Let’s get ready to repo! - speech by Victoria Saporta

Speech

Introduction

It’s a pleasure to be here today, particularly as this is my first speech as the Executive Director for Markets at the Bank. You may have heard of the American boxing announcer Michael Buffer and his trademark catchphrase: “let’s get ready to rumble!”^{footnote [1]}. Well, today my message is “let’s get ready to repo!” While I’m not averse to whipping a crowd into a frenzy, the key difference I should be clear about is that I don’t expect this to mark the start of any pugilism. Much the opposite, it’s a call for us to work together to prepare for the normalisation of the Bank of England’s balance sheet, in a world where the Bank is no longer taking extraordinary monetary policy actions and we supply most of our reserves by lending against collateral.

The question of what our balance sheet might look like after the process of unwinding our extraordinary monetary policy operations is over (in so-called “steady state”) and the journey to it (the “transition” to that steady state) sits right at the heart of central banking. A few weeks ago, in his LSE lecture, Governor Andrew Bailey set out^{footnote [2]} the critical role that central bank reserves play in the financial system and how that relates to our core objectives of implementing monetary policy and contributing to financial stability.

As Andrew has shown, the size of the Bank’s balance sheet is largely determined by the size of reserves we supply. This gives rise to the question of what is the best operational framework for supplying these reserves and what is the optimal quantum of them in steady state. Andrew and my predecessor Andrew Hauser have set out why we judge that monetary policy implementation and financial stability objectives are best delivered by a framework where we meet the system’s demand for reserves and no more.^{footnote [3]}

As I will argue in the main body of the speech, implementing such a demand-driven system implies the need for regular and flexible lending operations which can respond effectively to fluctuations in the demand for liquidity. We launched the Short-Term Repo facility (STR) in 2022 as one critical component of such a framework. Its aim was to ensure interest rate control at the point the Monetary Policy Committee (MPC) started the process of unwinding its asset purchases, and it’s encouraging to see it being used, as we intended it to be. But the STR is just one part of our operating framework. We need to have other operations ready to fulfil our steady state vision and our Sterling Monetary Framework (SMF) counterparties need to be ready to use them. This is the subject of my speech today.

To give you the three key takeaways and the single punchline upfront:

First, we welcome the increased use of the STR as a key mechanism in ensuring interest rate control as we normalise our balance sheet. To generate the reserves that will deliver on our vision of a demand-driven operating framework in steady state, we need to use other operations to complement the STR.

Second, out of the two most obvious ways of doing so, asset purchases and lending (e.g., through longer-term repo facilities), we think there is a good case to do so through repo. The Bank already has a facility in place that can play this role alongside the STR - our Indexed Long-Term Repo facility (ILTR), which supplies reserves in market-wide auctions for a 6-month period against a broad range of collateral. We are in the process of reviewing the calibration of the ILTR to ensure that it is effective and attractive enough to support potentially large provision of reserves. We will be engaging further with the market later in the year, including through a discussion paper.

Third, the transition from where we are now - with surplus reserves - to a demand-driven framework, is well underway. While we are still well above estimates of where the preferred minimum range for reserves demand may be, there is uncertainty about the accuracy of these estimates. It’s important therefore that our facilities are robust to this uncertainty, and that firms step up their preparations to ensure they are ready to use them sooner rather than later: the point at which banks will be required to borrow reserves to satisfy their payments and precautionary demand is firmly in sight. As part of this, the recent increase in the usage of STR demonstrates welcome willingness and operational readiness to use our facilities. Indeed, for the same reasons, we also expect and would welcome increased usage of the ILTR as we move further into the transition to steady state.

The single punchline is that both we, the Bank, and you, the market, need to prepare ourselves for increased usage of both our short-term and long-term repo operations.

Or in short, let’s get ready to repo!

Some basics about operational frameworks

Before we get to the core of the speech, let us recap some basics about operational frameworks.

