The central bank balance sheet

Introduction 

Good afternoon. It is a pleasure to be here today. 

Let me begin by thanking Isabel Schnabel for her numerous contributions during the past seven years at the ECB. Through her expertise, dedication and thoughtful engagement, Isabel has profoundly enriched the policy debate, both within the ECB and across the globe. I wish her all the best in her new position. 

The ECB Conference on Money Markets is a unique forum. It brings together academics, market participants and central bankers, offering a valuable platform for connecting analytical research with practical market and policy experience on questions related to money markets and central bank operations – issues central to central banking but often not in the limelight. 

Today I would like to discuss one of these core issues, which for a long time was very much a niche question: the size of a central bank’s balance sheet. 

Over the past two decades, central bank balance sheets have undergone a profound transformation. After the Great Financial Crisis, and again during the Covid‑19 pandemic, central banks in many economies expanded their balance sheets substantially. These interventions were essential in supporting monetary and financial stability during periods of severe stress and when monetary policy was constrained by the effective lower bound. 

As those crises receded and monetary space improved, attention gradually shifted from expanding balance sheets to reducing them. Here in the euro area, for example, the balance sheet has been declining since 2022. Many other central banks have followed a similar path. This has brought a broader question into focus: outside crisis periods, what are the costs and benefits of a large central bank balance sheet? 

Determining the appropriate size of a central bank balance sheet is not straightforward. 

For one, the appropriate size may be time-varying. As I just discussed, the balance sheet can be used to address financial instability in crises or to set the monetary policy stance, for example when the interest rate is constrained by the effective lower bound.

Today I want to abstract from these episodes and focus on the structural size of the balance sheet. More specifically, I want to focus on commercial banks’ holdings of reserves at the central bank – what I will refer to simply as reserves. I will examine the main factors determining the structural level of reserves in the system, the policy choices available to central banks and the trade-offs that those choices entail. These trade-offs need not be assessed in the same way across jurisdictions. Differences in financial systems, market structures, institutional frameworks and monetary policy objectives can lead central banks to different conclusions regarding the appropriate level of reserves.

Conceptually, the level of reserves is shaped by two forces.

The first is the central bank’s operational framework, which determines the supply of reserves.

The second is banks’ aggregate demand for holding reserves.

I will consider these in turn.

Operational frameworks

Let me start with monetary policy operational frameworks. They can be classified according to the segment of the demand curve where the system operates.¹

In a corridor system, also known as a scarce-reserves system, the central bank calibrates the reserve supply to intersect the downward-sloping portion of the demand curve. 

In a floor system with ample reserves, supply intersects the flat part of the demand curve, close to where the demand flattens (ie the kink). With abundant reserves, supply lies well beyond the kink. 

In a tiered system, the central bank engineers a second steep segment of the reserves demand curve by remunerating reserves at a lower rate above a specified threshold. 

Operational frameworks affect the marginal opportunity cost of holding reserves overnight.

The marginal opportunity cost is the return a bank forgoes by depositing its last unit of reserves at the central bank rather than lending it in the market.

Under a scarce reserves framework, the money market rate trades above the deposit facility rate and the marginal opportunity cost is positive, as currently in Brazil, India or Korea. 

In ample- and abundant-reserves systems, the money market rate trades close to the deposit facility rate or even below it.² This is currently the case in the euro area, as in many other advanced economies. 

The Swiss National Bank (SNB) operates a tiered framework. The marginal opportunity cost is positive because the money market rate trades between the higher and lower remuneration rates. Despite the higher complexity, the SNB introduced this framework to ensure active money market trading in an environment of abundant reserves.³ 

The desired level of the marginal opportunity cost remains subject to debate. 

As a starting point, we could apply the Friedman rule. As the marginal cost of supplying reserves is very small for the central bank, this would argue for an ample reserve framework with low marginal opportunity costs. Lorrie Logan (2023) made this point here three years ago. 

But while the central bank’s cost of supplying large reserves may be small, the costs to the financial system and society may be greater – as argued by Isabel Schnabel (2025) last year. For example, a substantial central bank footprint may distort resource allocation and disintermediate financial markets. Such negative externalities would imply positive marginal opportunity costs.

That said, there may also be positive externalities. For example, a large quantity of reserves may reduce the likelihood of financial stress by preventing fire sales and “dash-for-cash” episodes. If so, the liquidity needs of individual banks may not reflect the needs of the financial system as a whole. Recognising these positive externalities would imply a larger reserve supply.

