A name on the door
04 August 2026
In the seventh instalment of this ongoing series, Cyril Louchtchay de Fleurian, head of securities finance and balance sheet strategy at Capteo: Strategy & Management Consulting, talks risk
Image: stock.adobe.com/JW Studio
The history of banking risk is a history of late baptisms. Every major risk has followed the same sequence, without exception: diffuse, then named, then given an owner, then governable. The title of chief risk officer did not exist before 1993, when GE Capital created it for James Lam. Until then, the firm’s aggregate risk belonged to everyone, which is to say to no one. 15 years later, no systemic bank operated without one. Operational risk met the same fate: before Barings collapsed in February 1995 — £827 million of concealed losses, more than twice the bank’s capital — it sat in the taxonomies under ‘other risks’. Basel II gave it a name in 2004, a capital charge, and owners.
Counterparty risk offers the most instructive precedent. The Basel Committee would later establish that roughly two-thirds of the counterparty losses of the 2008 crisis came from credit valuation adjustments (CVA) and only one-third from actual defaults: the risk lay in the price of risk, which nobody owned. The CVA charge of December 2010 and the spread of XVA desks followed, greeted by the objections one hears today word for word: yet another desk, yet another layer, yet another turf war. Conduct risk, finally: more than US$9 billion in fines for LIBOR manipulation alone, in the region of £50 billion in redress in the UK for the mis-selling of payment protection insurance (the PPI scandal), and in total more than US$300 billion in penalties for the world’s largest banks between 2009 and 2016. The UK’s Senior Managers Regime drew the organisational conclusion in 2016 by attaching every risk to a name.
Each time, what looked like added complexity turned out to be a simplification: the replacement of implicit, scattered trade-offs with explicit accountability. Executable liquidity is the next risk on that list, for the reasons set out in the first six articles of this series. History suggests that the lag between the revealing shock and the organisational response is measured in years — between two and nine, depending on the case; it also suggests that the lag has never protected those who used it to wait.
Three myths and a funeral
Three solutions, all of them expected, all of them heard before — three myths risk departments tell themselves about the high-quality liquid asset (HQLA) buffer. First, more HQLA: the bank already has plenty, and recent cases show institutions wobbling with their buffers intact. Second, better models: the previous article established that measurement promises solvency, not execution. Third, stress tests: they exist — fragmented, unorchestrated, with no one to arbitrate between them; running more of them does not fix their main defect, which is that they belong to no one.
Liquidity is an execution capability, not a dormant insurance policy: it protects through its ability to be mobilised now, at a controlled cost and without friction, far more than through its nominal size. A buffer that is untested, unmonetised, and disconnected from market flows is an illusion of safety. The structuring rule fits in one number: 50–60 per cent of the buffer monetised in business-as-usual, in a controlled and reversible manner. It serves three purposes: continuously testing whether assets can actually be mobilised, keeping the transformation channels open (repo, triparty, securities lending, central bank), and avoiding the cliff effect under stress. By the time the crisis arrives, it is already too late.
The old model rested on a sanctified buffer and ad hoc — and therefore late — activation, with stress treated as an exceptional event. The new model builds stress into normal operations, with scenarios modelled and tested before they are needed. A crisis becomes the acceleration of a regime already running, not a change of mode. The liquidity one freezes out of prudence is precisely the liquidity that will no longer move in a crisis.
Spending the buffer to keep it intact
Monetise 50–60 per cent of the buffer: the proposal makes any risk committee jump, and the objection always lands in the same terms — the ratio will not allow it. The objection describes a constraint, not an impossibility; the Liquidity Coverage Ratio (LCR) caps monetisation, it does not prohibit it, and the whole doctrine consists in mapping the gap between the cap and the use.
The LCR buffer is not a block, it is a living inventory, made up of assets that are regulatorily defined and operationally constrained. It breaks down into LCR-eligible HQLA (Level 1/Level 2A/2B): a portion that is genuinely available — unencumbered, free of any charge, mobilisable without friction; and a portion that is not — encumbered, already committed (triparty, CCP, pledges, limited rehypothecation and so on) or technically unavailable (cut-offs, contractual rights, concentration, prohibitive haircuts).
Alongside the LCR buffer sit other pockets of potential liquidity: central bank eligible assets that do not always qualify as HQLA for LCR purposes, and non-HQLA assets that can be monetised in the market but do not count towards the buffer. In other words, eligible does not mean usable, and usable does not mean LCR-countable. Within that frame, monetising the buffer amounts to proving that it is an execution capability and not merely a theoretical ratio.