Implementing the desired monetary policy of policymakers to achieve the inflation target is a core aim of central banks like ours. Contributing to financial stability is the Bank’s other core aim.

But what do we precisely mean by a demand-driven system?

Let us first recall the classic motives for reserve demand: (i) transactions (to meet known or predictable outflows); (ii) precautionary (to meet potential outflows in a stress, some of which follows from risk appetite influenced by prudential regulation); and (iii) relative return (return on reserves relative to other liquid assets).

In so-called “supply-driven” operating frameworks, the supply of reserves by the central bank exceeds the minimum level of aggregate demand of market participants at the prevailing policy rate. These frameworks are also referred to as conditions of abundant reserves^{footnote [4]} and are illustrated in the right-hand side of Figure 1. The Bank has been operating with abundant reserves since Quantitative Easing (QE) programmes and lending schemes expanded reserve supply above aggregate demand. Under such a framework, banks’ payments and precautionary demand is satiated and marginal changes in demand are determined by the relative return motive. In that situation, if reserves had a relatively lower rate of return than other assets, banks would individually try to economise on their reserves balances, by lending them out to those more willing to hold them, or investing in other liquid assets, and that would drive short-term interest rates downwards. To avoid that, central banks operating such supply-driven frameworks maintain interest rate control by remunerating deposits at the policy rate. In this way, no bank has an economic incentive to lend out their reserves at a rate lower than the policy rate, as they can borrow reserves in the market and earn the policy rate by depositing them at the central bank, thereby ensuring interest rate control.

As our balance sheet normalises and we shift from the right-hand side of the figure to the left, there will come a point where supply reduces to the level consistent with the aggregate demand for transactional and precautionary balances. We do not know where this point is, but for planning purposes, we have a range of estimates of where it may lie which we call the Preferred Minimum Reserve Range (PMRR). This is depicted by the shaded grey area.

Were we to move further to the left of the PMRR, we would be in conditions of ‘reserves scarcity’. Under such conditions, banks would try to borrow reserves in the market, driving short term market rates upwards, and we would not maintain interest rate control, all else equal.

This is the reason the Bank introduced the STR at the point that the MPC started unwinding its asset purchases.^{footnote [5]}

In a demand-driven framework, we aim to ensure banks can obtain enough reserves at or just above the point of incipient reserves scarcity. We expect market interest rates to be relatively stable around Bank Rate in this region achieving interest rate control - a key and common objective of central bank operating frameworks. The net terms of our lending facilities will also influence the quantity demanded and will impact on the degree of volatility around Bank Rate. Put differently, quantity demanded is endogenous to the terms of supply.

We also expect a demand-driven framework to deliver financial stability benefits largely because it is designed to supply the reserves that banks demand for payments and precautionary purposes. Precautionary demand is in turn influenced by regulatory liquidity requirements on high quality liquid assets (HQLA) that depend on the size and flightiness of banks’ liabilities.

By comparison to the demand-driven system I have described, a scarce reserves framework would encourage banks to rely more on non-reserves HQLA to meet banks’ liquidity needs. Such assets are highly-liquid as the definition implies but in order to be used to meet cash outflows, would still need to be converted into reserves^{footnote [6]}. As a result, such a framework would place greater reliance on the ability to immediately and smoothly convert non-reserves HQLA into reserves in a stress. At the same time, such frameworks may promote more private money market activity and liquidity recycling, potentially supporting the depth of repo markets and the ability of banks to convert non-reserves HQLA into cash. By the same token, the footprint of the central bank tends to be lower with associated reputational, financial and political economy benefits.^{footnote [7]} On the other hand, if markets are not used to banks accessing lending facilities in normal times, central bank operations may get stigmatised during market stress.

As a second reference point, abundant reserves frameworks could encourage firms to rely too much on reserves to meet their HQLA needs and hold only limited amounts of other HQLA. If in addition these reserves were provided through lending operations encumbering assets that could otherwise be used to raise liquidity in a stress (reducing so called “dry powder”), this could reduce the overall liquidity resilience of the banking sector.

