The box nobody ticked
01 September 2026
In the final instalment of the series, Cyril Louchtchay de Fleurian, head of securities finance and balance sheet strategy at Capteo: Strategy & Management Consulting, talks lessons learnt
Image: stock.adobe.com/NexPix
One per cent
On 31 July 2026 the European Central Bank (ECB) published the results of an exercise announced in December 2025. Its most interesting finding may be buried in its least-read chapter: Geopolitical risk reverse stress test of euro area banks — 2026 SSM thematic stress test: final results, table 4. Here, 110 euro area banks were handed a target: a 300 basis point fall in the CET1 ratio — 20 per cent of banks went further — calibrated on the crisis episodes of the past 15 to 20 years. Each then had to imagine the geopolitical scenario capable of producing it. The exercise inverts the usual stress test, in which every bank takes the same shock — here each builds its own. It was framed around capital, with liquidity bolted on as a secondary dimension. The more interesting lesson comes from the second.
Several banks in the panel, having diligently described the catastrophe that might befall them, found that their liquidity held up remarkably well under that same scenario. Across the sample, the median Liquidity Coverage Ratio (LCR) falls from 186 per cent to 163 per cent over one year and stays comfortably above the regulatory minimum. The median hides a more troubling result: at several institutions, a scenario severe enough to take a deep bite out of capital produces almost no additional strain on the liquidity metrics. The finding could be read as resilience. The supervisor read it as a modelling weakness and has said it will follow up; in a crisis, it notes, solvency and liquidity are often strongly interrelated.
Asked what they would do if their own disaster scenario arrived, banks cite suspending dividends at 59 per cent, managing down exposures at 57 per cent, IT resilience at 74 per cent, optimising the funding mix at 35 per cent, strengthening collateral and margin at a modest 20 per cent, and tightening repo haircuts at just 1 per cent. If your institution is directly supervised by the ECB it is one of those 110 banks, and somebody in your building ticked some of these boxes last spring.
A second blind spot shows up in foreign currency liquidity: several institutions project no variability at all, even though the FX LCR is structurally tighter than the aggregate ratio and drops below 100 per cent at some banks.
From these answers the report draws an observation the exercise was bound to produce, since it added up 110 crisis plans: 40 per cent of banks intend to sell assets, 50 per cent to cut new business, 59 per cent to suspend distributions. Each action is plausible on its own; all of them are a good deal less so when half the system runs them on the same morning. Somebody will have to buy those assets. A plan written as though nobody else had one is an intention drafted in calm weather.
The same demand for realism, applied to liquidity, points to the operational chain that turns an asset into cash. A bank that says it can monetise its buffer without ever having tested that chain is describing exactly the same kind of action. That is the bridge to the argument these columns have been making for six months, and it comes down to one idea: when 110 banks list what they would do to survive, the collateral-repo-cash chain barely appears. It is more than a modelling blind spot; it also mirrors the organisation. That chain runs through treasury, ALM, the markets business, risk, collateral management, and operations, without any one function necessarily carrying end-to-end execution. Its near-absence from the answers sent to the ECB reflects, at least in part, the absence of a clearly assigned end-to-end responsibility.
Hormuz from a distance
Operational risk completes the picture. Cyber incidents and the disruption of third-party services stand out as the sharpest threats, with the ECB asking that operational resilience and cyber risk be built into stress testing frameworks. The supervisor arrives here, by another route, at what ‘What risk management cannot promise’ (SFT no. 406) called the convergence of outages.
There is a second reading, less comforting: the models worked perfectly. They measured what they were asked to measure — a level of liquidity. A crisis does not turn on a level, it turns on what the bank manages to mobilise, by which route, before which hour. That, liquidity models still measure poorly. It was the subject of a question put in these columns last June, about the Strait of Hormuz (SFT no. 405, ‘The weapon that never fires’), to which an answer was not expected quite so soon.
About a quarter of banks did include it, under a Middle East conflict, among their trigger events. Most assumed transmission through the real economy, with market impacts following, and the projected losses concentrated in sectors exposed to trade, energy, and logistics disruption: agriculture, construction, manufacturing, transport, accommodation, and food services. Hormuz entered the exercise as an oil shock. Not as a chokepoint — which remains its most striking feature. The allegory did not make the cut. So much for that.
