The role of holding limits for sterling-denominated systemic stablecoins and a potential digital pound

Executive summary

The Bank of England’s (Bank’s) approach to new forms of digital money aims to support innovation in payments while preserving the integrity and stability of the UK monetary system. This paper explores the key financial stability (FS) considerations arising from sterling-denominated systemic stablecoins (SSCs)^{footnote [2]} and a potential digital pound (collectively referred to as ‘digital money’ in this paper) and assesses how holding limits could address these risks.

The Bank has today published a consultation paper (CP) on the proposed regulatory regime for SSCs and is in the ‘Design Phase’ of the potential digital pound. SSCs would be privately issued by non-bank firms operating under the Bank’s proposed regime for systemic payment systems using stablecoins and a potential digital pound would be issued by the Bank. SSCs would be fully backed by a mix of central bank deposits and sterling-denominated UK government debt securities. The decision on whether or not to launch a digital pound and its timing will be taken in due course.

There is a risk of a reduced lending to businesses and households as the economy adjusts to these new forms of money. Currently in the UK, lending is largely supported by issuance of short-term deposits held in banks,^{footnote [3]} often in support of payments activities. Introduction of digital money could pose the following challenges for the real economy: (i) growth in digital money could lead to a greater decoupling of payments and credit, with potential to both increase the cost of credit and lower the availability of credit; and (ii) in a stress, deposit outflows from banks could impair their ability to lend when it may be needed the most. In particular, a disorderly transition to widespread adoption of digital money could pose risks to provision of credit to the UK economy. In contrast to the US, where capital markets play a larger role, UK households and businesses continue to rely heavily on the banking sector for credit provision. Significant and rapid outflows of bank deposits into digital money could therefore lead to potential reduction in credit for UK businesses and households if the banking system were unable to increase, at scale and at pace, its use of wholesale financing from non-banks.^{footnote [4]}

This paper sets out a methodological approach to quantify this risk. As noted in the Bank’s 2021 discussion paper on New Forms of Digital Money, the existence of digital money could result in households and businesses switching some of their commercial bank deposits to these digital monies, especially during a banking stress, due to their cash-like properties and perceived safety relative to commercial bank deposits.^{footnote [5]} The loss of deposits for commercial banks is known as ‘bank disintermediation’ and, depending on the speed and scale, could have implications for financial stability and growth. There are two scenarios in which disintermediation could take place: (i) in the transition to steady state; and (ii) in a stress scenario. This paper focuses on the risk of disintermediation in a stress scenario and outlines a methodology for quantifying this risk.

Recent academic literature suggests that holding limits can reduce this risk, and other central banks are considering similar risks. For example, BIS (2024)^{footnote [6]} highlighted their role in balancing the innovation benefits from digital money with prudential safeguards, and a study conducted by ECB (2023)^{footnote [7]} outlines an approach similar to this paper for calibrating the holding limit for the proposed digital Euro. Previous publications, including the Bank’s 2021 discussion paper and its responses, have made a case for the role of holding limits to manage financial stability risks in periods when the financial system is adjusting to digital money. This paper extends that further to consider both the potential for distributional effects across the banking system and potential outflows from the banking system.

This paper explores the effectiveness of different levels of holding limits. We developed a hypothetical ‘severe illustrative stress’ scenario to explore the potential impact of sharp deposit outflows across different banks in the presence of digital money, with a view to understanding the efficacy of individual holding limits in mitigating this liquidity stress scenario.^{footnote [8]} The scenario was based on severe assumptions, including outflows from a group of UK banks, both insured and uninsured deposits, that are much higher than those seen in previous crises. The analysis focuses on three metrics: (i) the number of banks falling below 100% Liquidity Coverage Ratio (LCR);^{footnote [9]} (ii) relatedly, the scale of central bank lending that banks might demand; and (iii) a residual shortfall to a 100% LCR which may need to be met through eg withdrawing existing credit lines and monetisation of assets in private markets.^{footnote [10]} This is based on the assumption that banks do not take any pre-emptive mitigating actions in anticipation of the above risks. The model evaluates these outcomes, first with no limits, and then across different levels of holding limits, ranging from £5,000 to £20,000 for individuals and £1 million, £10 million and £100 million for businesses.