The central constraint lies in the very mechanics of the ratio: the LCR rests on the effective availability of a stock of HQLA. Monetising the buffer credibly therefore means accepting a core of durably unencumbered HQLA, with any mobilisation beyond it remaining reversible at short notice to avoid any breaking point — a channel closing, a haircut jump, a missed cut-off.
Three rules frame that gap and define the acceptable space for putting the buffer to work. An untouchable floor: a minimum level of unencumbered HQLA held at all times (per currency where necessary), set ex ante and monitored intraday; this floor offers little room for optimisation. Reversibility inside T+1: any monetisation beyond the floor must be structured to unwind quickly (ideally before T+1), with known routes back — short maturities, substitution options, roll and term-out capacity, cut-offs under control. Channel diversification: monetisation never depends on a single channel, it is spread across several rails (central bank, CCP, triparty, bilateral repo), several counterparties, and several formats (overnight versus term), with concentration limits.
Monetising the buffer therefore means converting, in a controlled way, a fraction of the liquid asset stock (HQLA and central bank eligible) held as a liquidity reserve into usable funding capacity (cash or better collateral) through secured transactions — repo and reverse repo, triparty, CCP-cleared, central bank pledges, collateral upgrades — without degrading the regulatory and operational availability of the buffer.
In practice, it is the dynamic management of the pair ‘unencumbered HQLA versus secured funding capacity’. One temporarily pledges securities from the buffer to raise cash (or to free up better-quality collateral), while keeping control of haircuts and their convexity under stress, maturities and reversibility (roll and term-out, substitution), settlement and delivery cut-offs and frictions, channel concentration (central bank versus CCP versus triparty versus bilateral repo) and the level of encumbrance created — all while preserving a floor of HQLA that remains durably unencumbered and strictly ring-fenced.
Liquid — at least on paper
In a regime where liquidity is conditional, liquidity only exists if the collateral is pre-positioned. An asset that is eligible but not pre-positioned is economically liquid and operationally useless. Pre-positioning is not a nice-to-have; it is a structuring requirement — measurable and governable.
Pre-positioning comes down to one picture: before any stress, the assets are already lodged with, or technically mobilisable at, the key infrastructures — central banks, CCPs, triparty agents, major counterparties. The legal rights, the documentation, the operational chains, and the payment and collateral routings are activated and tested. The bank can convert those assets into liquidity within a measured, guaranteed timeframe. Pre-positioning turns a potential stock into immediate capacity to act. Recent episodes — Credit Suisse in particular — show a bank holding eligible assets in volume, but not in the right place, not at the right time, not with the correct rights activated. The loss of control was caused neither by a shortage of collateral nor by a regulatory breach, but by an inability to mobilise quickly assets that were, on paper, perfectly liquid. Pre-positioning is the tipping point between compliance and executability.
A liquidity buffer counts as executable only to the extent that it is pre-positioned, tested, and mobilisable through an identified channel. The principle implies a shift of paradigm: one does not manage a buffer by asset class, but by effective conversion channels. Pre-positioning rests on three pillars. Pillar 1 — pre-positioning by infrastructure: the buffer must be explicitly broken down by mobilisation channel (central banks, per currency, CCPs, triparty agents, custodian). Each asset carries a primary channel, a fallback channel and a measured conversion time (time-to-cash). Pillar 2 — legal and operational activation: an asset counts as pre-positioned only if the documentation is in place and valid, the accounts and links are open and operational, and the teams know how to execute without exceptional escalation. Pre-positioning is not a static legal exercise; it is a living operational capability. Pillar 3 — regular testability: pre-positioning must be tested in normal conditions, as business-as-usual practice, not discovered under stress. That means periodic access tests by channel, intraday mobilisation drills, and live exercises in beating the cut-offs.
The doctrine translates into hard numbers: the share of the central bank eligible buffer actually pre-positioned, per currency; the share of the buffer mobilisable within a set number of hours, per channel; median and worst-case time-to-cash per infrastructure; the success rate of monthly mobilisation tests; the share of the buffer resting on a single channel. These thresholds belong on the ExCo agenda, not in a technical committee.
The natural owner of the doctrine arbitrates the allocation of the buffer across channels, decides what pre-positioning may acceptably cost in business-as-usual, prioritises the tests and the critical channels, and reports on the liquidity that is genuinely mobilisable, not merely eligible. Without a mandate, pre-positioning remains fragmented, sub-optimised, and politically unowned. Failing to formalise pre-positioning as core doctrine means accepting that the bank will discover under stress which assets are actually usable, and will take whatever trade-offs the infrastructures impose.