That leaves what the supervisor does with these findings. The qualitative deficiencies identified can feed the supervisory dialogue, be taken into the governance element of the Supervisory Review and Evaluation Process (SREP) and, where relevant, affect Pillar 2 requirements — Pillar 2 guidance is explicitly outside the scope of the exercise. The areas flagged for supervisory attention also cover the Internal Capital Adequacy Assessment Process (ICAAP) and Internal Liquidity Adequacy Assessment Process (ILAAP) frameworks. With geopolitical risk remaining a supervisory priority for 2026–28, the prudential transmission channel is clear enough. The timing follows the supervisory cycle: ICAAP and ILAAP preparation, then SREP decisions are usually notified late in the year for application the year after.
None of that detracts from what is, overall, an excellent report, which moves the discussion from the level of the ratios to the credibility of the machinery meant to protect them once the scenario stops being theoretical. All of it is set out without the slightest sense of urgency — which is the surest way of ensuring that nobody feels any.
The bill
Having set out the change of regime in liquidity (SFT no. 399, 400, 401, 403), explored the new risks that come with it (SFT no. 405, 406) and shown why the Enterprise Liquidity Management (ELM) stands as the natural framework for managing the executability of liquidity (SFT no. 408, 409), it is worth closing this run with the four main lessons and paradoxes of that shift.
The first is: the bank that hoards its liquidity loses it; the one that uses it keeps it, and keeps it in working order. Turning prudence into imprudence took a certain genius; the industry managed it without effort.
The second will console the auditors: a compliant buffer is a citadel with rusted hinges. It reassures for as long as nobody pushes at the gate. We learnt to build solid walls, only to find that the battle was fought at the entrance — at the precise hour the gate had to open.
The third is the awkward one: somebody holds that gate. Repo decides who comes in, at what hour and at what price, without ever using the word sanction; it is the most courteous form of coercion — and the most effective.
Then the last, less a paradox than a bill. Liquidity has long been taken for a promise; a promise without a date is not prudence — it is an overdraft that does not know its own name.
One hundred and ten banks imagined their own catastrophe, and the promise held without being tested: liquidity ratios stayed compliant, the crisis plans barely brought into view the chain that makes them executable, and no end-to-end responsibility emerged with any clarity. The diagnosis is on the table; the decisions it calls for still have to be taken — and the timetable is now in your hands.
Liquidity has become a right of access, and therefore a matter of command
Liquidity used to be observed, through stocks, ratios, and stress tests. It is now proved by the ability to turn an asset into cash in the right place, at the right moment, through the right post-trade chains and inside real cut-offs. The consequence follows directly: once liquidity becomes conditional and kinetic, the absence of an owner stops being an organisational flaw and becomes a self-inflicted strategic risk.
That risk does not surface at the end of the chain, after a ratio breach; it emerges upstream, when margins, haircuts, and intraday windows dictate the sequence of actions faster than governance can decide it. In practical terms, liquidity has stopped being a prudential attribute and become a capacity to act under constraint. And a capacity to act has to be commanded. First item on the table.
Who holds the mandate to arbitrate intraday?
In practice, bank organisations are still calibrated for a world in which liquidity is presumed available and then managed by exception. This series argues the opposite: liquidity forms a coupled system, exposed to discontinuities, procyclicality, non-bank financial institutions (NBFIs) dependencies and infrastructures whose availability is never guaranteed. Its economics can no longer be separated from its ability to execute. Liquidity has become a resilience capability on the same footing as cybersecurity or business continuity, with one difference: it is also a source of P&L — and of power.
For as long as a bank declines to treat liquidity as a strategic execution function, it will keep overpaying for compliant buffers and keep discovering too late that it cannot act. The conclusion is less about strengthening liquidity than about taking back control of executable liquidity, which in this regime means governing access, governing velocity, and governing dependencies. Today, no single function clearly owns any of the three; they sit scattered across teams that are legitimate in their own right — but none has authority over end-to-end execution. Second item on the table.
Who explicitly carries responsibility for executable liquidity today, and with what decision rights?
Repo is an infrastructure of power, and the risks are already live. This series strips repo of its status as neutral infrastructure and repositions it as a chokepoint, and therefore as a filtering mechanism. A chokepoint by definition hands out a right of access — to liquidity, to leverage, to continuity of execution.