Without holding limits, or alternative policies that achieve the same outcome, the risk of a significant impact on lending to businesses and households could be high. In our hypothetical severe illustrative stress scenario, the number of firms at risk of low LCRs – indicative of potential risks of reduced bank lending to businesses and households – increases with the introduction of digital money, even at a low limit, which reflects the low average balance of the majority of deposits. The number of firms at risk increases as the individual limit is raised from £5,000 to £20,000, but more slowly, reflective of the distribution of deposit balances typically being less than £5,000. However, when there are no individual limits in place, the number of firms at risk of falling below 100% LCR is substantially higher. The potential demand for central bank lending and stressed firm behaviour increase proportionately over the same range, which suggests further potential for the risk of cuts to banking lending to the real economy or potential for sizeable asset sales as banks seek to monetise assets in private markets. If there were no individual limits, both would increase rapidly, with additional demand for central bank lending rising to c.£250 billion (from £112 billion for a £20,000 limit). The results under our hypothetical severe illustrative scenario suggest that the impact of digital money adoption is likely to vary across firms, with some facing a steeper adjustment curve – highlighting the importance of a well-managed transition to avoid disproportionate exposure.

However, limits could also reduce the ability of households and businesses to use digital money (‘usability’), reducing the ability to unlock their potential innovation benefits. This paper also highlights the potential implications of holding limits on usability. A lower limit may reduce financial stability (FS) risks but setting limits too low would lower the usability of digital money. This paper has considered the proportion of households who could receive regular income and retain rolling balances in their accounts (ie use of a digital money account as a bank current account equivalent). A higher limit could also allow for more high-value transactions, and a greater volume of transactions to benefit from the potential innovations offered by digital money.

As the financial system adapts to digital money, policymakers’ reliance on limits as a safeguard could reduce. The Bank expects that uncertainty around the potential FS risks from adoption of digital money would reduce as the financial system adapts. For example, over time, banks may choose to increase the amount of pre-positioned collateral (PPC) with the Bank, which could make it quicker for firms to access central bank lending in such circumstances. Under the Bank’s proposed regime for SSCs, limits would be in place only during transition, and the Bank would monitor their utilisation, impact and the remaining risk to inform the judgment on when to amend or remove limits, once it gains sufficient comfort that FS risks have been understood and suitably mitigated. The methodology outlined in this paper could continue to guide this judgement in the future.

In summary, this paper provides an analytical framework to understand the role of holding limits in mitigating the potential for bank disintermediation in stress, which has informed the Bank’s policy judgements on holding limits. The determination of the right level of holding limits for each of the form of digital money will involve policy judgements which extend beyond FS risks, such that the limits are set in a way that is both prudent and practical.

Section 1: Introduction

The Bank’s approach to new forms of digital money, including sterling-denominated SSCs and the potential digital pound aims to support innovation in payments while preserving the integrity and stability of the UK monetary system. This paper explores the key FS considerations arising from the SSCs and the potential digital pound (collectively referred to as ‘digital money’ in this paper).

In this paper, we consider a combination of digital money in a mixed money ecosystem where public and private monies coexist. This is in line with the Bank’s view that non-bank entities can play an important role in payments. SSCs would be privately issued, regulated by the Bank^{footnote [11]} and would be fully backed by a mix of central bank deposits and sterling-denominated UK government debt securities and the potential digital pound would be issued by the Bank. The Bank has today published a consultation paper on the proposed regulatory regime for SSCs and is in the ‘Design Phase’ of the potential digital pound. The decision on whether to launch a digital pound and its timing would be taken in due course.

As noted in the Bank’s 2021 discussion paper on New Forms of Digital Money, the existence of digital money could result in households and businesses switching some of their bank deposits to these digital monies, especially during a period of banking stress, due to their perceived safety relative to commercial bank deposits. The loss of deposits for commercial banks is known as ‘bank disintermediation’ and, depending on the speed and scale, could have FS implications, including lending to businesses and households.

This paper focuses exclusively on the risks from retail use cases of digital money, and does not address wholesale applications, which involve distinct operational and risk considerations.^{footnote [12]} FS risks could arise from two different types of disintermediation scenario:

