How much money have central banks really lost?

Central banks experienced widespread financial losses over the last three years. The problem has been most severe at banks whose balance sheets were bloated in the years of low interest rates and quantitative easing (QE). As far as the scale of the problem is concerned, however, first impressions can be misleading because central banks account for various assets and liabilities differently in reporting their net worth.

By the second half of 2022, as the whiplash effect of the inflation surge began to be felt in rising interest rates, central banks began to warn of future losses, but they communicated the state of their finances with varying degrees of candor. In their defense, quantification of the issue is not straightforward. A remarkably wide variety of accounting conventions is in use, obscuring many basic facts, especially when it comes to international comparisons. For example, some central banks use the marked-to-market method (termed "fair value”)—e.g., they list the value of the asset as its current market price, which is the amount of money you would get if you sold it on the open market. Other central banks use the book value accounting method—in effect, they report how much they paid for the asset when they bought it.

Restating the accounts to put the relevant balance sheet items on the same “fair value” (marked-to-market) basis reveals several anomalies. For example, some central banks that report positive net equity are actually under water; others report a negative equity figure that does not reflect the true value of their gold holdings. These data issues are illustrated by the four most important examples of central banks that have reported losses: the Swiss National Bank (SNB), the US Federal Reserve, the Eurosystem (comprising the European Central Bank and the national central banks of the countries that have adopted the euro), and the Bank of England.[1]

The biggest annual loss reported by any central bank has been that of the SNB, which reported a 2022 loss equivalent to 17 percent of that year’s Swiss GDP. This loss mainly reflected capital losses on the huge foreign exchange portfolio the bank had accumulated in the years of too-low inflation. Still, the SNB had built sizable balance sheet reserves in previous years and reported a positive total equity position every year. The scale of the 2022 reported loss also reflected the fact that the SNB’s accounting convention values the bulk of its portfolio at fair value; it made a huge profit in 2024.

Despite incurring heavy losses because of its sizable net interest payments, the US Federal Reserve System also continued to report positive total equity. But there is a catch. The Fed does not use the same accounting conventions as the SNB. If it did, it would have reported net negative equity of about $1.2 trillion at the end of 2024 (more than 4 percent of US GDP). Thus, although (unlike the SNB) the Fed is not allowed to build up reserves (beyond a very small amount specified in legislation), it can run up what are, in effect, very substantial negative reserves.

It is not as though the SNB uses systematically unflattering accounting conventions. The European Central Bank reported negative capital and reserves for 2024, but this figure would have been positive if it had used the same accounting conventions as the SNB. The same is true of the Bundesbank and the Banque de France, the largest of the euro area’s national central banks. But the experience of national central banks in the Eurosystem varies. Several Eurosystem national central banks had the opposite experience, reporting positive capital and reserves that would have been negative if they had used the SNB conventions.

The Bank of England reported a modest profit every year, but those figures do not take account of the losses of its Asset Purchase Facility (APF), the name given to its QE program. The APF has a separate balance sheet from the Bank of England and receives substantial indemnity payments from the UK Treasury reflecting continuing losses. Thus, although the Bank of England itself is not under water, if the marked-to-market deficiency of its APF were added to its own balance sheet, the cumulative actual and prospective losses being compensated for by the Treasury QE indemnity would represent more than 7 percent of GDP, larger than for any of the other central banks mentioned.

These accounting contrasts reflect three main aspects.

• Use of fair value: The SNB values essentially all of its financial assets at fair value (market price). The Fed and the euro area central banks do not mark their QE-related securities at fair value, using a historic book value instead.
• Use of revaluation accounts: The euro area central banks value their gold and foreign exchange holdings at fair value. Unlike the SNB, they segregate the huge unrealized capital gains on these assets in a revaluation account and do not include them in reported capital and reserves.
• Notional accounts for accumulated losses and indemnities from government: Even though cumulative losses exceed relevant reserves, resulting in a deficiency that will have to be covered before it resumes remittance payments to the US Treasury, the Fed does not explicitly report a deficiency (negative reserves). Instead, it balances the account by counterintuitively adding a notional asset “deferred asset: remittances to the Treasury.” In contrast, the Bank of England records the market value of the indemnity it receives from the UK Treasury as an asset.

Each of these central banks provides enough information to allow a pro forma balance sheet to be calculated adjusting each of these items to approximate the SNB’s Fair Value Through Profit and Loss (FVTPL) accounting approach.

The figure below shows the 2024 leverage ratios—calculated as marked-to-market capital as a percentage of total assets—for the four banks mentioned. (Data for other affected central banks are presented in my PIIE working paper [Honohan 2025]). The marked-to-market figures show a much stronger outcome than the reported equity for the Eurosystem (and for several of its main component national central banks, including the Bundesbank) and a much weaker outcome for the Fed and the Bank of England.