Liquidity becomes geopolitical from that point, not because it depends on states, but because it depends on eligibility rules, membership, margin models, and jurisdictions — which is to say on silos of sovereignty. That carries a structural idea many banks have yet to absorb: choosing an infrastructure (CCP, triparty, CSD, ICSD, and so on) determines far more than efficiency; it sets an operating regime, and at times a political position. Cleared repo, where the main attraction is balance sheet netting, is the plainest example.
In a fragmented ecosystem, the intrinsic quality of a piece of collateral tells only part of the story; its real value depends on whether it can actually be converted in the right pools, at the right hours, with legal certainty that holds. That is why the new risks read as a map of mechanisms through which an outside party, public or private, can constrain a bank without ever using the word sanction. The risk classification set out in SFT no. 406, ‘What risk management cannot promise’, offers a way in: some shocks cannot be recovered from, others are amplified by internal disorder, others erode ROE until the strategy can no longer be funded.
The critical point follows. Repo works as an instrument of constraint because it is technically defensible as risk management, immediate in its effects and politically discreet. In this logic of economic warfare, the vulnerability lies in chokepoints with no fallback and no governance. Third item on the table.
Which dependencies are we prepared to accept, and which must be neutralised by design?
That question is what calls for a forward view. The underlying trajectory is one of accumulating constraints — central clearing, margin requirements, balance sheet limits, the procyclicality of NBFI liquidity — whose combined effect is to multiply the points at which the chain can break. The best ratio offers no protection against an operational chokepoint; it can even manufacture false confidence and delay the decision. That makes liquidity a question of power: how fast a bank can mobilise, through how many routes, and how reliably it can repeat the exercise. If repo organises a right of access, the bank has to manage that right the way it manages credit risk today, with limits, governance, and a strategy for dependencies. Fourth item on the table.
The ELM turns liquidity from a buffer into an option
The ELM follows logically from that: it formalises a missing owner. Its perimeter still has to be drawn, or the term ends up containing everything and designating nothing. The ELM is the operating model that assigns ownership of executable liquidity, sets its decision rights, and measures its mobilisation capacity. It is not there to compete with treasury, ALM, the repo business, risk, collateral management, or operations. It gives them a common centre of gravity; it does not replace them.
Its first pivot is pre-positioning. Until collateral sits pre-positioned with central banks, CCPs and triparty agents — documentation signed, limits set, routing live — liquidity remains a promise without a date. The chain that runs from assets to collateral, repo, cash, and margin has to be workable in normal times to stay executable under stress.
Monetising part of the buffer regularly in normal times follows the same logic. It avoids the cliff effect: the crisis becomes the acceleration of a routine already practised rather than improvisation at the worst moment. The point is not to consume liquidity in order to prove it exists, it is to keep the routes, the market relationships, the operational capacity, and the know-how in working order for the day mobilisation is needed. Any P&L gain comes as a bonus.
The second pivot is the shift from measured liquidity to managed liquidity, through a Risk Appetite Framework (RAF) and operational KPIs: mobilisability, velocity, channel concentration, infrastructure dependency, implicit cost by activity. Banks already have plenty of ratios; what they lack is a time-to-cash they can stand behind, tested and monitored with the attention given to an intraday market risk.
The value of the ELM then runs beyond protection. It converts a defensive posture — pulling back to preserve yourself — into a capacity to choose where to commit the balance sheet, and so into competitive advantage when the market contracts. This is where the economic warfare dimension sits: resilience becomes the ability to keep operating when others cannot — selectively, with discipline, and on economically rational terms. Fifth item on the table.
Which strategic choices have we already given up — not by decision, but under liquidity constraint?
Making those choices possible again takes three things: a mandate, demonstrated execution capability, and an internal economic model.
The first is a clear mandate with enforceable rights. Intraday is not run by consensus. It requires an identifiable owner and explicit rights: prioritising cash, collateral and margin, switching on stress modes, single-point arbitration when they conflict. Without rights the orchestrator becomes an observer; without an owner, the decision fragments and arrives after the cut-off.
The second is proof by execution. Management starts from a minimal, auditable base: a control tower covering positions, encumbrance, eligibility, haircuts, and cut-offs; regular testing of the rails that convert into cash — central bank, CCP, triparty, bilateral, and securities lending; a T+0/T+1 playbook run to an hourly sequence. The objective stays operational: being able to answer, every morning, who pays what, with which collateral, through which channel, and how quickly.