1. Disintermediation of the banking sector in the transition to steady state. As digital money gains traction, deposits could migrate out of the banking system. The degree to which this decoupling of payments and credit takes place could reduce the availability of bank funding, potentially increasing the cost or reducing the quantity of banks’ credit provision over time, to the extent that banks have not yet adjusted their funding sources or business models. This could be particularly pronounced if it happens rapidly or is disorderly, which may be a greater risk in a period where individuals, businesses and financial institutions are adjusting to these new forms of money. Given the important role of the banking sector in providing credit to households and businesses in the UK, it will be crucial to ensure that the transition does not cause undue disruption to the supply of credit and support the real economy as it adjusts.
2. Disintermediation in a stress scenario. Both SSCs regulated under the proposed Bank regime and the potential digital pound, if launched, are intended to be new safe forms of money – with both sharing some properties of cash. Under the proposed Bank regime, SSCs would be backed by a combination of central bank money and short-term sterling-denominated UK government debt securities, and have access to central bank lending. Unlike the banking system, they will not undertake lending to the real economy. In previous stress scenarios, such as the global financial crisis, we observed depositors rapidly moving funds from some banks into other banks which were perceived to be safer, as well as into cash and other safe assets. In any future banking stress, digital money could provide additional perceived safe havens^{footnote [13]} for depositors to move into, potentially at speed. The banking sector could prove unprepared to withstand rapid large outflows of deposits and sterling money markets may also face disruptions – which could amplify stress with potential implications on availability and resilience of credit.

This paper focuses on the risk of disintermediation in a stress scenario and outlines a methodology for quantifying this risk.

Section 2 explores how the presence of digital money could increase financial stability risks if banks are disintermediated in a stress scenario. Whilst risk of liquidity stress for the banking sector already exists, the severity and likelihood could plausibly be affected by the existence of digital money if these risks are amplified by the systemic nature, perceived safety, and rapid scalability of digital money, especially in the context of retail payments.

In times of stress, their existence could potentially increase outflow rates for bank deposits if depositors migrate to these digital monies. This could exacerbate any stress on the banking system, with risks being more pronounced for uninsured deposits. All banks could see outflows and be less able to recycle liquidity than in an interbank stress. Together these factors could increase the risk of cuts to bank lending to the real economy, with potential for sizeable asset sales as banks seek to monetise assets in private markets. This also has implications for how much banks additionally seek to borrow from the Bank of England.

The degree of this additional risk is highly uncertain and will depend on the number and types of money that emerge, the scale and pace of changes in depositors’ preference over different forms of money, the response of banks and other non-bank lenders and wider economic factors at the time of transition.

The Bank has proposed holding limits for SSCs^{footnote [14]} and the potential digital pound^{footnote [15]} as a transitional safeguard, which would allow the Bank to learn more about the extent of FS risks. The Bank’s consultation paper for SSCs sets out that it would expect to loosen, and ultimately remove, such limits when it gains sufficient comfort that financial stability risks have been suitably understood and mitigated. The methodology outlined in this paper could be used to guide this judgement.

Policymaker judgement on the calibration of limits would need to be balanced against the ability of households and business to use digital money (‘usability’), which is important for unlocking their potential innovation benefits. A lower limit may reduce FS risks but setting limits too low would also lower the usability of digital money. To underscore the role of usability, we present analysis on the potential implications of holding limits on usability, which we measured as the ability of households and businesses to use digital money accounts to spend, receive and hold money.

In the academic literature, there are several papers which study the role of limits in containing financial stability risks (eg BIS (2024)), highlighting their role in balancing the innovation benefits from digital money with prudential safeguards. Other central banks are currently considering similar risks, for example a study conducted by European Central Bank (ECB) (2023) outlines an approach (which is similar to this paper) for calibrating the holding limit for the proposed digital Euro by looking at the potential impact of the introduction of the digital Euro across European banks. A survey on central banks exploring central bank digital currency (CBDC) conducted by BIS (2023) suggests that more than half of central banks are considering holding limits. Previous publications, including the Bank’s 2021 DP and its responses, have made a case for the role of holding limits to manage financial stability risks in periods when the financial system is adjusting to digital money. This paper extends that further to consider both the potential for distributional effects across the banking system and potential outflows from the banking system. Section 2 discusses our approach to modelling the potential for disintermediation under stress. Section 3 explains the role of holding limits in mitigating these risks and Section 4 provides details of the usability analysis. Finally, Section 5 outlines the potential policy implications of our analysis. 

Section 2: Modelling the potential for disintermediation in stress

This section explores how digital money could increase financial stability risks if there is sizeable bank disintermediation in a stress scenario.

We note that the 2021 DP also explored potential for disintermediation in stress. This paper extends that further to consider both the potential for distributional effects across the banking system as well as potential outflows from the banking system, ie we capture the possibility that some deposits will remain within the system as deposits move between banks. We also apply different outflow assumptions to the 2021 DP, including allowing some insured deposits to flow out of the banking system. The banking stress event in March 2023 (with separate shocks involving Credit Suisse, Silicon Valley Bank, First Republic and Signature Bank) highlighted the scale and pace at which stress events could take place with the advancement in digital technology. With the introduction of digital money, depositors could have additional options over safe assets to move their deposits in a stress. This prompted us to consider an illustrative scenario with a more severe stress as compared to March 2023. However, this should not be viewed as forecast, but a hypothetical tail event. This section includes further details on our key simplifications and assumptions.