Conversely, raising pre-positioning to a measurable capability turns liquidity into a sovereign execution capacity, drastically reduces the risk of forced liquidations, and makes governance a risk that is implicit today. Liquidity cannot be improvised at the moment it is required. It is prepared through pre-positioning, or it is lost by default.
Pre-positioning is also the only operational answer to what the previous article called the convergence of unavailabilities. Whether an asset is frozen by a sanction, immobilised by a custodian outage, or trapped behind a missed cut-off, the balance sheet effect is identical as is the defence. One cannot hedge a sanction, one cannot hedge an outage, one cannot hedge ransomware. But collateral that is already positioned, documented, and tested across several channels reduces the surface of every one of these unavailabilities at once, with no need to predict which will materialise. That is the whole value of the arrangement: it protects against the cause you did not see coming.
The empty chair
Who in the bank today holds both the mandate and the right to put the buffer to work daily, across every channel? As a matter of fact, nobody.
The mandate fits in one sentence: a responsibility for coordination, arbitration, and execution, cutting across Treasury, repo, asset and liability management (ALM), the scarce resources desk, and risk. Clear, enforceable, and without an equivalent anywhere on today’s organisation chart. Cross-functional liquidity execution does not create value out of thin air, and it does not aim to eliminate costs through aggressive optimisation. It makes them visible, arbitrable, and reversible; it turns silent, suffered losses into explicit, governed decisions. It does not add return on equity (ROE). It prevents several points of ROE from vanishing without anyone ever having attributed them.
The responsibilities fall into five core domains. Orchestrating the monetisation of the buffer: setting and steering the monetisation rate in business-as-usual, choosing the monetisation channels (bilateral repo, triparty, central bank and so on), managing substitutions, haircuts, and exclusions. Steering the kinetics of liquidity: measuring the speed of conversion from asset to cash, identifying operational bottlenecks, testing stress scenarios regularly. Monitoring the liquidity, balance sheet, and cost trade-offs: challenging the desks’ use of the balance sheet, turning away low-risk, high-cost volumes, exposing the silent destruction of ROE, demonstrating a recurring capacity to create value. Governing access to infrastructures: mapping dependencies on CCPs (concentration, and anticipating the cost of de-netting), triparty agents and central banks, securing critical access, anticipating the risks of exclusion or fragmentation. Coordinating the interface with risk and ALM: feeding the liquidity risk appetite framework with operational metrics, translating stress tests into executable decisions, keeping the regulatory LCR and the economic LCR consistent.
The mandate does not add risk, it reduces execution risk. It turns a risk that is currently implicit and endured into one that is explicit and managed. A monetised buffer reduces the risk of a sudden break; cross-functional accountability prevents liquidity from being captured by any single business line; the LCR is optimised at group level, not as the sum of local optimisations. Securing liquidity does not mean freezing it, it means keeping it permanently at work under control, so that a crisis is merely a change of intensity, not a leap into the unknown.
Such a mandate remains rare in continental Europe, because it rejects the illusion that safety is secured by the level of cash alone, accepts monetising the buffer in business-as-usual — a cultural taboo — treats liquidity as a flow rather than a stock, withdraws the implicit control of liquidity from the trading desks, and assumes an active management run on economic grounds. It breaks with the logic of the ring-fenced buffer, the confusion between liquidity and compliance, and the reactive management of stress.
The 3am phone call
The supervisor calls on a Thursday, late in the day, because a rumour is circulating and the deposits are moving. He asks three simple questions: how much collateral can be mobilised before 09:00 tomorrow, through which channels, and who decides. In most banks, the honest answer to the third question triggers a conference call with eight participants. That conference call is precisely what the mandate abolishes.
Today’s prudential frameworks are designed to assess the financial soundness of institutions under standardised conditions and, where necessary, to frame the management of stress situations. When funding spikes materialise, however, the dialogue with the supervisor can harden, and it no longer bears solely on compliance with the ratios but on the bank’s operational ability to work its liquidity levers in real time.
In those phases, the absence of a clearly identified point of accountability weakens the institution’s position. Fragmented information, a multiplicity of contacts and the difficulty of producing a consolidated, up-to-date view of executable liquidity expose the bank to unilateral prudential decisions, taken out of caution, for want of sufficient visibility.
The liquidity execution mandate professionalises that dialogue. It gives the supervisor a single counterpart, legitimate and technically equipped, able to demonstrate at any moment the effective mobilisation of the buffers, the kinetics of converting assets into liquidity, the dependencies on critical infrastructures, and the trade-offs made between liquidity, balance sheet, and operational continuity.