There is much debate about the degree to which these losses matter (Bell et al. 2023). Although the losses are a legacy of the years of easy money and QE, it would be a big mistake to regard them as a metric of the overall effectiveness of policies that were not designed to turn a profit but to achieve and cement economic recovery from the global financial crisis and the COVID-19 pandemic.

Shareholder governments grumble about multiyear interruptions in the payment of surplus income; in some cases, they have also been making indemnity or recapitalization payments to the central bank (Buiter 2024, Cecchetti and Hilscher 2024, Honohan 2023). Years ago, the flow of dividends was boosted by profits generated from the early years of QE, but (like eaten bread) these—and the wider benefits to economic activity and price stability from QE policies—have already been forgotten.

Conventional wisdom among financial economists is that despite some tentative evidence that undercapitalized central banks do not contain inflation as effectively as they otherwise might, the net equity of a central bank is at most of secondary importance, although it is generally understood that if it becomes too negative, the central bank’s theoretical ability to control inflation could be lost (Del Negro and Sims 2015). Central bankers themselves tend to fear a loss of de factopolicy independence if their equity is too low or goes negative (Archer and Moser-Boehm 2013, Bailey 2024, Goncharov et al. 2023).

Each of the central banks discussed here has been subject to specific criticism:

• The Swiss National Bank has been criticized for its conservative dividend policy, which needlessly restricts the funding of cantonal governments (Gerlach et al. 2025).
• The US Federal Reserve Banks have been criticized for insouciantly operating with what is effectively a negative net equity level, ignoring the risk that they could lose market credibility (Kupiec and Pollock 2024).
• The Eurosystem’s policy on the interest rates paid on deposits placed with it by commercial banks has been criticized as entailing an unnecessary and distorting subsidy from the public sector to the banking system (De Grauwe and Ji 2025).
• The Bank of England’s asset sales policy has been criticized for constraining the new British government’s ability to spend as much as it would like on its policy goals (House of Commons 2024, Mahon 2025).

Some of these criticisms are surely overstated, although that does not preclude their having a reputational impact.

Central bank losses can be thought of as transfers rather than a net economic cost. The sums involved are large but not so large as to destabilize the macroeconomy, and they follow years of profits. Still, some of their fiscal and distributional consequences could have been finessed with more refined policy choices, underscoring the importance of paying closer attention to dividend, recapitalization, and indemnity policies as well as to the remuneration regime for bank reserves.

It would surely help good policy design to establish a shared lens through which the scale of the losses and the underlying financial condition of the central banks can be viewed on a common cross-country basis.

References

Archer, David, and Paul Moser-Boehm. 2013. Central Bank Finances. BIS Paper 71. Basel: Bank for International Settlements.

Bailey, Andrew. 2024. The Importance of Central Bank Reserves. Speech at the London School of Economics, May 21.

Bell, Sarah, Michael Chui, Tamara Gomes, Paul Moser-Boehm, and Albert Pierres Tejada. 2023. Why Are Central Banks Reporting Losses? Does It Matter? BIS Bulletin 68. Basel: Bank for International Settlements.

Buiter, Willem. 2024. Central Bank Capital Adequacy: The Simple Analytics and Complex Politics. CEPR Discussion Paper 19057. Paris and London: Centre for Economic Policy Research.

Cecchetti, Stephen, and Jens Hilscher. 2024. Fiscal Consequences of Central Bank Losses. CEPR Discussion Paper 19088. Paris and London: Centre for Economic Policy Research.

De Grauwe, Paul, and Yuemei Ji. 2025. The High Price of the Fight against Inflation: The Case of the Euro Area . VoxEU column, April 30.

Del Negro, Marco, and Christopher A. Sims. 2015. When Does a Central Bank’s Balance Sheet Require Fiscal Support? Journal of Monetary Economics 73: 1–19.

Gerlach, Stefan, Yvan Lengwiler, and Charles Wyplosz. 2025. Equity and Profit Distribution. Report 10. SNB Observatory.

Goncharov, Igor, Vasso Ioannidou, and Martin C. Schmalz. 2023. (Why) Do Central Banks Care about their Profits? Journal of Finance 78, no. 5: 2991–3045.

Honohan, Patrick. 2023. Central Banks Are Incurring Losses, but Critics’ Concerns Are Overblown. RealTime Economics, January 18. Washington: Peterson Institute for International Economics.

Honohan, Patrick. 2025. How Much Capital Do Central Banks Really Have? PIIE Working Paper 25-15. Washington: Peterson Institute for International Economics.

House of Commons. 2024. Quantitative Tightening. London: UK Parliament.

Kupiec, Paul H., and Alex J. Pollock. 2024. Federal Reserve Losses and Monetary Policy. Washington: American Enterprise Institute.

Mahon, Christopher. 2025. Where the Bank of England’s QE Programme Went Wrong. FT Alphaville, Financial Times, April 7.

Note

[1]. Despite recent increases in Japanese bond yields, capital gains on the gold holdings of the Bank of Japan, another big user of QE, have so far been sufficient to keep its marked-to-market capital above zero.