In practice, both negative and positive externalities coexist, making it difficult to determine their net effect. In the following, I want to discuss some of the trade-offs of different operational frameworks for monetary policy implementation, market impact and financial stability.

Monetary policy implementation

Let me start with monetary policy implementation. 

All the operational frameworks I described can achieve monetary control. This is unsurprising: central banks would not adopt a framework that did not allow them to implement their desired policy stance effectively.

There are, however, some differences.

For one, operational frameworks differ in terms of how closely the operational target tracks the desired policy stance.⁴ In particular, with ample- and abundant-reserves systems, the operational target generally remains closer to the central bank’s policy rate than under scarce-reserves systems. You can see this when you compare the standard deviation of the spread between the operational target and the policy rate vis-à-vis reserve levels. 

One reason for the differing volatility of the operational target is the way changes in reserve demand and supply affect interest rate control. For example, changes in autonomous factors, such as higher public demand for currency in circulation or a rise in government deposits held at the central bank, affect the net supply of reserves. Under frameworks with high reserve levels, such changes do not need to be offset to maintain the desired monetary policy stance. With scarce reserves, by contrast, changes in reserve supply could push money market rates away from the middle of the corridor. To counter this, central banks have to intervene based on forecasts of developments in autonomous factors, which can be challenging.

That said, even in scarce-reserves systems, volatility in the deviations of the operational target from the policy rate is typically limited to a few basis points. What is more important, particularly in normal times, is that this volatility does not appear to be transmitted to other short-term interest rates. 

There are also other operational considerations. A high-reserves framework can support the central bank’s ability to increase the supply of reserves when needed without undermining interest rate control. In a scarce-reserves system, an increase in the supply of reserves, eg due to financial stability considerations, would push the operational target towards the floor, altering the monetary policy stance.

Market impact

Let me now turn to the impact reserve supply has on money markets.

Operational frameworks shape banks’ incentives to trade reserves and therefore influence interbank market activity.

With scarce reserves, banks face a positive marginal opportunity cost of holding reserves. Therefore, they have an incentive to trade them, which leads to an active interbank market. As the supply of reserves shifts along the demand curve – from scarce to ample and, ultimately, abundant – the opportunity cost declines and incentives to trade weaken. This tends to reduce market activity. A tiered framework provides support for interbank trading, because banks with reserves above their quotas are incentivised to trade with banks whose reserves fall below their quotas.

The interbank market may also fragment when the central bank provides ample or abundant reserves. Since the marginal opportunity costs are low in this case, some banks become inactive and rely solely on their central bank accounts to manage payment flows. Hence, the interbank market atrophies. For example, since Sveriges Riksbank started operating with high reserve levels in 2020, around 50% of central bank counterparties have not participated in any money market segment.⁵

Views on the importance of interbank markets differ among central banks.⁶ 

High levels of activity can facilitate the reserve redistribution and avoid market fragmentation. But reserves can also be distributed via broader money markets, such as the secured market, which has remained very active despite the large quantity of reserves. It is nevertheless unclear whether activity would have been even stronger if reserves had been less abundant.

High interbank activity can also provide signals about monetary policy implementation and potential financial stress. And it can support price discovery and market discipline. However, other market segments also provide market signals. And the experience from the Great Financial Crisis and available research points to mixed empirical evidence on how strong the disciplining role of interbank monitoring is. The function may also be performed effectively by regulation and supervision.

At the same time, there is a debate on whether the unsecured interbank market would be revived if reserves become scarce. For instance, a recent ECB paper finds that regulatory costs are substantially higher for unsecured lending than for secured transactions when backed by high-quality collateral.⁷ This cost differential suggests that unsecured market activity may remain weak. That said, unsecured interbank markets in Japan recovered after the Bank of Japan introduced a new tiering framework.⁸  

Financial stability

Let me next discuss the link between operational frameworks and financial stability.

Since central bank reserves are the ultimate means of payment, ensuring their adequate and timely provision is critical for maintaining financial stability. 

Again, all operational frameworks I reviewed can achieve this. But they do so in different ways.

Central banks operating with ample or abundant reserves provide excess reserves ex ante, beyond what is technically needed to implement monetary policy. In contrast, central banks operating with scarce reserves provide additional reserves ex post, once stress materialises. 

There are trade-offs associated with these different approaches.