The objective is not to challenge the supervisor, but to reduce information asymmetry which, under stress, often leads to uniform, undifferentiated precautionary measures. By demonstrating command of its execution levers, the bank strengthens the credibility of its trajectories and retains negotiating room on the sequencing and intensity of prudential measures. In practice, a bank with such an identified responsibility enters a period of stress with a clear, documented view of its executable liquidity, decisions already taken, documented and traceable, and the ability to explain, in real time, the choices made and their impact.
That posture changes the nature of the dialogue: the supervisor no longer faces a reactive, fragmented institution, but an organisation able to actively steer its liquidity under constraint. The responsibility does not substitute for prudential requirements, it secures their operational application, turning the crisis dialogue into a structured exchange between accountable peers. The supervisor still decides; the mandate determines on what basis the supervisor decides.
Rights, not reporting lines
Granting a liquidity execution mandate raises an architectural question the industry almost always frames badly: should one create a desk with direct execution capabilities, or a cross-functional orchestration function with operational rights? The question is badly framed because it is binary; the real variable is the level of rights the organisation is prepared to transfer. An orchestrator without rights recommends, and an organisation that recommends under stress reverts to its reflexes, a desk without a cross-functional mandate cannibalises and recreates the very problem it was meant to solve. Everything else — the strengths and weaknesses of both models — flows from that single variable.
Figure 1
The desk model pushes the rights to their maximum, and its strengths are real. It hard-wires accountability in the full sense: one owner, one chain of command, results measurable in time-to-cash, in fails, in funding costs, in encumbrance, all the way to the P&L. It brings speed where speed matters: under stress, a desk acts instead of recommending, arbitrates on the spot, catches the cut-offs and clears settlement frictions while others would be scheduling a meeting. It makes optimisation genuinely actionable, because when optimisation requires trading — substitutions, switches, cleared versus bilateral arbitrage — only the one who executes controls the last mile. And it carries weight externally: facing CCPs, triparty agents, and market peers, a desk is an identified operational centre of gravity, which no committee will ever be.
Those strengths come at a price, and it is a steep one. A desk poorly delineated from repo, ALM and Trading guarantees turf wars, diluted accountability and endless political horse-trading. It requires the full apparatus of a front office — positions, limits, integration into the market risk framework, a control environment covering VaR, stress, limits, and model risk — which makes it heavy and slow to build. It creates an alignment risk: the moment a desk carries a P&L with a performance target, one introduces incentives potentially adverse to the stability mandate that justified its creation. It threatens, finally, to duplicate existing infrastructure — booking, confirmations, settlement, controls — rebuilding instead of orchestrating amounts to adding complexity in the name of reducing it.
The orchestrator model inverts the trade-off. It governs without cannibalising: it sets the rules, the priorities, and the escalation thresholds, and steers the action of the existing desks through defined rights — playbooks, runbooks, accountability matrices. It is light on balance sheet and capital, carrying little or no position of its own and consuming no limits. It deploys fast — on data, decisions, and escalations — without hiring a full front office or rebuilding the front-to-back chain. And it is natively scalable, since it operates as a system across repo, ALM, collateral, operations, and risk, standardising as it extends.
Its weaknesses are the exact mirror of its lightness. Without a dedicated team, day-to-day steering falls apart. Without enforceable rights — pre-agreements, delegations, triggers — it merely recommends, and the organisation goes back to its habits at the first sign of stress. It leaves open the awkward question: who executes when everything tightens? Depending on another desk to act can cost a cut-off and turn a correct decision into an operational failure. It exposes the bank to the worst scenario of all — diffuse accountability, collective and therefore unattributable, with everyone entitled to assume the decision belonged to another function. And it rests entirely on impeccable data, orchestrating without an intraday golden source covering positions, encumbrance, eligibility, haircuts, and cut-offs amounts to flying an instrument whose dials are masked.
The robust architecture therefore combines the two, and the combination is nothing like a soft compromise: an orchestrator by default, holder of the rules and of the consolidated view, endowed with a limited, conditional execution power, activated on defined triggers. The rights are calibrated in peacetime and exercised in wartime. It is the level of those rights, and that alone, which will separate real mandates from decorative ones.
The next baptism
Every risk in this story waited for its baptism: aggregate risk until 1993, the price of counterparty risk until 2010, conduct until 2016. Executable liquidity is waiting for its own. The objections standing in the way are familiar — yet another desk, yet another layer, yet another turf war; they are, word for word, the ones that greeted the XVA desks, and history has delivered its verdict on their foresight.