The third is an internal economic model. Resistance rarely comes from the principle; it comes from the economics of the arrangement, and from the question of who pays for liquidity, who carries the redundancy and which business loses P&L. An intraday Liquidity Transfer Price (LTP) makes low-risk, high-cost volumes visible, aligns incentives and removes implicit arbitrage. Exceptions remain, not least to protect a franchise, provided they are explicit, recorded and bounded: an exception that is owned is a choice, an exception left undocumented is a risk.
All three should leave evidence behind: time-to-cash by channel, median and worst case, cut-off performance, the share of unencumbered high-quality liquid asset (HQLA) against the floor, intraday incidents and resolution times, availability of alternative routes, costs made visible and then progressively allocated. No audit is needed to check any of it. An organisation that meets the conditions answers three questions on the spot: what is its median and worst-case time-to-cash by channel? What share of its buffer was actually converted into cash over the past 12 months, incidents excluded? What proportion of its collateral sits in a jurisdiction where a unilateral decision could block access to it? An organisation that has to convene eight people to answer has just obtained its answer. Sixth item on the table.
The decisions the ExCo can no longer delegate
At the end of this run, comparative advantage is measured by the ability to convert the buffer into cash — quickly, by several routes, without improvised arbitration. The level of a compliant buffer says nothing about that ability. What is at stake comes down to three decisions, and none of them can be delegated.
Set a target level of execution sovereignty: which channels have to stay open, in which currencies, with what time-to-cash, and what fallback routes. Accept the explicit cost of that redundancy: pre-positioning, operational capacity, documentation, limits, and testing. Decide which dependencies are tolerable and which have to be neutralised before they become constraints. Leave these choices unmade and the ELM stays a governance intention. Make them and it becomes a strategic capability: liquidity stops being a ratio and becomes an option — and the option becomes the power to act. Seventh item on the table.
The price of silence
The risk is already live, and it is silent. It does not show up in the ratios; it shows up in the decisions a bank can no longer take, or delays, or takes at a cost it has never made visible. The next stress will not ask how much HQLA sat in the buffer at the previous night’s close; it will reveal how much of it could become cash, by which route, under which dependencies, and how quickly.
Banks know how to measure, how to produce the ratios and how to survive crises. They are less good at arbitrating a cost, spotting a slow erosion, assigning a responsibility, or correcting a structural inefficiency. The difficulty owes less to competence than to the legacy of 15 years of reform — each one justified on its own terms, the accumulation rather less so.
The stock grew while its use narrowed. Buffers have never been larger, nor the room to mobilise them tighter; collateral has never been more abundant, nor less mobile, segregation, operational constraints, and fragmentation having turned velocity into the adjustment variable.
Measurement and governance have proliferated without delivering what was expected of them. Prudential indicators have multiplied faster than the understanding of the economic mechanisms they describe; committees have been added faster than responsibilities have been assigned, to the point that many liquidity costs still have no owner at all. As for the layers of protection, they are paid for twice — once upfront and again every day, through friction. The sum of individual prudence ends up producing collective fragility: optimised in isolation, each institution withdraws from the system a little of the elasticity they all depend on the same morning. An uncomfortable rule emerges from all of this: the most expensive risks set off no alarm, and silent erosion outruns, in the end, the price of visible crises.
Several shifts are already shaping the decade ahead; the supervisory cycle, for its part, is annual, and it has already started.
The first is the shift from compliance to optimisation: meeting the ratios is an obligation, minimising their economic cost is the job. The second is to make hidden costs visible — those of fragmentation and structural inefficiency — and to assign them to somebody. The third puts in place an economic governance of liquidity, cross-functional by design: treasury, risk, finance, the repo businesses and infrastructures. The fourth erases boundaries that are already fading, those separating liquidity, solvency, capital, leverage, and the balance sheet — managed apart and consumed together. The fifth concerns the circulation of collateral, where speed now counts for more than the quantity available.
A polite market
On 31 July 2026 the supervisor measured what can be measured. The rest will appear in no supervisory report; it will appear, or not, in a set of minutes. We began with the least-read chapter of a public document; we end with the least-read document in a bank. Both take 10 minutes to read.
One box remains. One hundred and ten banks imagined their worst day, then listed what they would do to survive it; 1 in 100 planned to tighten its repo haircuts. The other 99 have therefore modelled a crisis in which their counterparties — which is to say themselves — would remain accommodating. The LCR barely moves. It all adds up.