A high level of excess reserves strengthens banks’ ability to absorb liquidity shocks, both individually and at the system level. Other things equal, this reduces the likelihood of stress. Banks’ holdings of excess reserves reduce the need to rely on unsecured interbank markets, to monetise assets, such as non-reserve high-quality liquid assets (HQLA), or the need to access central bank lending facilities. Both can be difficult due to frictions. 

A higher share of reserves in HQLA also reduces banks’ sovereign debt holdings, limiting the risk of a sovereign-bank nexus.

Moreover, in stress money markets can fragment as banks become less willing to lend to one another. This can arise because of heightened counterparty credit risk and liquidity hoarding, and can lead to persistent reserve imbalances. Unless reserves are supplied ex ante, banks facing reserve shortages may need to borrow from central bank facilities, often at penalty rates and against stringent collateral requirements. 

However, the likelihood of stress could rise as the financial system endogenously adjusts to the high levels of reserves. For instance, as discussed in detail later, monetisation frictions may increase with the quantity of reserves, and banks’ balance sheets may become more fragile. 

Demand for central bank reserves and its drivers

Let me now move on to the second piece of the puzzle: the demand for central bank reserves. 

There is evidence that reserve demand has increased over the last decade, especially in major advanced economies.⁹ What factors have driven this shift, and can the central bank do something about them? In what follows, I will discuss some factors that could have contributed to changes in reserve demand, including the possibility that reserve demand may respond endogenously to the central bank’s operational framework.

Liquidity regulation

Liquidity regulation is often mentioned as one of the key drivers of increased reserve demand. However, it is important to note that different types of liquidity regulation can affect reserve demand in different ways.

Some regulations effectively function as reserve requirements. This is, for example, the case for intraday liquidity requirements in some jurisdictions. It is also the case where supervisory pressure induces banks to hold a fraction of HQLA in reserves. Such requirements increase reserve demand. 

By contrast, the Liquidity Coverage Ratio (LCR) regulations do not necessarily increase reserve demand. The reason is that LCR and similar liquidity regulations can be met by non-reserve HQLA. Banks will choose the optimal mix of reserves and non-reserve HQLA from a risk-return perspective. Indeed, there appears to be little relation between the LCR and the level of reserves. However, as reserves rise and the opportunity cost declines, reserves account for a larger share of HQLA. 

Frictions affecting reserve demand

To meet liquidity outflows, banks can draw on their central bank accounts, borrow in secured or unsecured money markets, or access central bank facilities. Various frictions can affect banks’ choices among these options and, consequently, their demand for reserves. 

If the money market were frictionless, the banking system would require no excess reserves. Banks with surplus reserves would simply lend them to banks facing shortages. However, when reserves cannot be redistributed seamlessly, some banks face a risk of ending the day with a reserve shortfall, especially when payment flow volatility is higher. This increases the insurance value of holding reserves. 

Recent policy debate has highlighted two factors that may have increased money market frictions. 

One such factor is increased monetisation frictions. The frictions flatten the reserve demand curve and shift its kink outwards. 

From a liquidity risk management perspective, reserves and other HQLA are not perfect substitutes. Liquidating non-reserve HQLA in money markets takes time. This friction may be particularly severe when banks need to liquidate large volumes of non-reserve HQLA in stress. Stressed outflows may also have become more demanding as runs have accelerated, as evidenced by the banking turmoil in 2023. 

Monetising HQLA at the central bank could also be hampered. This might be the case because of stigma. It might also arise because of a lack of operational readiness that constrains banks’ ability to access the central bank. This issue was also evident during the Silicon Valley Bank crisis. Again, the implication of this is a higher insurance value of holding reserves. 

A second factor that may have affected money market frictions is the growing presence of non-bank financial institutions (NBFIs). A larger NBFI footprint has an ambiguous effect on money market frictions. On the one hand, more liquidity providers or takers increases the diversity of market participants and thus resilience. In stress, though, the flightiness of certain NBFIs may amplify swings in money markets and undermine liquidity.

Endogeneity of reserve demand 

Reserve demand may also respond endogenously to the central bank’s operational framework.

One channel works through interbank market fragmentation that can increase reserve demand and lead to unexpected shifts. Evidence from Sweden suggests that, as reserve supply increased and the interbank market fragmented, banks raised their precautionary reserve buffers.¹⁰ With fragmented markets, reserve demand can also shift unexpectedly because inactive banks become like autonomous factors.¹¹ Their net payment flows to active banks give rise to unexpected shifts in the effective reserve demand curve. These shifts can be especially pronounced if there are additional negative reserve supply shocks that active banks need to absorb. An example is a large tax payment if the government banks with the central bank. 