Approach to quantifying FS risk

Though considered generally safe, Banks today already face (tail) risks from rapid deposit outflows, even in the absence of digital money. Prudential regulations require banks to maintain liquid assets (reserves and high-quality liquid assets - HQLA) against the risk of outflows of various types of deposits based on their perceived withdrawal risk levels. As such, uninsured deposits need to be backed by more liquid assets in comparison to insured deposits because they are more vulnerable to withdrawal risk. Banks are also eligible to use the Bank of England’s liquidity facilities under the Sterling Monetary Framework (SMF).

Digital money could increase households and businesses’ marginal propensity to withdraw funds. Under certain stress scenarios, these options may be perceived to be more attractive for general payments purposes than bank deposits, money market funds (MMFs), government savings (eg NS&I)^{footnote [16]} products, cross-border options, or cash.

In times of banking stress, the existence of digital money could potentially increase outflow rates for bank deposits as depositors migrate to these digital monies. The presence of digital money could increase the likelihood that deposits leave the banking system rather than flow from one bank to another. This could exacerbate the stress on the banking system, with risks being more pronounced for uninsured deposits. All banks could see outflows and be less able to recycle liquidity, with implications for the provision of bank lending to the real economy.

To explore this potential increase in FS risk, we have modelled a hypothetical ‘severe illustrative stress’ scenario. As noted above, this is intended to set out dynamics of interest and potential orders of magnitude under a hypothetical tail event, rather than act as a forecast.

We begin with a ‘baseline stress scenario’ guided by observed outflows during the banking stress event in March 2023 (Credit Suisse, Silicon Valley Bank, First Republic and Signature Bank), to determine a set of baseline outflow rates that proxy the liquidity risks faced by the banking sector in a severe stress scenario without new forms of digital money. We also assume that deposits also move within the banking system, and proxy this by assuming that depositors would shift deposits away from smaller, and towards larger banks.

We introduce three SSCs and a potential digital pound into our model by assuming an initial aggregate uptake (pre-stress) of digital money of approximately 15% of aggregate sight deposits for both individuals and businesses.^{footnote [17]} We then construct a hypothetical ‘severe illustrative scenario’ by increasing deposit outflow rates from banks to model the extent of bank disintermediation during stress. Our modelled outflows represent total outflows across all four coins.

The incremental severity in our scenario is driven by the modelled increase in outflow rates during stress due to the presence of digital money, detailed in Table 2.A. Our scenario assumes an outflow rate of 75% for individual uninsured deposits (100% for business deposits) and c.10% for individual insured deposits (c.40% for business deposits). For simplicity, we assume uniform depositors’ behaviour across digital money products in both the baseline and illustrative stress scenarios in that they do not prefer one type of product above another.^{footnote [18]}

In the next section, we introduce holding limits into the modelling. Where holding limits bind, total outflows to digital money are capped – and so in such cases, we assume that depositors would (consistent with our baseline scenario) instead move their deposits from smaller banks to larger ones. By design, the scenario has a greater impact on smaller banks than larger banks.

As shown in Table 2.A, the assumed uninsured outflow rates are higher than the worst idiosyncratic cases in the March 2023 baseline, and insured outflow rates are twice as large as LCR standards. This reflects a more significant stress than observed in the recent past and a severe illustrative representation of how a hypothetical system-wide stress may change with the introduction of digital money.

This paper deliberately models a severe scenario to account for uncertainty around how the presence of digital money may affect outflow dynamics in the future. We recognise therefore that the past may not necessarily be a good guide to the future – for example, events in March 2023 highlighted that the potential scale and pace of bank outflows could be more accelerated than previous events. Our hypothetical scenario provides an approach to considering the risks of bank outflows, with a view to illustrating the role limits might play across the full distribution of the banking sector, including any potential tipping points. This allows us to test how well we could manage increased risk to ensure financial stability in a highly uncertain world.

