2015 CSRC rule change and its role as the crash's trigger, the structural comparison between brokerage-financed and shadow-financed margin accounts including the Pingcang Line mechanism, the 1.43x and 6.62x average leverage figures, the price-limit and suspension selling-intensification finding, the 5 percentage point fire-sale underperformance figure and its reversion window, and the Shanghai Stock Exchange's own reported 85% retail trading volume share.
• Federal Reserve Bank of St. Louis, retrieved as CSV and recomputed for this page: CBOE Volatility Index, VIXCLS, used to confirm the 24 August 2015 close of 40.74 and that no higher close occurred between 4 October 2011 and 24 August 2015; and China / U.S. Foreign Exchange Rate, DEXCHUS, used to compute the 2.82% two-day yuan move following the 11 August announcement and the roughly 3.3% cumulative move from 10 to 25 August.
• Yale Program on Financial Stability, New Bagehot: China Reserve Requirements, 2015 to 2016: confirmation of the 25 August 2015 reserve requirement ratio cut of 50 basis points for financial institutions generally, with an additional 300 basis point cut for financial-leasing and auto-financing companies.
• Reuters, via Business Standard: China central bank cuts rates, reserve ratio to aid stumbling economy, August 25, 2015: the 25 basis point benchmark lending rate cut to 4.6% announced on 25 August 2015, cited here as secondary reporting corroborating an institutionally-confirmed reserve requirement move; this specific rate figure was not independently fetched from a primary PBoC document this session and is included on the strength of two independent contemporaneous outlets and its arithmetic consistency with the 27 June 2015 rate level reported by the U.S.-China Economic and Security Review Commission above.

Figures deliberately not stated. This page does not give an exact closing level for the Shanghai Composite or CSI 300 Index on any date, including the 12 June 2015 peak, because no primary or institutional source verified here supplied index closing levels precisely enough to publish with confidence; the percentage moves and P/E ratios cited throughout come directly from the BIS and USCC sources above. This page does not give a precise total size for the shadow-financed margin lending system, because that system operated largely outside any single regulator's reporting and estimates found during research were not independently verifiable against a primary source. This page does not give an exact S&P 500 or Dow Jones Industrial Average closing index level for August 2015, because the Federal Reserve Bank of St. Louis's public daily series for those indices does not extend back that far under its current licensing; the percentage moves cited come from the BIS source above. This page does not give a specific figure for offshore yuan interbank funding rates during the August selloff, because no primary or institutional source verified here supplied that data with confidence.

Method note. The two-day and cumulative renminbi moves are calculated directly from the Federal Reserve's DEXCHUS daily series rather than copied from any secondary source, using the closing rate on 10 August 2015 as the pre-announcement baseline. The VIX claim that 24 August 2015 was the highest close since 4 October 2011 was verified by comparing the full daily VIXCLS series from 1 January 2011 through 24 August 2015 and confirming no intervening close exceeded that level. Percentages describing index and market moves in the June to August 2015 window are taken directly from the Bank for International Settlements and U.S.-China Economic and Security Review Commission sources cited above and are not independently recomputed by Swoopr, since the underlying daily Chinese index series were not available through a primary source accessible this session.

This is educational content about a historical episode. It is not investment advice, it is not a forecast, and nothing here is a claim about current Chinese monetary policy, current equity valuations, or any market today. China's reserve requirement ratio, benchmark policy rates, margin regulations including the Pingcang Line, and the yuan's daily trading band are all rules that can and do change; figures describing them here reflect August 2015 only.

Frequently Asked Questions