Given the trade-off involved between the financial stability costs of potentially stigmatised facilities associated with scarce reserves systems and the potential financial stability costs of less dry powder that could be associated with abundant reserve systems, a demand-driven system seems to strike the right financial stability balance^{footnote [8]}.

Having argued that a demand-driven framework achieves interest-rate control and best aids financial stability, the next question is how to assess the demand for reserves. In practice, we regularly survey banks on their preferred minimum level of reserves holdings and the factors that drive the PMRR. We cross-check this with estimates of the level of the PMRR derived from models^{footnote [9]} and by using information from banks’ prudential liquidity returns. Critically, any estimate is uncertain and will change over time. For example, if the world became structurally a riskier place, we would expect precautionary demand for reserves to go up. That is why our framework is explicitly designed to be robust to that uncertainty – to stand ready to provide reserves whether the true revealed demand turns out to be higher or lower than our estimates.

That implies we must have a set of flexible and useable facilities to meet reserves demand as it arises. Given that on all our measures we expect a higher level of reserves demand through repo than prior to the Global Financial Crisis, those flexible supply arrangements must be able to do two things:

1. Supply a potentially large stock of reserves in the event that the revealed demand for reserves turns out to be large;
2. Be responsive to higher frequency changes in the demand for reserves, both to maintain interest rate control as conditions evolve but also responding to meet demand in stress.

The latter implies the need to supply at least some reserves elastically via repo so that changes in demand are broadly matched by changes in supply, rather than by causing fluctuations in market interest rates. This is the primary purpose of the STR, in that once a week, the supply of reserves is perfectly elastic. Banks can get as many reserves as they want for a short tenor of one week, against high quality (level A, e.g. gilt) collateral at Bank Rate. Banks retain incentives to manage their liquidity over short horizons, as they can borrow reserves each day through private markets and they can also make use of our Operational Standing Facilities and the discount window, which are there to be used, albeit those are priced at a spread to Bank Rate.

But how to best deliver the former requirement for our supply arrangements is more of an open question - so now let me explore that in more detail focussing on the steady state balance sheet. I will come to the transitional issues later in the speech.

We need a way of supplying a stock of reserves

At the simplest level, there are two choices for the Bank in deciding how to supply the stock of reserves: we could supply a large portion of reserves by purchasing assets directly – which forcibly injects a fixed quantity of reserves into the system (until those purchases are unwound). In our case the most natural asset to purchase would be UK government bonds, i.e., gilts. Alternatively, we could supply a large proportion of reserves by offering lending facilities (i.e., repo). As I have explained, the demand-driven nature of the operating framework we want to implement already certainly implies some role for short-term lending operations to supply the marginal reserve. But here, the question is how much of the ‘stock’ of reserves should be supplied by asset purchases versus repo operations.

In his LSE speech, Andrew already gave you the headline, that is, that we think there is a strong case against a set of principles for a majority of reserves to be supplied by our repo operations in future. In coming to that view, we have assessed how a repo-led and a gilt-led framework would perform against a set of principles.

There are several strong reasons to favour a repo-led framework from the perspective of our core policy objectives.

First a repo-led framework will allow the Bank’s balance sheet to shrink or grow in response to the changing level of reserve demand in the system, as well as from individual firms. The result is likely to be both a more efficient and responsive balance sheet and a more variable one – in which changes in size from one period to another come to be expected by the market. Moreover, a repo-led framework for reserves supply normalises the use of Bank lending facilities which brings financial stability benefits. Firms are more likely to borrow from the Bank in scale during periods of stress if they already do so in normal market conditions. And by operating one which accepts a broad range of collateral, we ensure the largest possible number of firms can access our facilities^{footnote [10]}.