That leaves the question of timing. Banking rarely reforms in anticipation; it needs its fallen angels, and the next one will appear where nobody is looking — in a missed cut-off, a piece of collateral that will not move, a payment queue growing longer. Yet nothing obliges anyone to wait for it. Everything the mandate requires already exists in every bank: the assets, the channels, the teams, the infrastructures. All that is missing is a name on the door. The mandate has a name: Enterprise Liquidity Management (ELM). The next article will describe how it works. One can always rely on banks to give a risk an owner, once every cheaper alternative has been exhausted.
Counterparty risk offers the most instructive precedent. The Basel Committee would later establish that roughly two-thirds of the counterparty losses of the 2008 crisis came from credit valuation adjustments (CVA) and only one-third from actual defaults: the risk lay in the price of risk, which nobody owned. The CVA charge of December 2010 and the spread of XVA desks followed, greeted by the objections one hears today word for word: yet another desk, yet another layer, yet another turf war. Conduct risk, finally: more than US$9 billion in fines for LIBOR manipulation alone, in the region of £50 billion in redress in the UK for the mis-selling of payment protection insurance (the PPI scandal), and in total more than US$300 billion in penalties for the world’s largest banks between 2009 and 2016. The UK’s Senior Managers Regime drew the organisational conclusion in 2016 by attaching every risk to a name.
Each time, what looked like added complexity turned out to be a simplification: the replacement of implicit, scattered trade-offs with explicit accountability. Executable liquidity is the next risk on that list, for the reasons set out in the first six articles of this series. History suggests that the lag between the revealing shock and the organisational response is measured in years — between two and nine, depending on the case; it also suggests that the lag has never protected those who used it to wait.
Three myths and a funeral
Three solutions, all of them expected, all of them heard before — three myths risk departments tell themselves about the high-quality liquid asset (HQLA) buffer. First, more HQLA: the bank already has plenty, and recent cases show institutions wobbling with their buffers intact. Second, better models: the previous article established that measurement promises solvency, not execution. Third, stress tests: they exist — fragmented, unorchestrated, with no one to arbitrate between them; running more of them does not fix their main defect, which is that they belong to no one.
Liquidity is an execution capability, not a dormant insurance policy: it protects through its ability to be mobilised now, at a controlled cost and without friction, far more than through its nominal size. A buffer that is untested, unmonetised, and disconnected from market flows is an illusion of safety. The structuring rule fits in one number: 50–60 per cent of the buffer monetised in business-as-usual, in a controlled and reversible manner. It serves three purposes: continuously testing whether assets can actually be mobilised, keeping the transformation channels open (repo, triparty, securities lending, central bank), and avoiding the cliff effect under stress. By the time the crisis arrives, it is already too late.
The old model rested on a sanctified buffer and ad hoc — and therefore late — activation, with stress treated as an exceptional event. The new model builds stress into normal operations, with scenarios modelled and tested before they are needed. A crisis becomes the acceleration of a regime already running, not a change of mode. The liquidity one freezes out of prudence is precisely the liquidity that will no longer move in a crisis.
Spending the buffer to keep it intact
Monetise 50–60 per cent of the buffer: the proposal makes any risk committee jump, and the objection always lands in the same terms — the ratio will not allow it. The objection describes a constraint, not an impossibility; the Liquidity Coverage Ratio (LCR) caps monetisation, it does not prohibit it, and the whole doctrine consists in mapping the gap between the cap and the use.
The LCR buffer is not a block, it is a living inventory, made up of assets that are regulatorily defined and operationally constrained. It breaks down into LCR-eligible HQLA (Level 1/Level 2A/2B): a portion that is genuinely available — unencumbered, free of any charge, mobilisable without friction; and a portion that is not — encumbered, already committed (triparty, CCP, pledges, limited rehypothecation and so on) or technically unavailable (cut-offs, contractual rights, concentration, prohibitive haircuts).
Alongside the LCR buffer sit other pockets of potential liquidity: central bank eligible assets that do not always qualify as HQLA for LCR purposes, and non-HQLA assets that can be monetised in the market but do not count towards the buffer. In other words, eligible does not mean usable, and usable does not mean LCR-countable. Within that frame, monetising the buffer amounts to proving that it is an execution capability and not merely a theoretical ratio.
The central constraint lies in the very mechanics of the ratio: the LCR rests on the effective availability of a stock of HQLA. Monetising the buffer credibly therefore means accepting a core of durably unencumbered HQLA, with any mobilisation beyond it remaining reversible at short notice to avoid any breaking point — a channel closing, a haircut jump, a missed cut-off.
