Stephen Cecchetti is the Rosen Family Chair in International Finance at Brandeis International Business School, Brandeis University, and Jens Hilscher is a Professor at the University of California, Davis
“As the sole issuer of euro-denominated central bank money, the euro system will always be able to generate additional liquidity as needed. So, by definition, it will neither go bankrupt nor run out of money. And in addition to that, any financial losses, should they occur, will not impair our ability to seek and maintain price stability.”
ECB President Christine Lagarde, Committee on Economic and Monetary Affairs Monetary Dialogue, 19 November 2020.
During and after the Global Financial Crisis of 2008-09, central banks engaged in large-scale asset purchase programmes, significantly increasing the size of their balance sheets. Purchases continued during the COVID-19 pandemic in the early 2020s.
Figure 1 plots the consolidated assets of the Federal Reserve System and the Eurosystem. In both cases, total assets peaked at nearly ten times their level in 2008.
Figure 1. Total assets of the Federal Reserve and Eurosystem, January 2006 to March 2024, weekly
Source: FRED.
Importantly, central banks purchased bonds when long-term rates were low. So, when interest rates rose in 2022 and 2023, their holdings started to generate losses. This sparked a debate both over whether it was prudent to amass these portfolios and if the mounting losses would undermine the ability of central banks to meet their price stability objectives.
Indeed, some observers argue that it is important to focus on both central bank capital and profitability, concepts that were previously of limited interest to researchers and policymakers. The idea seems to be that the valuation effects and implications for net interest margins – ie. losses associated with large asset purchases – are of sufficient size to have welfare implications for society. In contrast, the above quote suggests that the central banks themselves are much less concerned.
In a recent paper (Cecchetti and Hilscher 2024), we develop a framework for understanding the medium- and long-run implications of the losses arising from central banks’ large-scale asset purchases. Specifically, we suggest that they are best viewed as a form of fiscal policy.
We arrive at this conclusion by consolidating the balance sheets of the central bank and the finance ministry. Consolidation clarifies that bond purchases change the maturity structure (and in some cases the gross quantity) of outstanding government debt.
For example, large issuance of short-term reserves and concurrent purchases of long-term debt reduce the maturity structure of government liabilities, while leaving the overall quantity unchanged.
When monetary policymakers initially purchase bonds, their objective is to reduce longer-term interest rates, compressing some combination of sovereign term spreads and risk premia on private sector bonds. If successful, these policies stimulate aggregate demand, stabilising inflation, growth, and employment, and the financial system.
Ideally, this reduces the length and severity of recessions, thereby increasing aggregate welfare. Policymakers generally do not focus on the potential for future profits or losses1.
Nevertheless, to the extent that policies are enacted when risk-free rates are particularly low or term spreads are especially large, asset purchases also may generate large cash-flow gains from the fact that bond coupon payments exceed funding costs. However, as time passes and interest rates rise to their longer-run steady state level, funding costs will rise and the central bank will begin to suffer losses.
If interest rates rise further, for example due to policies designed to combat higher-than-expected inflation, these losses will increase2. As a result, remittances from the central bank to the fiscal authority can shrink or cease altogether.
Depending on the structure of the indemnity agreement with the finance ministry, as well as the accounting rules that are in place, central bank losses (negative payments) may trigger an explicit fiscal expenditure either immediately or at some time in the future. It is in this sense that central bank balance sheet policies have direct fiscal consequences.
The nature of the losses, and their resulting economic impact, depends on what it is that the central bank chooses to purchase: domestic bonds or securities issued by a foreign entity. Purchases of domestic sovereign bonds change the maturity structure of privately held government debt, implying no direct ex-ante transfer of wealth, though there may be transfers ex post. The addition of more short-term liquid government bonds reduces the expected size of interest expenses but increases its variability (Greenwood et al 2015).
Indeed, the realised path of interest rates resulted in higher interest costs. The impact on ex-ante welfare derived from a change in the maturity structure depends on the relative importance of these two effects – the benefits of liquidity services relative to the deadweight loss from taxation (Barro 1974, Lucas and Stokey 1983).
Similarly, purchasing private sector bonds can create ex-post transfers. But since owners of the bonds’ issuing companies are domestic, this is a within-country transfer. In contrast, purchases of foreign securities result in transfers to foreigners.
Importantly, however, central bank losses are distinct from central bank solvency. As a technical matter, central banks have very little capital. For example, the Eurosystem as a whole had capital equal to €120 billion supporting total assets of €8.8 trillion in October 2022 – a leverage ratio of roughly 75.
Similarly, the Federal Reserve System has $43 billion of capital with peak assets of $9.0 trillion, implying leverage of over 200. As a result, even very modest losses can lead to insolvency. Our view accords with that of Reis (2015) and Buiter (2008) who suggest that this is a technical issue with only tangential relevance for the ability of central banks to carry out policy.
However, solvency may be essential for credibility (Wessels and Broeder 2022). Importantly, with the appropriate fiscal support (which is generally not in question), solvency will never be an issue.
Notably, though, the form and timing of the fiscal support is important. Two aspects are relevant: when the central bank realises losses and when the fiscal authority transfers funds to make the central bank whole again.