By contrast, supplying the majority of reserves through gilt purchases is likely to reduce the scale and breadth of participation in Bank facilities and risks producing periods of excess reserves supply, since the central bank needs to decide the size of the portfolio^{footnote [11]}.

Second, a repo-led portfolio will leave monetary policy makers with more room to manoeuvre. Providing the majority of reserves via repo would preserve the potential headroom for any future QE programmes relative to a gilt portfolio. Reserves supplied via repo can easily be displaced by QE operations, when monetary conditions require such a stimulus^{footnote [12]} whilst keeping the risk of possible market distortions from the central bank purchasing a significant fraction of gilts lower.

Third, a majority gilt approach in normal times would expose the Bank and the public sector at large to greater interest rate risk, whilst not providing greater monetary and financial stability benefits. Andrew’s LSE speech covers this area fully, so there is no need for me to elaborate further.

Although we believe the policy case for moving towards a repo-led framework is strong, there are also a number challenges that we, and firms, will need to navigate.

First, by tying up collateral for use in our reserve supply operations, a repo-led framework may result in the banking sector having less dry powder in the form of unencumbered assets to use if a market stress did occur, relative to a gilt-led system.^{footnote [13]}

Our analysis to date suggests that a repo-led framework would only result in a modest rise in overall encumbrance levels although aggregates may conceal pockets of encumbrance within the system. Critically, the UK banking system has a large amount of potentially eligible collateral. By way of illustration, we estimate the potentially eligible universe of sterling loan collateral that exists today is around three times the amount that would be needed to generate estimates of the PMRR.^{footnote [14]} But currently only around a third of that potentially eligible loan collateral is pre-positioned with us to use if needed (Chart 1).

That highlights the need for us to collectively work on further pre-positioning. And Chart 1 shows just the sterling denominated loan collateral (a subset of Level C collateral) – we accept a much broader range than that (Levels A, B and C, all in sterling and non-sterling).

Second, it will be a major change in operating practice, both for us and firms. The Bank has not supplied a large stock of reserves via a repo portfolio for an extended period of time. That’s what I will turn to now.

The Indexed Long-Term Repo (ILTR) facility can play this role

In order to supply the potentially large stock of reserves that the system may demand in steady state – judging from our current survey-based PMRR estimates of £345-490bn - we need to complement the STR facility with a longer-term facility. Even at the bottom end of this range, it would not make operational sense either for us or our counterparties to keep rolling over such large amounts in using solely short-term facilities repeatedly. These longer-term lending facilities also need to be collateralised with a broader range of assets making them usable for the widest range of firm business models admitted into the sterling monetary framework. This will aid the distribution of reserves where they are needed. And by accepting non-HQLA collateral in return for reserves (suitably valued and risk managed), a longer-term facility against broad collateral offers a way for the banking sector to adjust its aggregate holdings of HQLA as it deems necessary – since a liquidity upgrade against non-HQLA adds to the supply of liquid assets in the financial system, which can be beneficial for financial stability. ^{footnote [15]}

We already have such a facility. The ILTR facility provides liquidity in the form of central bank reserves, for a term of 6 months, against the full range of eligible collateral (from the most liquid levels A – such as gilts or Treasuries, to the least liquid level C – such as loan pools). It is designed to be flexible and responsive to evolving market conditions, by providing more liquidity to the market as demand for liquidity increases.

The ILTR has been used by the market in size before but, unsurprisingly, usage has only been moderate in recent years. With ample liquidity in the system, including the presence of cheap term funding in the form of the Covid-era Term Funding Scheme with additional incentives for SMEs (TFSME) more recently, there has been less of a role for the ILTR (Chart 2). That means the ILTR has mainly been associated with providing liquidity insurance which is why most of the use we’ve seen in the past has been during periods of stress.