Three rules frame that gap and define the acceptable space for putting the buffer to work. An untouchable floor: a minimum level of unencumbered HQLA held at all times (per currency where necessary), set ex ante and monitored intraday; this floor offers little room for optimisation. Reversibility inside T+1: any monetisation beyond the floor must be structured to unwind quickly (ideally before T+1), with known routes back — short maturities, substitution options, roll and term-out capacity, cut-offs under control. Channel diversification: monetisation never depends on a single channel, it is spread across several rails (central bank, CCP, triparty, bilateral repo), several counterparties, and several formats (overnight versus term), with concentration limits.
Monetising the buffer therefore means converting, in a controlled way, a fraction of the liquid asset stock (HQLA and central bank eligible) held as a liquidity reserve into usable funding capacity (cash or better collateral) through secured transactions — repo and reverse repo, triparty, CCP-cleared, central bank pledges, collateral upgrades — without degrading the regulatory and operational availability of the buffer.
In practice, it is the dynamic management of the pair ‘unencumbered HQLA versus secured funding capacity’. One temporarily pledges securities from the buffer to raise cash (or to free up better-quality collateral), while keeping control of haircuts and their convexity under stress, maturities and reversibility (roll and term-out, substitution), settlement and delivery cut-offs and frictions, channel concentration (central bank versus CCP versus triparty versus bilateral repo) and the level of encumbrance created — all while preserving a floor of HQLA that remains durably unencumbered and strictly ring-fenced.
Liquid — at least on paper
In a regime where liquidity is conditional, liquidity only exists if the collateral is pre-positioned. An asset that is eligible but not pre-positioned is economically liquid and operationally useless. Pre-positioning is not a nice-to-have; it is a structuring requirement — measurable and governable.
Pre-positioning comes down to one picture: before any stress, the assets are already lodged with, or technically mobilisable at, the key infrastructures — central banks, CCPs, triparty agents, major counterparties. The legal rights, the documentation, the operational chains, and the payment and collateral routings are activated and tested. The bank can convert those assets into liquidity within a measured, guaranteed timeframe. Pre-positioning turns a potential stock into immediate capacity to act. Recent episodes — Credit Suisse in particular — show a bank holding eligible assets in volume, but not in the right place, not at the right time, not with the correct rights activated. The loss of control was caused neither by a shortage of collateral nor by a regulatory breach, but by an inability to mobilise quickly assets that were, on paper, perfectly liquid. Pre-positioning is the tipping point between compliance and executability.
A liquidity buffer counts as executable only to the extent that it is pre-positioned, tested, and mobilisable through an identified channel. The principle implies a shift of paradigm: one does not manage a buffer by asset class, but by effective conversion channels. Pre-positioning rests on three pillars. Pillar 1 — pre-positioning by infrastructure: the buffer must be explicitly broken down by mobilisation channel (central banks, per currency, CCPs, triparty agents, custodian). Each asset carries a primary channel, a fallback channel and a measured conversion time (time-to-cash). Pillar 2 — legal and operational activation: an asset counts as pre-positioned only if the documentation is in place and valid, the accounts and links are open and operational, and the teams know how to execute without exceptional escalation. Pre-positioning is not a static legal exercise; it is a living operational capability. Pillar 3 — regular testability: pre-positioning must be tested in normal conditions, as business-as-usual practice, not discovered under stress. That means periodic access tests by channel, intraday mobilisation drills, and live exercises in beating the cut-offs.
The doctrine translates into hard numbers: the share of the central bank eligible buffer actually pre-positioned, per currency; the share of the buffer mobilisable within a set number of hours, per channel; median and worst-case time-to-cash per infrastructure; the success rate of monthly mobilisation tests; the share of the buffer resting on a single channel. These thresholds belong on the ExCo agenda, not in a technical committee.
The natural owner of the doctrine arbitrates the allocation of the buffer across channels, decides what pre-positioning may acceptably cost in business-as-usual, prioritises the tests and the critical channels, and reports on the liquidity that is genuinely mobilisable, not merely eligible. Without a mandate, pre-positioning remains fragmented, sub-optimised, and politically unowned. Failing to formalise pre-positioning as core doctrine means accepting that the bank will discover under stress which assets are actually usable, and will take whatever trade-offs the infrastructures impose.
Conversely, raising pre-positioning to a measurable capability turns liquidity into a sovereign execution capacity, drastically reduces the risk of forced liquidations, and makes governance a risk that is implicit today. Liquidity cannot be improvised at the moment it is required. It is prepared through pre-positioning, or it is lost by default.
Pre-positioning is also the only operational answer to what the previous article called the convergence of unavailabilities. Whether an asset is frozen by a sanction, immobilised by a custodian outage, or trapped behind a missed cut-off, the balance sheet effect is identical as is the defence. One cannot hedge a sanction, one cannot hedge an outage, one cannot hedge ransomware. But collateral that is already positioned, documented, and tested across several channels reduces the surface of every one of these unavailabilities at once, with no need to predict which will materialise. That is the whole value of the arrangement: it protects against the cause you did not see coming.
