Market Infrastructure Makes Liquidity Look Plentiful Until Everyone Needs It at Noon.

The Balance Sheet Is Early to Lunch

Market infrastructure has an agreeable way of making liquidity look plentiful.

It nets obligations, accepts collateral, supplies queues, records a handsome closing balance, and gently encourages the firm to believe that a pile of assets and a modest net exposure amount to a settled payment.

Then, at noon, it asks for a particular amount of money, in a particular currency, through a particular account.

Your assets remain handsome. They are simply not there yet.

This is not an argument against netting.

Netting is one of the civilising achievements of the financial system: it takes a thicket of bilateral claims and reduces the amount that must finally move.

Nor is it an argument against collateral, central-bank money or an intraday facility.

All useful inventions.

The trouble begins when a firm treats the private fact that it is solvent, or the market fact that it can sell something eventually, as evidence that it can make an infrastructure payment now.

The distinction is almost offensively simple.

Market liquidity is the capacity to trade a position at a stated size and state. Funding liquidity is the capacity to obtain the cash or eligible collateral required on schedule.

A tight two-cent spread on five hundred shares says nothing decisive about selling fifty thousand, and much less about paying an obligation before a settlement cut-off.

A quote is a price for a small immediate order. It is not a promise from the payment system.

The systems themselves know this perfectly well.

A government-bond trade can be economically agreed at 10:15 CET.

The trade is matched.

Both sides enjoy the pleasant sensation of having done business.

Delivery-versus-payment still requires an available position in the delivering account, and cash or approved intraday liquidity on the receiving side. A securities account holding a comfortable balance offers rather less if part of it is already earmarked.

That is the point at which the story stops being about a bond and becomes about plumbing.

A matched trade is an agreement to settle.

A final transfer changes the records from which somebody may actually use the bond, the cash or the collateral.

Between the two sits an interval, and it is rarely funded by the people who have just congratulated one another on the net number.

The Seven Clocks, All of Them Annoyingly Real

The useful retrospective question is not whether there was liquidity.

It is what you had usable, in the right place, at each clock.

There are seven timing mismatches.

Omitting any one of them is how a calm treasury pack acquires a small but very attentive audience.

Valuation time comes first.

A mark says what the asset is thought to be worth at a defined moment.

It is neither a sale nor a payment.

A screen price can belong to a security already pledged, subject to a transfer restriction, or unacceptable to the particular clearing system.

The collateral schedule wants currency, issuer, maturity, legal ownership, concentration, valuation source and settlement location. It is tiresome because it is trying to establish whether the asset can do the job.

Margin-call time is second.

Variation margin is a daily, sometimes intraday, demand for collateral following a mark-to-market loss.

The market can say quite truthfully that the portfolio remains valuable. The counterparty can say quite truthfully that it is owed eligible resources now. Both remarks may be made over the same coffee, by two people who will have very different afternoons.

Collateral-mobilisation time is third.

A sovereign bond supports a high proportion of its market value; a volatile or thinly traded security supports considerably less.

Those haircuts reserve for the cost and uncertainty of a rapid sale. They do not confer on the operations team the ability to move a bond through custody, repo and settlement by moral force.

Encumbrance, legal ownership, transferability, concentration and the route to the relevant account decide whether the recognised value arrives before the call becomes overdue.

FX-conversion time is fourth.

An obligation has a currency.

So does the collateral.

That both have a number beside them is charming, and insufficient.

Conversion must occur, the converted funds must settle, and the receiving account must be able to use them. The necessary currency may be trapped behind a market close, a payment cut-off, a correspondent bank, or a queue with developed views about priority.

The settlement cut-off is fifth.

Operator rulebooks turn general principles into something actionable: membership, account structures, message formats, queue processing, collateral eligibility and cut-offs are not supervisory poetry.

Fedwire Securities closes its secondary-market transfer channel at 3:15 p.m. ET and ends the securities business day at 7:00 p.m.

ET.

A trade agreed a minute after the cut-off is perfectly valid. It has merely missed that day’s channel, a distinction which becomes unusually vivid when the incoming payment was meant to fund an outgoing one.

Credit-draw time is sixth.

An approved source of intraday liquidity is not a balance in the settlement account.

It has a borrower, a facility agreement, eligible collateral, valuation and haircut terms, operating instructions, a currency, a settlement-bank route, and a moment at which the proceeds become usable. A firm that counts the facility before testing those conditions is not forecasting liquidity.

It is admiring the word committed.

Payment-finality time is seventh.

PFMI gives market infrastructures a shared vocabulary for exactly this problem.

For an exchange-of-value system the relevant principle is not that both parties feel reassured. It is that final settlement of one obligation occurs if and only if final settlement of the linked obligation occurs.

Until then an expected payment is still expected.

A green dashboard is not a legal state.