Banks’ endogenous response to high reserve levels could also increase their liquidity dependence and increase financial vulnerabilities.¹² For example, evidence from the United States suggests that as reserves expanded, banks increased uninsured demandable deposits and extended more credit lines to firms. If these claims on liquidity do not decline proportionally when reserves are reduced, banks might be left more exposed to liquidity risk and contribute to greater financial fragility. 

Conclusion

Let me conclude.

Today, I have discussed the determinants of the structural size of a central bank balance sheet – away from crisis times and when monetary policy is not constrained by the effective lower bound. And I have focused on central bank reserves, as it is the liabilities that largely determine the structural balance sheet size and reserves are the item most directly controlled by the central bank.

I have discussed that the size of reserves is affected by both the supply – driven by the operational framework – and banks’ demand for reserves. I also highlighted several potential trade-offs across frameworks. 

How to weigh them in practice is difficult. Different countries inevitably make different choices regarding their operational frameworks depending on country-specific experiences and circumstances.

While I have focused today on the structural level of reserves, an important issue is whether the various endogenous effects I have discussed can be reversed when balance sheets are reduced. This remains an open question. That said, recent experiences in many jurisdictions, including the euro area, suggest that it is possible to achieve a smooth balance sheet reduction. 

References

Acharya, V, R Chauhan, R Rajan and S Steffen (2024): “Liquidity dependence and the waxing and waning of central bank balance sheets”, NBER Working Papers, no 31050.

Afonso, G, D Giannone, G La Spada and J Williams (2022): “Scarce, abundant, or ample? A time-varying model of the reserve demand curve”, Staff Reports, no 1019, Federal Reserve Bank of New York.

Åsberg-Sommar, P, M Drehmann, D Hansson and V Shreeti (2026): “What determines banks’ excess demand for reserves?”, BIS Working Papers, no 1376.

Bank of Japan (2024): “Developments in the Japanese money markets and their functioning with excess reserves”, BOJ Reports & Research Papers, 6 November.

Bindseil, U (2016): "Evaluating monetary policy operational frameworks", in Designing resilient monetary policy frameworks for the future, proceedings of the Economic Policy Symposium, Jackson Hole, Wyoming, Federal Reserve Bank of Kansas City.

Cavallino, P, M Drehmann, R Finlay and J Remache (2025): “Monetary policy operational frameworks – a new taxonomy”, BIS Quarterly Review, September.

Cavallino, P, M Drehmann and A Mehrotra (2026): “The central bank balance sheet, the operational framework and reserve demand: what are the trade-offs?”, forthcoming.

Drehmann, M and X Fu (2026): ”Hard to shift, easy to reshape: central bank reserve demand and frictions”, BIS Working Papers, no 1372.

Faïk, E and O Vergote (2026): “Regulatory incentives and the structure of euro area money markets”, ECB Occasional Paper Series, no 401. 

Kim, K, A Martin and E Nosal (2020): “Can the US interbank market be revived?”, Journal of Money, Credit and Banking, vol 52, no 7.

Logan, L (2023): “Ample reserves and the Friedman rule”, speech at the European Central Bank Conference on Money Markets 2023, Frankfurt am Main, 10 November.

Maechler, A and T Moser (2022): “Return to positive interest rates: why reserve tiering?”, speech at the Money Market Event, Geneva, 17 November. 

Potter, S (2016): "Commentary: Evaluating monetary policy operational frameworks", in Designing resilient monetary policy frameworks for the future, proceedings of the Economic Policy Symposium, Jackson Hole, Wyoming, Federal Reserve Bank of Kansas City.

Schnabel, I (2025): “Towards a new Eurosystem balance sheet”, speech at the ECB Conference on Money Markets 2025, Frankfurt am Main, 6 November.

Soares, C (2025): “Liquidity dependencies in the euro area”, ECB Working Papers, no 3056.

Thedéen, E (2025): “The banks need to have more active liquidity management”, speech at a meeting of monetary policy counterparts, Sveriges Riksbank, Stockholm, 11 September.

Ugai, H and T Osada (2026): “Reversal of the BoJ’s balance sheet policy and liquidity dependence”, The Japanese Economic Review, vol 77.

Ungarelli, F and T Kostka (2026): “Estimating the time-varying reserve elasticity of money market rates in the euro area”, ECB Economic Bulletin, no 1/2026.