If you buy shares from someone, there is a gap between agreeing the trade and completing it. In that gap, either side could fail.
The institution that removes that risk is a central counterparty, and it is one of the more consequential pieces of financial infrastructure that almost nobody outside the industry has heard of.
What it actually does
Through a process called novation, the original contract between buyer and seller is replaced by two contracts — buyer with clearing house, clearing house with seller.
Which means neither party is exposed to the other. Both are exposed to the clearing house.
The counterparty risk in the market is concentrated into one institution rather than spread across a web of bilateral exposures nobody can see.
Why concentrating risk is an improvement
This sounds like it should be worse. Concentration is usually bad.
The argument is that a visible, regulated, heavily capitalised concentration point is preferable to an invisible network of bilateral exposures where nobody knows who is exposed to whom.
The 2008 experience with over-the-counter derivatives is the case in point — institutions could not assess their own exposure to a failing counterparty because it ran through chains they could not see.
The regulatory response pushed large categories of derivatives into central clearing precisely for this reason.
How it protects itself
The clearing house does not simply absorb risk. It manages it through a layered structure.
Initial margin is collateral posted against every position, sized to cover expected losses if that position had to be closed out in stressed conditions.
Variation margin moves daily as positions gain and lose value, so losses are settled continuously rather than accumulating.
A default fund, contributed by all clearing members, sits behind the individual margin.
The clearing house's own capital sits in the waterfall too, deliberately positioned so it loses money before the surviving members' contributions are touched.
The default waterfall
The order in which resources are consumed when a member fails is specified in advance, and it is the most important document at any clearing house.
First the defaulting member's own margin. Then its default fund contribution. Then a tranche of the clearing house's capital. Then the surviving members' default fund contributions.
Beyond that, powers to call for additional contributions, and in extremis tools to allocate remaining losses.
Each layer is sized against stress scenarios that are tested regularly and published in summary form.
Where margin calls come from
This is the connection to things people have actually noticed.
When volatility rises, margin requirements rise, because the potential close-out loss on any position grows.
Which means brokers must post more collateral, quickly, at exactly the moment markets are stressed.
A broker unable to meet that call must reduce its positions, restrict its customers, or fail.
The trading restrictions that appeared at several retail brokers during a period of extreme volatility in a small number of shares were, mechanically, this.
The explanations offered at the time were poor, which is a reasonable criticism. The underlying mechanism was not a conspiracy.
The procyclicality problem
The obvious objection is that margin rises when stress rises, which amplifies the stress.
This is a recognised and genuinely difficult problem. Regulators have pushed for margin models that look through short-term volatility spikes, using longer lookback periods and floors.
Which reduces the procyclicality and means margin is higher than strictly necessary in calm periods.
That trade-off has no clean answer. Every approach either undercollateralises in calm or overcollateralises in stress.
Who watches them
Clearing houses that matter are designated as systemically important and supervised accordingly, with requirements on governance, capital, stress testing and recovery planning.
Recovery and resolution planning for clearing houses is an active and unresolved area, because a failing clearing house is a problem with no good precedent.
The honest position among people who work on this is that the arrangements have never been tested by an actual failure of a major clearing house, and nobody is eager to find out.
Why bother knowing this
Because it converts a category of market events from inexplicable to mechanical.
Trading halts, broker restrictions, sudden deleveraging, the specific way stress propagates from a price move to a funding problem — these have a plumbing explanation, and it is usually margin.
Netting and why the numbers shrink
The other function, and it is where most of the efficiency comes from.
Because the clearing house stands between every trade, it can net a member's obligations across all its trades in a security into a single figure.
A member that bought and sold the same instrument many times settles the difference rather than each trade individually.
Which reduces the volume of actual settlement enormously — the netted figure is typically a small fraction of the gross trading volume.
That reduction lowers operational risk, settlement failure rates and the collateral required, and it is the quiet reason the system can handle the volumes it does.
The interoperability question
Where clearing houses compete, arrangements between them allow members of one to trade with members of another.
Which requires the two to hold collateral against each other, creating a link that regulators watch carefully, since it connects the concentration points that were supposed to be separate.