On 31 July 2026 the European Central Bank (ECB) published the results of an exercise announced in December 2025. Its most interesting finding may be buried in its least-read chapter: Geopolitical risk reverse stress test of euro area banks — 2026 SSM thematic stress test: final results, table 4. Here, 110 euro area banks were handed a target: a 300 basis point fall in the CET1 ratio — 20 per cent of banks went further — calibrated on the crisis episodes of the past 15 to 20 years. Each then had to imagine the geopolitical scenario capable of producing it. The exercise inverts the usual stress test, in which every bank takes the same shock — here each builds its own. It was framed around capital, with liquidity bolted on as a secondary dimension. The more interesting lesson comes from the second.
Several banks in the panel, having diligently described the catastrophe that might befall them, found that their liquidity held up remarkably well under that same scenario. Across the sample, the median Liquidity Coverage Ratio (LCR) falls from 186 per cent to 163 per cent over one year and stays comfortably above the regulatory minimum. The median hides a more troubling result: at several institutions, a scenario severe enough to take a deep bite out of capital produces almost no additional strain on the liquidity metrics. The finding could be read as resilience. The supervisor read it as a modelling weakness and has said it will follow up; in a crisis, it notes, solvency and liquidity are often strongly interrelated.
Asked what they would do if their own disaster scenario arrived, banks cite suspending dividends at 59 per cent, managing down exposures at 57 per cent, IT resilience at 74 per cent, optimising the funding mix at 35 per cent, strengthening collateral and margin at a modest 20 per cent, and tightening repo haircuts at just 1 per cent. If your institution is directly supervised by the ECB it is one of those 110 banks, and somebody in your building ticked some of these boxes last spring.
A second blind spot shows up in foreign currency liquidity: several institutions project no variability at all, even though the FX LCR is structurally tighter than the aggregate ratio and drops below 100 per cent at some banks.
From these answers the report draws an observation the exercise was bound to produce, since it added up 110 crisis plans: 40 per cent of banks intend to sell assets, 50 per cent to cut new business, 59 per cent to suspend distributions. Each action is plausible on its own; all of them are a good deal less so when half the system runs them on the same morning. Somebody will have to buy those assets. A plan written as though nobody else had one is an intention drafted in calm weather.
The same demand for realism, applied to liquidity, points to the operational chain that turns an asset into cash. A bank that says it can monetise its buffer without ever having tested that chain is describing exactly the same kind of action. That is the bridge to the argument these columns have been making for six months, and it comes down to one idea: when 110 banks list what they would do to survive, the collateral-repo-cash chain barely appears. It is more than a modelling blind spot; it also mirrors the organisation. That chain runs through treasury, ALM, the markets business, risk, collateral management, and operations, without any one function necessarily carrying end-to-end execution. Its near-absence from the answers sent to the ECB reflects, at least in part, the absence of a clearly assigned end-to-end responsibility.
Hormuz from a distance
Operational risk completes the picture. Cyber incidents and the disruption of third-party services stand out as the sharpest threats, with the ECB asking that operational resilience and cyber risk be built into stress testing frameworks. The supervisor arrives here, by another route, at what ‘What risk management cannot promise’ (SFT no. 406) called the convergence of outages.
There is a second reading, less comforting: the models worked perfectly. They measured what they were asked to measure — a level of liquidity. A crisis does not turn on a level, it turns on what the bank manages to mobilise, by which route, before which hour. That, liquidity models still measure poorly. It was the subject of a question put in these columns last June, about the Strait of Hormuz (SFT no. 405, ‘The weapon that never fires’), to which an answer was not expected quite so soon.
About a quarter of banks did include it, under a Middle East conflict, among their trigger events. Most assumed transmission through the real economy, with market impacts following, and the projected losses concentrated in sectors exposed to trade, energy, and logistics disruption: agriculture, construction, manufacturing, transport, accommodation, and food services. Hormuz entered the exercise as an oil shock. Not as a chokepoint — which remains its most striking feature. The allegory did not make the cut. So much for that.