We then quantify the FS impact of a banking sector stress scenario across three metrics:

1. Number of firms that fall below 100% LCR: this acts as a proxy for bank liquidity stress – this is not equivalent to firm failure, as the Bank encourages the use of liquidity buffers in stress, but we note in the past that firms have taken pre-emptive action to avoid this. Hence, this could be interpreted as a behavioural indicator of potential cuts to lending to businesses and households. This assumes that banks can and do make full use of the Bank of England facilities, deposit insurance remains partially effective in limiting outflows and banks don’t make any pre-emptive adjustments to deal with the increase in risk posed by digital money (eg further increase LCR or additional collateral positioning).
2. Additional demand for central bank lending: Related to (1), we make a simplifying assumption that the assumed monetisation of collateral and HQLA significantly reduces the share of firms falling below 100% LCR. All of the Bank’s market operations in support of monetary and financial stability are designed to be used by firms and are ‘open for business’.^{footnote [19]} For modelling purposes, this is based on projected pre-positioned collateral (PPC) and non-reserve HQLA from the firms in our sample, assuming that firms fully utilise these balances.
3. Firm actions in stress: a residual deficit representing the liquidity shortfall to 100% LCR – known in terms of size (based on estimated number of firms falling below 100% LCR), but unknown in potential for financial market risks arising from stressed firm behaviour, which could increase the risk of cuts to banking lending to the real economy or potential for sizeable asset sales as banks seek to monetise assets in private markets. We assume that firms are able to monetise assets during the stress, and do not take additional precautionary actions to build liquidity buffers in stress. As a simplifying modelling assumption, our estimates also assume firms take sequential decisions where existing reserves are insufficient: (i) fully exhaust existing PPCPPC; then (ii) seek liquidity from private markets, rather than eg positioning new collateral with the Bank.

Other key assumptions

• We assume prior uptake equivalent to c.15% of household sight deposits, distributed across three SSCs and a potential digital pound.
• Prior uptake of digital money causes a reduction in deposits, liquid assets and LCRs across the system, requiring firms to rebuild liquid resources. In line with assumptions from the DP 2021, we assume firms purchase gilts to restore LCR to levels prior to the introduction of digital money, before allowing them to repo those gilts for reserves.
• We assume no frictions in moving to multiple forms of money at once – this could reflect a number of technological solutions, for example, Know Your Customer (KYC) and Anti-Money Laundering (AML) checks being carried out at the wallet-level to allow for easy transfers.
• We used the ONS Wealth and Asset survey to compute deposit distributions. This informs the impact of holding limits on the share of insured and uninsured deposits that could be withdrawn.
• We take into account the projected impacts of quantitative tightening (QT) and repayment of Term Funding Scheme with additional incentives for small and medium-sized enterprises (TFSME) by applying a conservative projection of reductions in reserves, before evaluating the FS risks from digital money ie we reduce the level of existing reserves in the banking system before introducing digital money in our model.
• For simplicity, our modelling only considers sterling deposits in a group of UK banks, and we assumed no deposit recycling ie when a bank loses retail deposits to digital money, it is not recirculated back into banking system. We note, subject to settlement timing, there could be a degree of deposit recycling as SSC issuers buy HQLA and HQLA sellers come to hold deposits in exchange. However, this would still lower banks’ LCR positions as retail deposits are replaced by wholesale deposits which have a higher outflow rate.
• For simplicity, we also assumed no other deposit flows for eg salary payments, during the same period.
• Unless otherwise stated, our results are presented as holding business limits constant at £10 million to understand the incremental impact of progressively increasing individual limits. The analysis on business limits is presented separately in the annex.

Impact with no holding limits in place

Below, we compute the three metrics under our hypothetical severe illustrative stress scenario, with no holding limits. We introduce holding limits into our model and assess their impact in the next section.

Chart 2.1 contrasts a system-wide interbank flight to safety without digital money against the illustrative stress, assuming a post-QT environment with lower liquidity buffers.

Chart 2.1: Impact of system-wide outflows without limits

Outflow rates for ‘March 23 baseline stress’ and ‘Illustrative stress’ correspond to Table 2.A for insured and uninsured deposits ^{(a) (b) (c)}

Footnotes

• Sources: Regulatory returns and Bank calculations.
• (a) Per cent of firms which fall below 100% LCR, after monetising PPC are depicted with orange circles (left axis).
• (b) Corresponding to (a), the amount of additional demand for central bank lending (£ billions) is depicted with purple diamonds (right axis).
• (c) Firms’ stressed actions (£ billions) is depicted by aqua diamonds (right axis).

Our baseline stress scenario (‘March 2023 Baseline stress’, left axis) illustrates the impact on firms before digital money is introduced, so that we can isolate the incremental additional risk under our illustrative scenario (‘Illustrative stress’, right axis).

