China Market Report 2015 — Leverage, Not Growth

The Chinese equity market rose substantially into mid-2015 and then fell sharply. The episode is frequently explained by reference to Chinese economic growth — a rise on optimism, a fall on emerging concerns.

The evidence supports a simpler explanation. The rise and fall track the expansion and contraction of margin financing, and the mechanics of that cycle are sufficient to explain the price path without reference to economic developments at all.

Margin financing allows an investor to borrow to buy shares, using the shares as collateral. It had expanded very substantially in the period before mid-2015, and the resulting dynamic is well understood:

• Rising prices raise collateral value, permitting more borrowing, which funds more buying, which raises prices.
• The process is self-reinforcing upward, with each increment of price supporting the next increment of leverage.
• Decline reverses it exactly, as falling collateral triggers margin calls, forcing sales, pushing prices lower and triggering further calls.
• The unwind is faster than the build, because margin calls have deadlines and buying does not.

This is a positioning event in the sense the 2018 global report establishes: the size of the move is determined by what was owned rather than by new information. The same structure appears in the 2018 volatility unwind and the 2021 crypto leverage unwind.

Two features made the Chinese case distinctive. Retail participation was unusually high, and retail investors using leverage are less likely to have planned for a margin call or to have the resources to meet one. And the policy response was direct — trading suspensions, restrictions on selling by large holders, and state-directed purchasing — which is a different intervention model from most markets.

Separately and more importantly, the economy was undergoing a genuine structural transition from investment-led to consumption-led growth. That is a slow process operating on a timescale of years, and it had almost nothing to do with the market episode.

Reading the margin cycle

The mechanics are worth setting out in detail because they were observable at the time in publicly disclosed data.

Margin balances are reported. The Shanghai and Shenzhen exchanges published outstanding margin financing balances. An investor watching that series could observe the leverage build directly, without inference.

The build produces a specific vulnerability:

• The buying power is borrowed, so it is not a durable holding. It must be maintained by continued collateral value.
• The collateral is the same asset being bought. This is the critical feature: the loan is secured by the thing the loan purchased, so a price decline impairs the collateral and the position simultaneously.
• Margin calls are mechanical. When collateral falls below a threshold, the position is sold — not by a decision but by a rule.
• The selling is concentrated in time. Many positions hit their thresholds within a narrow range, so the selling arrives at once.

What made the unwind severe:

• The leverage build had been rapid, so a large share of positions had been established at prices near the peak, meaning modest declines triggered calls.
• Retail holders had less capacity to meet calls with additional capital, so positions were liquidated rather than supported.
• Trading suspensions, intended to halt declines, had an unintended effect: with many stocks suspended, investors facing calls had to sell whatever remained tradeable, transmitting pressure to unsuspended stocks.

The market did not fall because the economy weakened. It fell because the buying had been borrowed, and borrowed buying has a deadline.

The general lesson generalises well beyond this market. When assessing a rapid price rise, the question worth asking is: what is funding this? If the answer is leverage, the rise contains the mechanism of its own reversal, and the reversal's timing depends on collateral thresholds rather than on fundamentals.

Why the currency adjustment mattered more than the equity fall

The August 2015 currency adjustment was far smaller in magnitude than the equity decline and considerably more consequential. Understanding why illustrates something general about how markets process information.

What happened. Authorities adjusted the mechanism by which the currency's daily reference rate was set, resulting in a depreciation against the dollar. The move was small in percentage terms.

Why the reaction was large. Markets had operated on an assumption that the authorities would maintain currency stability, treating it as a fixed feature of the landscape rather than as a choice. The adjustment did not disprove that entirely, but it demonstrated that stability was a policy decision — and policy decisions can change.

That prompted reassessment across several dimensions simultaneously:

• Whether the economy was weaker than official figures suggested, if authorities felt the need to support exports.
• Whether further depreciation would follow, which would give holders of domestic assets reason to convert — the capital outflow pressure the 2016 regional report describes as the following year's dominant story.
• Whether other exporting economies would respond in kind, with competitive implications across the region.

The general principle: the significance of an event is not proportional to its magnitude but to how much it changes the distribution of expected outcomes. A small move that invalidates an assumption matters more than a large move that confirms one.

There is a second reading worth noting. As the 2015 global report describes, a currency managed against the dollar imports US monetary policy. With the Fed moving toward tightening and the dollar strengthening, maintaining that link meant importing a tightening that domestic conditions did not warrant. The adjustment was, in part, a response to that pressure — which is the policy trilemma the 2016 regional report sets out, encountered in practice.

The composition shift

Underneath the market episode, a genuine structural transition was underway that operated on a completely different timescale.

The shift. The growth model that had driven the preceding decades relied heavily on investment — infrastructure, property, industrial capacity — and on exports. A transition toward consumption-led growth was underway, driven by rising incomes, an expanding middle class, and policy direction.