That leaves what the supervisor does with these findings. The qualitative deficiencies identified can feed the supervisory dialogue, be taken into the governance element of the Supervisory Review and Evaluation Process (SREP) and, where relevant, affect Pillar 2 requirements — Pillar 2 guidance is explicitly outside the scope of the exercise. The areas flagged for supervisory attention also cover the Internal Capital Adequacy Assessment Process (ICAAP) and Internal Liquidity Adequacy Assessment Process (ILAAP) frameworks. With geopolitical risk remaining a supervisory priority for 2026–28, the prudential transmission channel is clear enough. The timing follows the supervisory cycle: ICAAP and ILAAP preparation, then SREP decisions are usually notified late in the year for application the year after.
None of that detracts from what is, overall, an excellent report, which moves the discussion from the level of the ratios to the credibility of the machinery meant to protect them once the scenario stops being theoretical. All of it is set out without the slightest sense of urgency — which is the surest way of ensuring that nobody feels any.
The bill
Having set out the change of regime in liquidity (SFT no. 399, 400, 401, 403), explored the new risks that come with it (SFT no. 405, 406) and shown why the Enterprise Liquidity Management (ELM) stands as the natural framework for managing the executability of liquidity (SFT no. 408, 409), it is worth closing this run with the four main lessons and paradoxes of that shift.
The first is: the bank that hoards its liquidity loses it; the one that uses it keeps it, and keeps it in working order. Turning prudence into imprudence took a certain genius; the industry managed it without effort.
The second will console the auditors: a compliant buffer is a citadel with rusted hinges. It reassures for as long as nobody pushes at the gate. We learnt to build solid walls, only to find that the battle was fought at the entrance — at the precise hour the gate had to open.
The third is the awkward one: somebody holds that gate. Repo decides who comes in, at what hour and at what price, without ever using the word sanction; it is the most courteous form of coercion — and the most effective.
Then the last, less a paradox than a bill. Liquidity has long been taken for a promise; a promise without a date is not prudence — it is an overdraft that does not know its own name.
One hundred and ten banks imagined their own catastrophe, and the promise held without being tested: liquidity ratios stayed compliant, the crisis plans barely brought into view the chain that makes them executable, and no end-to-end responsibility emerged with any clarity. The diagnosis is on the table; the decisions it calls for still have to be taken — and the timetable is now in your hands.
Liquidity has become a right of access, and therefore a matter of command
Liquidity used to be observed, through stocks, ratios, and stress tests. It is now proved by the ability to turn an asset into cash in the right place, at the right moment, through the right post-trade chains and inside real cut-offs. The consequence follows directly: once liquidity becomes conditional and kinetic, the absence of an owner stops being an organisational flaw and becomes a self-inflicted strategic risk.
That risk does not surface at the end of the chain, after a ratio breach; it emerges upstream, when margins, haircuts, and intraday windows dictate the sequence of actions faster than governance can decide it. In practical terms, liquidity has stopped being a prudential attribute and become a capacity to act under constraint. And a capacity to act has to be commanded. First item on the table.
Who holds the mandate to arbitrate intraday?
In practice, bank organisations are still calibrated for a world in which liquidity is presumed available and then managed by exception. This series argues the opposite: liquidity forms a coupled system, exposed to discontinuities, procyclicality, non-bank financial institutions (NBFIs) dependencies and infrastructures whose availability is never guaranteed. Its economics can no longer be separated from its ability to execute. Liquidity has become a resilience capability on the same footing as cybersecurity or business continuity, with one difference: it is also a source of P&L — and of power.
For as long as a bank declines to treat liquidity as a strategic execution function, it will keep overpaying for compliant buffers and keep discovering too late that it cannot act. The conclusion is less about strengthening liquidity than about taking back control of executable liquidity, which in this regime means governing access, governing velocity, and governing dependencies. Today, no single function clearly owns any of the three; they sit scattered across teams that are legitimate in their own right — but none has authority over end-to-end execution. Second item on the table.
Who explicitly carries responsibility for executable liquidity today, and with what decision rights?
Repo is an infrastructure of power, and the risks are already live. This series strips repo of its status as neutral infrastructure and repositions it as a chokepoint, and therefore as a filtering mechanism. A chokepoint by definition hands out a right of access — to liquidity, to leverage, to continuity of execution.
Liquidity becomes geopolitical from that point, not because it depends on states, but because it depends on eligibility rules, membership, margin models, and jurisdictions — which is to say on silos of sovereignty. That carries a structural idea many banks have yet to absorb: choosing an infrastructure (CCP, triparty, CSD, ICSD, and so on) determines far more than efficiency; it sets an operating regime, and at times a political position. Cleared repo, where the main attraction is balance sheet netting, is the plainest example.