Under our hypothetical illustrative stress scenario, there is a significant increase in the share of firms which fall below 100% LCR (21%), and a sizeable increase in the assumed utilisation of the Bank of England’s balance sheet (c.£250 billion) and in firms’ incremental stressed actions (£28 billion). We note that the assumed monetisation of PPC significantly reduces the share of firms falling below 100% LCR, without this, the share of firms would double under our illustrative scenario (42%).

Section 3: The role of holding limits in mitigating potential for disintermediation in stress

In this section, we introduce individual holding limits into our modelling.

We use our hypothetical severe illustrative scenario to guide our understanding of incremental risk as we introduce and then increase individual holding limits. Chart 3.1 illustrates the impact of varying individual holding limits on banks' LCRs and the corresponding additional demand for central bank (CB) lending.

Chart 3.1: Waterfall of the effectiveness of individual limits on mitigating impact on banks’ LCR and additional demand for central bank lending under our hypothetical ‘severe illustrative stress’ scenario ^{(a) (b) (c)}

Footnotes

• Sources: PRA regulatory returns and Bank calculations.
• (a) Baseline stress represents risks present before the introduction of digital money.
• (b) Aqua bars depict the change in % of firms with LCR falling below 100% (left axis) for different levels of holding limit (in £).
• (c) Demand for additional central bank lending (right axis, £ billions) is shown by orange shaded area.

The baseline stress represents risks present before the introduction of digital money. We then introduce digital money at an individual limits of £5,000, whilst also permitting business limits of £10 million (see annex). We observe that an increase in affected firms comes from introducing digital money (from 4% in the baseline to 9%) even at a low level of holding limit (indicated by the aqua bar above £5,000). This reflects the low average balance of the majority of individual deposits.

The percentage of firms falling below 100% LCR rises between the £5,000 limit to £20,000 – but the increase is small – before peaking at around 20% when no limit is applied. Notably, all affected firms are small banks, highlighting potential challenges for smaller banks. Under our scenario, large banks are able to manage the risk of falling below 100% LCR – however this is underpinned by an assumption that they can make significant use of their PPC at the Bank to access CB lending, as illustrated by the right axis.

Under our hypothetical scenario, the additional CB lending demanded by banks, as proxied by our projections of PPC and non-reserve HQLA, increases with the increase in limits. From £21 billion for the £5,000 limit, CB lending rises sharply to approximately £58 billion for £10,000 limit and £112 billion for £20,000 limit. The required CB funding for a no limit scenario reaches approximately £250 billion. The trajectory highlights growing liquidity demands, especially for small banks, and underscores the Bank’s critical role in maintaining financial stability. Large firms are more reliant on CB lending to avoid falling below 100% LCR.

Chart 3.2 compares the level of CB lending demanded (discussed above) with the levels observed during previous financial crises – the global financial crisis of 2008 (GFC) and the dash for cash episode in 2020 (D4C). We note that whilst the potential additional demand for CB lending could be lower than the GFC, it is comparable and could be higher than what was seen during the dash for cash in March 2020.

Chart 3.2: Effectiveness of individual limits on additional demand for CB lending, in comparison with previous financial crises ^(a)

Footnotes

• Sources: PRA regulatory returns and Bank calculations.
• (a) GFC refers to global financial crisis (2008) and D4C refers to dash for cash (2020).

Under our hypothetical scenario, even after drawing on central bank lending, a liquidity shortfall may persist for some banks. As noted in Section 1, this shortfall could increase the risk of cuts to banking lending to the real economy, with potential for sizeable asset sales as banks seek to monetise assets in private markets. Chart 3.3 illustrates the potential scale of such behaviours under different holding limits, ranging from approximately £7 billion at a £5,000 limit to over £9 billion at a £20,000 limit and £28 billion in case of no limits.

Chart 3.3: Effectiveness of individual limits under the illustrative stress scenario on the potential for stressed actions by banks ^(a)

Footnotes

• Sources: PRA regulatory returns and Bank calculations.
• (a) For the benchmark, please refer to the Table 3.A below.

Table 3.A provides a summary of our key results as discussed above. We have provided benchmark comparisons from past financial crises and Bank’s recent system-wide exploratory scenario.

In summary, we observe that by capping user balances, individual holding limits could mitigate much of the impact on banking sector disintermediation in stress, and therefore potential implications on real economy lending, as the system adjusts to introduction of digital money.