The empty chair
Who in the bank today holds both the mandate and the right to put the buffer to work daily, across every channel? As a matter of fact, nobody.
The mandate fits in one sentence: a responsibility for coordination, arbitration, and execution, cutting across Treasury, repo, asset and liability management (ALM), the scarce resources desk, and risk. Clear, enforceable, and without an equivalent anywhere on today’s organisation chart. Cross-functional liquidity execution does not create value out of thin air, and it does not aim to eliminate costs through aggressive optimisation. It makes them visible, arbitrable, and reversible; it turns silent, suffered losses into explicit, governed decisions. It does not add return on equity (ROE). It prevents several points of ROE from vanishing without anyone ever having attributed them.
The responsibilities fall into five core domains. Orchestrating the monetisation of the buffer: setting and steering the monetisation rate in business-as-usual, choosing the monetisation channels (bilateral repo, triparty, central bank and so on), managing substitutions, haircuts, and exclusions. Steering the kinetics of liquidity: measuring the speed of conversion from asset to cash, identifying operational bottlenecks, testing stress scenarios regularly. Monitoring the liquidity, balance sheet, and cost trade-offs: challenging the desks’ use of the balance sheet, turning away low-risk, high-cost volumes, exposing the silent destruction of ROE, demonstrating a recurring capacity to create value. Governing access to infrastructures: mapping dependencies on CCPs (concentration, and anticipating the cost of de-netting), triparty agents and central banks, securing critical access, anticipating the risks of exclusion or fragmentation. Coordinating the interface with risk and ALM: feeding the liquidity risk appetite framework with operational metrics, translating stress tests into executable decisions, keeping the regulatory LCR and the economic LCR consistent.
The mandate does not add risk, it reduces execution risk. It turns a risk that is currently implicit and endured into one that is explicit and managed. A monetised buffer reduces the risk of a sudden break; cross-functional accountability prevents liquidity from being captured by any single business line; the LCR is optimised at group level, not as the sum of local optimisations. Securing liquidity does not mean freezing it, it means keeping it permanently at work under control, so that a crisis is merely a change of intensity, not a leap into the unknown.
Such a mandate remains rare in continental Europe, because it rejects the illusion that safety is secured by the level of cash alone, accepts monetising the buffer in business-as-usual — a cultural taboo — treats liquidity as a flow rather than a stock, withdraws the implicit control of liquidity from the trading desks, and assumes an active management run on economic grounds. It breaks with the logic of the ring-fenced buffer, the confusion between liquidity and compliance, and the reactive management of stress.
The 3am phone call
The supervisor calls on a Thursday, late in the day, because a rumour is circulating and the deposits are moving. He asks three simple questions: how much collateral can be mobilised before 09:00 tomorrow, through which channels, and who decides. In most banks, the honest answer to the third question triggers a conference call with eight participants. That conference call is precisely what the mandate abolishes.
Today’s prudential frameworks are designed to assess the financial soundness of institutions under standardised conditions and, where necessary, to frame the management of stress situations. When funding spikes materialise, however, the dialogue with the supervisor can harden, and it no longer bears solely on compliance with the ratios but on the bank’s operational ability to work its liquidity levers in real time.
In those phases, the absence of a clearly identified point of accountability weakens the institution’s position. Fragmented information, a multiplicity of contacts and the difficulty of producing a consolidated, up-to-date view of executable liquidity expose the bank to unilateral prudential decisions, taken out of caution, for want of sufficient visibility.
The liquidity execution mandate professionalises that dialogue. It gives the supervisor a single counterpart, legitimate and technically equipped, able to demonstrate at any moment the effective mobilisation of the buffers, the kinetics of converting assets into liquidity, the dependencies on critical infrastructures, and the trade-offs made between liquidity, balance sheet, and operational continuity.
The objective is not to challenge the supervisor, but to reduce information asymmetry which, under stress, often leads to uniform, undifferentiated precautionary measures. By demonstrating command of its execution levers, the bank strengthens the credibility of its trajectories and retains negotiating room on the sequencing and intensity of prudential measures. In practice, a bank with such an identified responsibility enters a period of stress with a clear, documented view of its executable liquidity, decisions already taken, documented and traceable, and the ability to explain, in real time, the choices made and their impact.