In a fragmented ecosystem, the intrinsic quality of a piece of collateral tells only part of the story; its real value depends on whether it can actually be converted in the right pools, at the right hours, with legal certainty that holds. That is why the new risks read as a map of mechanisms through which an outside party, public or private, can constrain a bank without ever using the word sanction. The risk classification set out in SFT no. 406, ‘What risk management cannot promise’, offers a way in: some shocks cannot be recovered from, others are amplified by internal disorder, others erode ROE until the strategy can no longer be funded.
The critical point follows. Repo works as an instrument of constraint because it is technically defensible as risk management, immediate in its effects and politically discreet. In this logic of economic warfare, the vulnerability lies in chokepoints with no fallback and no governance. Third item on the table.
Which dependencies are we prepared to accept, and which must be neutralised by design?
That question is what calls for a forward view. The underlying trajectory is one of accumulating constraints — central clearing, margin requirements, balance sheet limits, the procyclicality of NBFI liquidity — whose combined effect is to multiply the points at which the chain can break. The best ratio offers no protection against an operational chokepoint; it can even manufacture false confidence and delay the decision. That makes liquidity a question of power: how fast a bank can mobilise, through how many routes, and how reliably it can repeat the exercise. If repo organises a right of access, the bank has to manage that right the way it manages credit risk today, with limits, governance, and a strategy for dependencies. Fourth item on the table.
The ELM turns liquidity from a buffer into an option
The ELM follows logically from that: it formalises a missing owner. Its perimeter still has to be drawn, or the term ends up containing everything and designating nothing. The ELM is the operating model that assigns ownership of executable liquidity, sets its decision rights, and measures its mobilisation capacity. It is not there to compete with treasury, ALM, the repo business, risk, collateral management, or operations. It gives them a common centre of gravity; it does not replace them.
Its first pivot is pre-positioning. Until collateral sits pre-positioned with central banks, CCPs and triparty agents — documentation signed, limits set, routing live — liquidity remains a promise without a date. The chain that runs from assets to collateral, repo, cash, and margin has to be workable in normal times to stay executable under stress.
Monetising part of the buffer regularly in normal times follows the same logic. It avoids the cliff effect: the crisis becomes the acceleration of a routine already practised rather than improvisation at the worst moment. The point is not to consume liquidity in order to prove it exists, it is to keep the routes, the market relationships, the operational capacity, and the know-how in working order for the day mobilisation is needed. Any P&L gain comes as a bonus.
The second pivot is the shift from measured liquidity to managed liquidity, through a Risk Appetite Framework (RAF) and operational KPIs: mobilisability, velocity, channel concentration, infrastructure dependency, implicit cost by activity. Banks already have plenty of ratios; what they lack is a time-to-cash they can stand behind, tested and monitored with the attention given to an intraday market risk.
The value of the ELM then runs beyond protection. It converts a defensive posture — pulling back to preserve yourself — into a capacity to choose where to commit the balance sheet, and so into competitive advantage when the market contracts. This is where the economic warfare dimension sits: resilience becomes the ability to keep operating when others cannot — selectively, with discipline, and on economically rational terms. Fifth item on the table.
Which strategic choices have we already given up — not by decision, but under liquidity constraint?
Making those choices possible again takes three things: a mandate, demonstrated execution capability, and an internal economic model.
The first is a clear mandate with enforceable rights. Intraday is not run by consensus. It requires an identifiable owner and explicit rights: prioritising cash, collateral and margin, switching on stress modes, single-point arbitration when they conflict. Without rights the orchestrator becomes an observer; without an owner, the decision fragments and arrives after the cut-off.
The second is proof by execution. Management starts from a minimal, auditable base: a control tower covering positions, encumbrance, eligibility, haircuts, and cut-offs; regular testing of the rails that convert into cash — central bank, CCP, triparty, bilateral, and securities lending; a T+0/T+1 playbook run to an hourly sequence. The objective stays operational: being able to answer, every morning, who pays what, with which collateral, through which channel, and how quickly.
The third is an internal economic model. Resistance rarely comes from the principle; it comes from the economics of the arrangement, and from the question of who pays for liquidity, who carries the redundancy and which business loses P&L. An intraday Liquidity Transfer Price (LTP) makes low-risk, high-cost volumes visible, aligns incentives and removes implicit arbitrage. Exceptions remain, not least to protect a franchise, provided they are explicit, recorded and bounded: an exception that is owned is a choice, an exception left undocumented is a risk.