Section 4: Usability analysis

The ability of households and businesses to use digital money (‘usability’) is important to unlocking their potential innovation benefits. We have analysed household usability through three main channels: (i) the ability to receive regular income; (ii) the ability to make and receive payments as individuals currently do with banking current accounts; and (iii) the ability to make or receive some illustrative consumer purchases. These should be interpreted as measures for how households could use digital money accounts to spend, receive and hold money.

Ensuring broad usability is important for digital money. If launched, usability for the digital pound, would support the objectives of (1) ensuring the role of public money as a monetary anchor, and (2) promoting innovation, choice and efficiency. With respect to SSCs specifically, this can help a larger share of payments benefit from innovations offered by DLT technologies – such as automation.^{footnote [20]}

There is one important caveat to our analysis: analysis of consumer behaviours suggests that consumers will set themselves a lower limit than is required by a given policy limit if this is a hard limit. Behavioural theories suggest that this can arise from fear of hitting the limit and having their payment fail (loss aversion) or uncertainty about the risks and consequences of reaching their limits (uncertainty aversion). Users will therefore typically aim to avoid approaching their maximum limit, by building in a buffer, leading to a lower effective usability. The level of aversion is likely to depend on the reliability, efficiency and user understanding of mechanisms or automated tools to support consumers in making payments near the limits. These tools are explored further below.

Usability as measured by ability to receive income

The first channel concerns the ability of consumers to receive income into digital money accounts, and how this might relate to the potential limits implied by the analysis in the previous sections.

Relative to different calibrations of limits, we test the ability for consumers to: (i) enter a period with a balance equivalent to one month’s income; (ii) receive a payment equivalent to another month’s income; and (iii) also receive a payment of 10% of monthly income (interpreted as the equivalent of a ‘bonus’). Chart 4.1 shows whether consumers at different income percentiles would be able to receive these payments (ie payments worth 2.1x their monthly income), at different calibrations of limits, assuming no additional outflows. A version of this chart was previously shared in the 2023 Consultation Paper on the digital pound, and this represents an updated version using the latest available data on incomes.

This updated analysis suggests that 68% of consumers would be able to receive these payments on top of their existing balance if the applicable limit was £10,000. This increases to 94% if the applicable limit was instead £20,000.

Chart 4.1: 94% of consumers would be able to hold balances equivalent to 2.1x their monthly household income, if the applicable individual limit is £20,000 ^(a)

Footnotes

• Sources: Office for National Statistics, XpertHR and Bank calculations.
• (a) 2023 CP refers to Chart D.9 in The digital pound: a new form of money for households and businesses? Consultation Paper.

Usability as measured by rolling balances in current accounts

The second channel concerns the ability of consumers to make and receive payments in the ways they currently do using current accounts, and how that might relate to the potential limits implied by the analysis in the earlier sections.

We analysed a sample of UK consumers’ current account balances over 2019–2023, and how these balances evolved in light of payments made and received. We were then able to model the percentage of consumers that could have held their full current account balances, taking into account payments and actual usage of those accounts, within different limit values.

Chart 4.2 shows the percentage of consumers in our analysis whose account values remained within different limit thresholds. It shows that 75% of accounts did not exceed £10,000, while 90% did not exceed £20,000.

Chart 4.2: 90% of consumers would be able to hold balances and make payments in the same way as they presently do with a current account, at limits of £20,000

Footnotes

• Sources: ClearScore data and Bank calculations.

Usability as measured by payment use cases

The third channel concerns the ability of consumers to make or receive some illustrative payments, including high-value payments.

Table 4.A provides some use cases for payments, that could involve digital money in the future, as based on prices today.

The larger the applicable limit, the greater the number of one-off payment use cases that could be supported. Additional functionalities could help to support these use cases where they exceed limits.

• Sweeping: if a user account exceeds the limits for a certain period of time, a regular automated payment to sweep excess balances into a nominated deposit account could bring the account back within the limit.
• Waterfall: this functionality could automatically move excess balances into a nominated deposit account, in the instance a user has an incoming payment that would take their digital money balance over the limit.
• Reverse-waterfall: if a user wishes to make a payment that exceeds their balances, a reverse-waterfall could automatically convert deposits from a nominated account into digital money.

Interoperability with existing payment systems is key to enabling the additional functionalities. While these features expand the potential use cases for digital money, they also introduce their own risks and disadvantages. For example:

• Programmability: Having a successful set of programmable payments would require both the underlying infrastructure and network that enables money to move from a payer to a payee to support programmability, and the rail used by non-digital money may not. There would be more points of failure in a transaction, increasing the risk of user error and payment failure.
• Financial inclusion: These functionalities require users to have a commercial bank account, which reduces the ability for digital money to serve the unbanked.
• Resilience: Using a second rail creates a strategic dependency on another payment system and increased resilience risk (the payment system equivalent of tiering).
• Cost: Individual transactions would need to cover the cost of using multiple payment rails.