As we transition to our repo-led, demand-driven framework, we expect firms to use the ILTR in normal times to source reserves for payments and precautionary reasons. Critically, we do not view it as a facility to be used solely for liquidity insurance purposes. Using a phrase deployed by Mark Carney when he was the Governor of the Bank of England, we are open for business^{footnote [16]} and firms should use our repo facilities, both short and long term, to source reserves in the future. And as I have explained, this means that, unlike in a system of supply-driven abundant reserves, moving forward the ILTR plays a greater role in both supplying the reserves necessary for interest rate control, as well as supporting financial stability.

We are therefore also in the process of reviewing the calibration of the ILTR to ensure it is effective and attractive enough so that it is used in business-as-usual, consistent with our policy objectives.

By the end of 2024, we plan to further step up our dialogue with market participants around how our facilities are calibrated, should evolve over time, and how markets will adapt to a changing liquidity environment, including through publishing a discussion paper, where we will be asking firms’ views on calibration amongst other issues. We’ll also be doing sessions with firms next year to explain how our facilities work and what firms need to do to be ready to use them.

The transition from supply-led abundant reserves towards the PMRR is underway

As I have already alluded, we’re not starting from a blank slate. Our current assets and liabilities reflect multiple crisis-era interventions, and by far the dominant driver of reserves has been unconventional monetary policy. The lion’s share of that reflected funding for the MPC’s QE programme, via reserves created to finance the Asset Purchase Facility (APF) (Chart 3).

In addition, Covid stimulus such as the TFSME added further reserves into the system. TFSME aimed to support the pass-through of low interest rates and support lending to the real economy through covid-related disruption, and its unwind represents a normalisation of the Bank’s balance sheet from such crisis-era funding schemes.

As reserves fall as a result of QT and TFSME unwind, money markets are moving from years of plentiful reserves and relative collateral scarcity to the present period of falling, though still ample, reserves supply, and more available collateral (including as government debt issuance continues). We are beginning this transition away from a supply-driven framework towards the demand-driven framework I talked about earlier. We are moving from the right-hand side of Figure 1, towards the PMRR.

As that process continues, banks’ preferred holdings of reserves as well as non-reserves HQLA will evolve depending, amongst other things, on asset availability, relative rates of risk and return and regulatory rules as well as perceived ease of monetisation in a stress. We are beginning to see some of the changes in this new liquidity environment already in the behaviour of banks, as they have increased the share of gilts in their HQLA buffers (Chart 4).

As we are moving towards the PMRR and before we reach it, we expect frictions in the distribution of reserves to lead to pockets of reserve demand appearing which could lead to periods of market rates moving upwards. Factors such as the return on reserves relative to other forms of HQLA could push up or down on the demand for sterling reserves (perhaps even above the PMRR) even for extended periods of time, and that may be reflected by usage of our facilities. That makes it possible that reserves could settle for periods above the PMRR, and correspondingly we will expect firms to come to our repo facilities - either the STR or ILTR - to source reserves.

Indeed, we think we have been seeing some of these frictions, albeit in a temporary form, recently. Money market rates have been slightly more likely - than in the recent period of especially abundant reserves and correspondingly stable rates - to move upwards in response to demand for cash, especially around month and quarter ends, when bank balance sheets are most constrained. We had an example of this at end-April when a large gilt maturity (and a related fall in reserves) coincided with global factors increasing the relative demand for sterling through FX swaps. Repo rates rose to around 10bps above Bank Rate (Chart 5), and STR usage increased shortly thereafter, as intended in periods like these^{footnote [17]}. We saw a moderation in the level of market interest rates, thereafter, as the temporary factors unwound.

Turning to unsecured markets, in recent years, when reserves have been abundant, the overnight unsecured rate (SONIA) has sat beneath Bank Rate. As reserves have been falling, we have seen an increase in SONIA rate and the SONIA to Bank Rate wedge compressing, with the spread compressing by 1 basis point from around 6bps in Q1 2024 to 5bps^{footnote [18]} (Chart 6).