All three should leave evidence behind: time-to-cash by channel, median and worst case, cut-off performance, the share of unencumbered high-quality liquid asset (HQLA) against the floor, intraday incidents and resolution times, availability of alternative routes, costs made visible and then progressively allocated. No audit is needed to check any of it. An organisation that meets the conditions answers three questions on the spot: what is its median and worst-case time-to-cash by channel? What share of its buffer was actually converted into cash over the past 12 months, incidents excluded? What proportion of its collateral sits in a jurisdiction where a unilateral decision could block access to it? An organisation that has to convene eight people to answer has just obtained its answer. Sixth item on the table.
The decisions the ExCo can no longer delegate
At the end of this run, comparative advantage is measured by the ability to convert the buffer into cash — quickly, by several routes, without improvised arbitration. The level of a compliant buffer says nothing about that ability. What is at stake comes down to three decisions, and none of them can be delegated.
Set a target level of execution sovereignty: which channels have to stay open, in which currencies, with what time-to-cash, and what fallback routes. Accept the explicit cost of that redundancy: pre-positioning, operational capacity, documentation, limits, and testing. Decide which dependencies are tolerable and which have to be neutralised before they become constraints. Leave these choices unmade and the ELM stays a governance intention. Make them and it becomes a strategic capability: liquidity stops being a ratio and becomes an option — and the option becomes the power to act. Seventh item on the table.
The price of silence
The risk is already live, and it is silent. It does not show up in the ratios; it shows up in the decisions a bank can no longer take, or delays, or takes at a cost it has never made visible. The next stress will not ask how much HQLA sat in the buffer at the previous night’s close; it will reveal how much of it could become cash, by which route, under which dependencies, and how quickly.
Banks know how to measure, how to produce the ratios and how to survive crises. They are less good at arbitrating a cost, spotting a slow erosion, assigning a responsibility, or correcting a structural inefficiency. The difficulty owes less to competence than to the legacy of 15 years of reform — each one justified on its own terms, the accumulation rather less so.
The stock grew while its use narrowed. Buffers have never been larger, nor the room to mobilise them tighter; collateral has never been more abundant, nor less mobile, segregation, operational constraints, and fragmentation having turned velocity into the adjustment variable.
Measurement and governance have proliferated without delivering what was expected of them. Prudential indicators have multiplied faster than the understanding of the economic mechanisms they describe; committees have been added faster than responsibilities have been assigned, to the point that many liquidity costs still have no owner at all. As for the layers of protection, they are paid for twice — once upfront and again every day, through friction. The sum of individual prudence ends up producing collective fragility: optimised in isolation, each institution withdraws from the system a little of the elasticity they all depend on the same morning. An uncomfortable rule emerges from all of this: the most expensive risks set off no alarm, and silent erosion outruns, in the end, the price of visible crises.
Several shifts are already shaping the decade ahead; the supervisory cycle, for its part, is annual, and it has already started.
The first is the shift from compliance to optimisation: meeting the ratios is an obligation, minimising their economic cost is the job. The second is to make hidden costs visible — those of fragmentation and structural inefficiency — and to assign them to somebody. The third puts in place an economic governance of liquidity, cross-functional by design: treasury, risk, finance, the repo businesses and infrastructures. The fourth erases boundaries that are already fading, those separating liquidity, solvency, capital, leverage, and the balance sheet — managed apart and consumed together. The fifth concerns the circulation of collateral, where speed now counts for more than the quantity available.
A polite market
On 31 July 2026 the supervisor measured what can be measured. The rest will appear in no supervisory report; it will appear, or not, in a set of minutes. We began with the least-read chapter of a public document; we end with the least-read document in a bank. Both take 10 minutes to read.
One box remains. One hundred and ten banks imagined their worst day, then listed what they would do to survive it; 1 in 100 planned to tighten its repo haircuts. The other 99 have therefore modelled a crisis in which their counterparties — which is to say themselves — would remain accommodating. The LCR barely moves. It all adds up.
NO FEE, NO RISK
100% ON RETURNS If you invest in only one securities finance news source this year, make sure it is your free subscription to Securities Finance Times
100% ON RETURNS If you invest in only one securities finance news source this year, make sure it is your free subscription to Securities Finance Times
.jpg)