That posture changes the nature of the dialogue: the supervisor no longer faces a reactive, fragmented institution, but an organisation able to actively steer its liquidity under constraint. The responsibility does not substitute for prudential requirements, it secures their operational application, turning the crisis dialogue into a structured exchange between accountable peers. The supervisor still decides; the mandate determines on what basis the supervisor decides.
Rights, not reporting lines
Granting a liquidity execution mandate raises an architectural question the industry almost always frames badly: should one create a desk with direct execution capabilities, or a cross-functional orchestration function with operational rights? The question is badly framed because it is binary; the real variable is the level of rights the organisation is prepared to transfer. An orchestrator without rights recommends, and an organisation that recommends under stress reverts to its reflexes, a desk without a cross-functional mandate cannibalises and recreates the very problem it was meant to solve. Everything else — the strengths and weaknesses of both models — flows from that single variable.
Figure 1
The desk model pushes the rights to their maximum, and its strengths are real. It hard-wires accountability in the full sense: one owner, one chain of command, results measurable in time-to-cash, in fails, in funding costs, in encumbrance, all the way to the P&L. It brings speed where speed matters: under stress, a desk acts instead of recommending, arbitrates on the spot, catches the cut-offs and clears settlement frictions while others would be scheduling a meeting. It makes optimisation genuinely actionable, because when optimisation requires trading — substitutions, switches, cleared versus bilateral arbitrage — only the one who executes controls the last mile. And it carries weight externally: facing CCPs, triparty agents, and market peers, a desk is an identified operational centre of gravity, which no committee will ever be.
Those strengths come at a price, and it is a steep one. A desk poorly delineated from repo, ALM and Trading guarantees turf wars, diluted accountability and endless political horse-trading. It requires the full apparatus of a front office — positions, limits, integration into the market risk framework, a control environment covering VaR, stress, limits, and model risk — which makes it heavy and slow to build. It creates an alignment risk: the moment a desk carries a P&L with a performance target, one introduces incentives potentially adverse to the stability mandate that justified its creation. It threatens, finally, to duplicate existing infrastructure — booking, confirmations, settlement, controls — rebuilding instead of orchestrating amounts to adding complexity in the name of reducing it.
The orchestrator model inverts the trade-off. It governs without cannibalising: it sets the rules, the priorities, and the escalation thresholds, and steers the action of the existing desks through defined rights — playbooks, runbooks, accountability matrices. It is light on balance sheet and capital, carrying little or no position of its own and consuming no limits. It deploys fast — on data, decisions, and escalations — without hiring a full front office or rebuilding the front-to-back chain. And it is natively scalable, since it operates as a system across repo, ALM, collateral, operations, and risk, standardising as it extends.
Its weaknesses are the exact mirror of its lightness. Without a dedicated team, day-to-day steering falls apart. Without enforceable rights — pre-agreements, delegations, triggers — it merely recommends, and the organisation goes back to its habits at the first sign of stress. It leaves open the awkward question: who executes when everything tightens? Depending on another desk to act can cost a cut-off and turn a correct decision into an operational failure. It exposes the bank to the worst scenario of all — diffuse accountability, collective and therefore unattributable, with everyone entitled to assume the decision belonged to another function. And it rests entirely on impeccable data, orchestrating without an intraday golden source covering positions, encumbrance, eligibility, haircuts, and cut-offs amounts to flying an instrument whose dials are masked.
The robust architecture therefore combines the two, and the combination is nothing like a soft compromise: an orchestrator by default, holder of the rules and of the consolidated view, endowed with a limited, conditional execution power, activated on defined triggers. The rights are calibrated in peacetime and exercised in wartime. It is the level of those rights, and that alone, which will separate real mandates from decorative ones.
The next baptism
Every risk in this story waited for its baptism: aggregate risk until 1993, the price of counterparty risk until 2010, conduct until 2016. Executable liquidity is waiting for its own. The objections standing in the way are familiar — yet another desk, yet another layer, yet another turf war; they are, word for word, the ones that greeted the XVA desks, and history has delivered its verdict on their foresight.
That leaves the question of timing. Banking rarely reforms in anticipation; it needs its fallen angels, and the next one will appear where nobody is looking — in a missed cut-off, a piece of collateral that will not move, a payment queue growing longer. Yet nothing obliges anyone to wait for it. Everything the mandate requires already exists in every bank: the assets, the channels, the teams, the infrastructures. All that is missing is a name on the door. The mandate has a name: Enterprise Liquidity Management (ELM). The next article will describe how it works. One can always rely on banks to give a risk an owner, once every cheaper alternative has been exhausted.
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 Â鶹ӰÊÓ´«Ã½ 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 Â鶹ӰÊÓ´«Ã½ Times