Why the transition is slow:

• Investment-led growth produces measurable output quickly. Building infrastructure adds to GDP immediately.
• Consumption-led growth depends on household income and confidence, which change gradually.
• The transition creates excess capacity in industries built for the old model, and that capacity takes years to work off.
• The financial system had been built to fund investment, and redirecting credit toward consumption requires institutional change.

The investment implications were substantial and are frequently confused with the market episode:

• Industries serving investment demand — heavy industry, construction materials, industrial equipment — faced structural rather than cyclical headwinds.
• Industries serving consumption — consumer goods, services, healthcare, entertainment — faced structural tailwinds.
• Commodity demand growth moderated, since investment is far more commodity-intensive than consumption. This is the mechanism behind the 2015 commodity collapse's demand component and is why Australia's position deteriorated per the regional report.
• The composition of imports changed, affecting trading partners differently depending on what they sold.

The key analytical point: this transition was structural, slow and largely independent of the equity market episode. Investors who conflated the two — reading the market fall as evidence about the transition — were combining a leverage event with a decade-long structural process.

The transition was real and mattered enormously for anyone with exposure to Chinese demand. The equity episode was a margin cycle. They occurred in the same year and had almost nothing to do with each other.

What the intervention revealed

The policy response to the equity decline — trading suspensions, restrictions on selling by large shareholders, and state-directed purchasing — is worth analysing on its own terms, because interventions of this kind have consequences that outlast the episode.

The immediate objective was to halt a self-reinforcing decline. Given the margin mechanics described above, that objective is coherent: a forced-selling cascade does not stop on its own until the leverage is exhausted, and stopping it earlier limits the damage.

The instruments had specific side effects:

• Trading suspensions transmitted pressure rather than containing it. With a large share of listed stocks suspended, investors facing margin calls could only sell what remained tradeable. The suspension protected the suspended stocks and concentrated the selling onto the others. A liquidity restriction applied unevenly redirects pressure; it does not absorb it.
• Restrictions on selling by large holders created an overhang. A holder prevented from selling has not stopped wanting to sell. The restriction defers the transaction and creates a known future supply, which weighs on price until it clears.
• State-directed purchasing established an expectation. Once a market participant believes the authorities will support prices, that belief changes risk-taking — the same mechanism the 2019 global report describes for policy asymmetry in developed markets, and with the same dependency on the belief remaining true.

The durable consequence was for foreign investor assessment. An intervention of this kind is a data point about how the market operates, and it entered foreign investors' models as a permanent feature: prices in this market are subject to administrative measures that can suspend trading, restrict exit, or support levels. That is not a judgement about whether the interventions were correct. It is an observation that it changes what an investor is buying.

The practical expression is that foreign investors began pricing an execution risk that had not previously been explicit — the possibility that an intended exit could be prevented or delayed by measures unrelated to the company or to market conditions. That risk is real, it is not diversifiable within the market, and it is one of the reasons the 2023 regional report describes allocators eventually separating this market from the rest of the region.

What an allocator could act on

Watch margin balances. They were published by both exchanges throughout the episode. A leverage build of that magnitude and speed is visible in the data, requires no inference, and identifies the vulnerability before it expresses itself. The most useful single series in this market in 2015 was free and updated regularly.

Ask what is funding a rapid price rise. If the answer is borrowed money secured by the asset being bought, the rise contains the mechanism of its own reversal and the timing depends on collateral thresholds rather than on fundamentals.

Separate the market episode from the economic transition. These were two different things occurring in the same year on entirely different timescales. Positioning for the composition shift — away from industries serving investment demand, toward those serving consumption — was a multi-year structural decision. It was unaffected by the equity episode, and investors who conflated the two either abandoned a sound structural position or read a leverage event as confirmation of an economic view.

Price administrative risk explicitly. After 2015, the possibility of trading suspension or exit restriction is a known feature. It should appear in the model as a discount rather than in a footnote as a caveat, and the size of the discount is a judgement the investor should make rather than avoid.

Follow the commodity channel to its destinations. The composition shift's largest cross-border effect was on commodity demand, since investment is far more commodity-intensive than consumption. Investors with no direct exposure to this market held substantial indirect exposure through Australian resources, through commodity currencies, and through the energy and materials sectors globally. The relevant exposure map was wider than the country allocation suggested.

What 2015 established for China analysis

• The equity episode was a positioning event, observable in disclosed margin data throughout.
• Retail leverage produces faster and more complete unwinds than institutional leverage, because capacity to meet calls is lower.
• Trading suspensions can transmit rather than contain pressure, by forcing selling into whatever remains tradeable.
• The currency adjustment mattered more than its magnitude, because it reclassified an assumption as a choice.
• The composition shift was structural and slow, and conflating it with the market episode was a category error.