Put all seven into one time-bucketed forecast: valuation, margin call, collateral mobilisation, FX conversion, settlement cut-off, credit draw, payment finality. The apparent surplus has to survive every one of them.

Elaborate?

Certainly.

The elaborate version is the one that pays on time.

Netting Changes the Invoice, Not the Noon Payment

Netting is regularly credited with removing your liquidity risk.

It does not.

It changes the invoice.

That is already a fine achievement, and it should not be mistaken for an alchemy in which gross operational demands evaporate because the final exposure looks modest in a report.

A CSD may settle individual gross instructions or net certain obligations before the final cash transfer. Either design still needs a debitable securities account and a debitable cash account at the moment the platform attempts settlement.

The matched instruction waits in a queue until both legs can complete. It protects the buyer from paying and then waiting for the bond, and the seller from delivering and then hoping to be paid. It does not reserve either resource when the traders receive their confirmation.

The distinction matters at scale.

T2S settles something close to a million transactions a day, and its settlement efficiency by volume runs in the low nineties as a percentage.

The remainder is not evidence of scandal, still less a legal finding. It is a reminder that a settled market consists of eligible instructions, available balances, queues, rules and exceptions, rather than of trades that looked plausible at execution.

Suppose the queue reaches the delivering bank and part of the bond position turns out to be pledged elsewhere. The system does not deliver what it can and compose a thoughtful note about the rest.

It holds the instruction pending, or fails it under the rulebook, while the buyer keeps its cash.

The record then has to say whether the problem was missing securities, missing cash, an unmatched instruction, an expired cut-off, a system outage or an account restriction, and whether the trade later settled, was cancelled, or was bought in.

Failed is not a diagnosis. It is a doorway to several meetings.

This is also why market liquidity and funding liquidity deteriorate together. Dealer capital, secured financing, margin models, clearing arrangements, benchmark hedges, investor withdrawals, currencies and settlement channels are shared infrastructure.

Forty names diversify a borrower default.

They do not establish forty independent sources of noon cash.

A sharp reduction in available dealer funding can pull quotes in three unrelated asset classes without changing any issuer’s revenues. The apparently separate exits discover that they have all been using the same door.

The Ratio Is Not the Day

Prudential coverage ratios are useful precisely because they make a large and complicated funding structure visible.

They are not a substitute for the intraday timetable.

A coverage ratio is a statement about a specified stress scenario over a defined horizon. A structural funding ratio is a statement about maturity transformation over a much longer one.

Both are meaningful.

Neither is evidence that the surplus sits in the required currency, in the required entity, in the required account, on the required settlement route, at the exact moment a CSD or payment system calls for it.

That is not a criticism of either ratio.

It is an objection to their being asked to do a job they were never designed for.

An intraday payment forecast has to add currencies, entity boundaries, encumbrance, concentration, account location, payment windows, facility conditions and finality.

A ratio is a representation of explicit assumptions.

It is not a dinner jacket in which every funding risk becomes respectable.

Who Got the Certainty

Participants benefit from netting, because fewer final resources need to move, and from delivery-versus-payment, because they need not surrender cash or securities on faith.

They benefit from faster certainty, because a legally final receipt can support a new sale, a collateral pledge or a corporate-action entitlement.

Infrastructures benefit from prompt resources, because a queue that clears on time is much less likely to become a bilateral argument conducted across several time zones.

There is a public benefit as well.

PFMI’s insistence on legal basis, governance, risk management, settlement finality, custody protection and operational resilience does not make a market incapable of stress. It makes the transfer point legible when stress arrives.

A system that knows whether a payment is final can stop unwinding queued obligations and release the collateral.

Less glamorous than announcing liquidity.

Considerably more useful.

Who Financed the Interval

Firms pay, by holding buffers that are unencumbered, eligible, in the right currency and close enough to the settlement route to matter.

That is more expensive than holding a handsome portfolio and assuming it will behave as cash when addressed firmly.

Banks pay, by providing intraday credit, funding inventory, and deciding which collateral and account structures they will accept. Their balance sheets are the interval between a private promise and a final payment.

Markets pay when the interval is not funded at all. A forced seller does not get a patient seller’s price.

The all-in liquidation cost takes in average execution price, commissions, market impact and the risk carried while the order is worked. The public quote is only a lower bound for trading size.

A one-basis-point spread can coexist with wretched block capacity. Which is the sort of discovery that makes a risk committee unexpectedly interested in market microstructure.

So the distribution is neither heroic nor sinister.

Netting and prompt finality benefit those receiving certainty and the systems protected by usable resources. Buffers, intraday credit and the cost of forced sales are borne by the firms that must supply the interval.

The policy did not abolish liquidity risk.

It made clear that you are expected to finance it before noon.

Measure liquidity at the time the obligation falls due, not at the time the balance sheet looks handsome.