If the central bank realises losses when securities are sold or mature, then the timing of sales becomes relevant. This is the case for the Bank of England. At the same time, if the central bank uses fair value accounting, ie. if the central bank uses mark-to-market valuation of its portfolio, then losses are generated as soon as interest rates increase. This is the case for the Sveriges Riksbank.
Turning to when these losses trigger payments from the fiscal authority, there are two possibilities. One is that losses are paid for as they arise, while the second is that payments are delayed in some way.
To get some sense of the magnitude of the losses, and the fiscal implications, we consider two examples: the Bank of England and the Federal Reserve3. Both hold primarily domestic sovereign bonds. But the specific holdings, size and treatment of losses are different.
We propose merging the balance sheets of the central bank and the fiscal authority into one consolidated government balance sheet. This clarifies the consequences of central bank asset purchases, emphasising that these actions have fiscal consequences
Importantly, the UK has indemnification rules which mean that any realised losses translate almost immediately into a fiscal expenditure. In the case of the Fed, the losses simply postpone the date when transfers to the US Treasury will restart (they are currently on hold).
Finally, the undiscounted sum of the Bank of England’s losses appears to be larger than that of the Fed – 7.8% versus 0.8% of nominal GDP – and could stretch out for decades as opposed to lasting for only a few years.
When the central bank purchases foreign securities, the transfers are between domestic residents and foreigners. Assuming that countries place different welfare weights on domestic and foreign residents, transfers are qualitatively different.
To get a sense of the potential magnitude, consider the case of the Swiss National Bank. In stark contrast to the Fed and the Bank of England, the central bank’s losses, which are mainly transfers to foreigners, are quite large: annual net income varies from +8% to –17% of GDP.
Finally, we look at the case of the Eurosystem. While the (currently 20) national central banks each have their own balance sheets, they share some portion of both income and risk. The details are complex, but the result is that there are transfers among countries that depend on the relative sizes and movements of interest rates. But those transfers appear to be small and the overall consolidated losses look to be even smaller than in the US.
To conclude, our paper makes three contributions. First, we propose merging the balance sheets of the central bank and the fiscal authority into one consolidated government balance sheet. This clarifies the consequences of central bank asset purchases, emphasising that these actions have fiscal consequences.
Second, using the consolidated balance sheet approach, we can classify central bank purchases into two types – domestic bonds and foreign securities. We trace out the effects of these – a change in maturity structure resulting in potentially higher variation of tax liabilities as well as intra- and inter-country transfers.
Third, using ex-post realised losses, we are able to quantify and describe differences in losses and transfers across central banks. For this purpose, we consider four examples: the Bank of England, the Federal Reserve, the Eurosystem, and the Swiss National Bank.
In the first three, the losses are relatively small, ranging from 0.3% to 1.5% of GDP in a given year. By contrast, the Swiss National Bank is sustaining losses up to 17% of GDP. However, compared to the potential gains of avoiding more severe recessions and earlier benefits from holding high-yielding securities funded by low-interest reserves, losses seem relatively small.
Institutional arrangements also play an important role. Realised losses have different effects depending on accounting treatments and indemnification policies. If losses appear on the balance sheet only when sales are made, the timing of sales affects the timing of losses. This is true for the Bank of England, which is currently selling a substantial fraction of bond holdings every year.
In contrast, sales are not relevant for the timing of losses if they become part of a deferred asset that is then offset by future profits, as is the case for the Federal Reserve. The timing of fiscal consequences thus is related to the institutional arrangements.
Endnotes
1. For example, Humann et al (2023) discuss how central bank policies generated significant profits during disinflationary periods like the 1980s.
2. Pallotti et al (2024) highlight the unequal impact of the high euro area inflation, underscoring the importance of a more restrictive monetary policy stance in response to the high inflation that hit advanced economies in 2021-22.
3. For a comprehensive examination of how central bank and fiscal authorities handle transfers associated with profits and losses, see Chaboud and Leahy (2013).
References
Barro, R J (1974), “Are Government Bonds Net Wealth?”, Journal of Political Economy 82(6): 1095-1117.
Buiter, W (2008), “Can Central Banks Go Broke?”, CEPR Policy Insight 24.
Cecchetti, SG and J Hilscher (2024), “Fiscal Consequences of Central Bank Losses”, CEPR Discussion Paper 19088.
Chaboud, A and M Leahy (2013), “Foreign Central Bank Remittance Practices,” memo to the FOMC, 8 March (published 11 January 2019).
Greenwood, R, SG Hanson and JC Stein (2015), “A Comparative Advantage Approach to Government Debt Maturity,” Journal of Finance 70: 1683–722.
Humann, T, K Mitchener and E Monnet (2023), “Disinflation policies and central bank finances,” VoxEU.org, 12 July.
Lucas, RE and NL Stokey (1983), “Optimal fiscal and monetary policy in an economy without capital,” Journal of Monetary Economics 12(1): 55–93.
Pallotti, F, G Paz-Pardo, J Slacalek, O Tristani, and G Violante (2024), “The unequal impact of the 2021-2022 inflation surge on euro area households,” VoxEU.org, 1 March.
Reis, R (2015), “Different Types of Central Bank Insolvency and the Central Role of Seigniorage,” NBER Working Paper No. 21226.
Wessels, P and D Broeders (2022), “Central Bank Capital,” SUERF Policy Brief 420.
This article was originally published on VoxEU.org.