Section 5: Policy implications, including other considerations for limit calibration

In summary, this paper presents analysis which is indicative of what might happen if there is a rapid take up of digital money during a potential banking stress as households and businesses try to take advantage of a new technology, for example, SSCs operating on DLTs, in particular focusing on the potential for bank disintermediation in stress, which could have implications on lending to households and businesses. As noted earlier, there are other FS risks, including disintermediation of the banking sector in the transition to steady state. Any policymaker calibration of limits would need to be balanced against other risks, including risk-setting limits too low as it would also lower the usability of digital money.

Our analysis of limits is based on FS implications of a hypothetical ‘severe illustrative stress’ scenario as the economy transitions to regular usage of digital money. The scenario assumes significantly higher outflows in comparison to recent crises, with increased outflow rates driven by perceived safety and ease of transfer to digital money. This scenario is designed to consider tail risks rather than to forecast future deposit outflow. We quantify the FS risk in terms of the percentage of firms that see their LCRs falling below 100%, the demand for additional central bank lending against PPC and non-reserve HQLA, and the residual shortfall – each of which might suggest a cut in bank lending to businesses and households in transition. We have evaluated these metrics across different levels of individual holding limits, to probe the potential efficacy of limits more generally.

The number of firms at risk of low LCRs increases with the introduction of digital money, even at a low limit, which reflects the low average balance of the majority of deposits. The increase in risk from moving from £5,000 to £10,000 and £10,000 to £20,000 are relatively smaller. Despite this relatively small increase, there is some potential for individual banks to face distress, consistent with historical stress events. Our modelling suggests that risks from having no holding limits are substantially higher.

It is important to note the limitations of our approach and the results discussed. Like any other modelling exercise, our results are subject to a number of simplifications. We have applied severe assumptions on a hypothetical scenario (eg assumed outflows are more severe than seen historically) – to reflect that there remains a significant degree of uncertainty about how stress events could unfold in future, in the presence of digital money.

This analysis has informed the Bank’s policy judgements, making a case for transitional holding limits for retail payments use cases, including under the proposed regime for SSCs as communicated in today’s Consultation Paper. The determination of the right level of holding limits for each form of digital money will involve policy judgements which extend beyond the FS risks explored in this paper, such that the limits are set in a way that is both prudent and practical.

Annex: Business limits

Like individual holding limits, any calibration of business limits will be guided by policy judgement aimed at balancing financial stability risks with broader system objectives.

Higher business limits could enhance the appeal of UK-issued digital money and support the development of robust ecosystems for digital money. However, they also increase the risk of disintermediation in the banking sector and disruption to SONIA (Sterling Overnight Index Average) and money markets as corporates move their deposits out of banks and money market funds (MMFs) and into digital money.

The combination of individual and business limits, as mitigants to liquidity stress, should be considered in tandem. We used our illustrative stress scenario to explore the effectiveness of business limits in mitigating the impact of potential stress dynamics. We modelled this at three intervals – £1 million, £10 million and £100 million – to study the combined effects of individual and business limits.

Our central case for individual limits is based on a business limit of £10 million, for two reasons: (i) the £1 million cap was considered overly restrictive; and (ii) risks of corporate transactions migrating to digital money rise more rapidly with transactions above £40 million (equivalent to £10 million across four issuances of digital money), with consequent risks for SONIA.^{footnote [21]}

Business access to digital money could allow large businesses to move money out of money markets into these digital monies. An internal Bank assessment explored the potential impact of a business limit for digital money on the integrity of SONIA, which suggests that a business limit of £10 million could ensure that SONIA volumes remain manageable.^{footnote [22]}

As explained in Table A, in our illustrative scenario a lower business limit is less risky than a higher business limit combined with the same individual limit. However, the impact of moving between £10 million to £100 million is relatively modest for individual holding limits up to £20,000 – the estimated % of firms falling below 100% LCR does not increase under the illustrative scenario reflecting that outflow factors for wholesale deposits are higher than retail deposits in the LCR calculation. However, we note that additional demand for CB lending does increase. This implies that the same subset of firms may be affected at each limit level, but the scale of liquidity support required grows with higher limits due to increased outflows.