Who pays for QE losses? The invisible tax the ECB is considering for banks

The European Central Bank has not sent banks a bill. It is considering something quieter: stopping interest payments on an additional slice of their money.

Euro area banks must currently keep an average amount at their national central bank equal to 1% of certain deposits and other short-term liabilities. These minimum reserves have earned nothing since September 2023. Under the present regime, reserves above that floor are remunerated at the deposit facility rate, which has stood at 2.25% since June 2026.

On 23 July, Christine Lagarde confirmed that the Governing Council would discuss a higher reserve requirement. She also said the topic had not been considered at that meeting. According to Reuters, the option under debate would double the ratio from 1% to 2%. No decision has been taken and other options remain in play.

The stakes fit into one multiplication:

€173.561 billion × 2.25% = €3.905 billion per year

The €173.561 billion is the average minimum reserve requirement reported by the ECB for the maintenance period from 6 May to 16 June 2026. If the reserve base stayed unchanged, moving from 1% to 2% would add roughly the same amount to the zero-interest bucket. This matches Reuters’ order of magnitude: nearly €4 billion less in annual interest for banks and, symmetrically, less interest expense for the Eurosystem.

This static calculation is neither a profit forecast nor an estimate of the effect on credit. It simply reveals the transfer at the centre of the debate.

The bill in three moves

The story starts during the years of very low rates.

1. The Eurosystem buys bonds through the APP and PEPP. At the end of 2025, these portfolios still held €3.745 trillion of securities.
2. Settlement of those purchases creates reserves in the banking system. The reserves sit on the liability side of central bank balance sheets.
3. From 2022, policy rates rise rapidly. The old bonds continue to pay low coupons, while the cost of bank reserves resets almost immediately with the policy rate.

In short, the asset is slow and the liability is fast. The central bank bought long-dated, low-rate securities, then financed them with an overnight liability that became expensive.

The Bundesbank offers the clearest version of the problem. Its net interest income improved from -€13.1 billion in 2024 to -€4.2 billion in 2025. Its annual loss fell from €19.8 billion to €8.6 billion, but accumulated losses carried forward have reached €27.8 billion.

The accounts of the ECB itself are separate from those of the Bundesbank and other national central banks. The ECB lost €1.254 billion in 2025, down from €7.944 billion in 2024. Its net interest expense almost vanished, falling from €6.983 billion to €178 million, partly because average rates were lower and the balance sheet contracted. These figures should not be added together as if they described a single company.

The €4 billion switch

To understand the 2% option, picture two buckets in a bank’s account at the Eurosystem.

• The first holds minimum reserves. It earns 0%.
• The second holds the surplus. It earns 2.25%.

Changing the reserve ratio from 1% to 2% widens the first bucket and shrinks the second. In a snapshot where total reserves do not change, the Eurosystem pays less interest without selling a single bond or changing its policy rate.

The word tax needs a qualification. Legally, this is not a tax voted and collected by a government. Economically, forcing a bank to hold an asset at 0% when a neighbouring asset earns 2.25% creates an opportunity cost. Research from the Bank for International Settlements notes that unremunerated reserve requirements are traditionally modelled as a tax on bank intermediation.

The ECB already accepts the principle. When it cut remuneration on minimum reserves to zero in 2023, it said the move would reduce the total amount of interest needed to implement monetary policy. The 2026 debate is about the size of the bucket.

Banks cannot make aggregate reserves disappear

The word “locked” can mislead. A bank must meet its requirement on average over a six-to-seven-week maintenance period. It can move reserves from one day to the next and lend them to another bank. More importantly, the banking system as a whole cannot turn reserves into household loans at the click of a button. Banks can transfer reserves among themselves, but only changes in the Eurosystem’s balance sheet can reduce their aggregate amount.

With a little more than €2 trillion of excess liquidity still in the system, a move to 2% would initially be a change in price. Reuters reports that only a few banks would need to raise additional cash, while most already hold more reserves than the proposed level.

That conclusion will not hold forever. The ECB projects reserves to decline by roughly €470 billion per year as its portfolios mature. Its survey of bank treasurers suggests that institutions representing half of euro area banking assets could approach their preferred reserve levels by the end of 2026. A requirement raised in an ocean of liquidity can become more binding as the water level falls.

This is why the equation “€174 billion more in required reserves means €174 billion less credit” is wrong. The funding effect would depend on reserve distribution, the cost faced by banks that need to acquire reserves, their margins and their commercial choices.

The chain of possible payers

The first payer is identifiable: the commercial bank loses interest on the reclassified slice. What follows is not automatic.

A bank can absorb the shortfall in earnings. It can also seek to protect profitability through several channels: lower deposit remuneration, wider margins on some loans, higher fees or reduced spending. Shareholders may receive less profit. No public evidence currently allocates the €3.9 billion among these outcomes.

Differences across banks would matter. A deposit-rich bank with abundant reserves is not the same as an institution dependent on market funding. National deposit markets are also highly uneven. The ECB’s May 2026 Financial Stability Review places the return on equity of large euro area banks near 10% in 2025, while highlighting differences in margins and deposit competition.

Depositors are a plausible candidate for sharing the bill, not a certainty. The ECB’s own research finds that sight deposits respond only slowly and unevenly to policy rates. That weak pass-through supported bank profits after rates rose. It identifies a possible channel, not proof that a higher reserve requirement would be passed to customers euro for euro.

At the other end of the chain, the Eurosystem’s accounts improve. That can shorten the period during which national central banks stop paying profits to their governments. The ECB made no profit distribution to national central banks for 2025. Yet one euro of avoided expense is not immediately one euro for a public budget: it may first reduce a loss carried forward or rebuild a buffer.

Does Frankfurt need the money?

Two readings can be true at once.

The first is monetary. The ECB announced in March 2024 that it would review the parameters of its operational framework in 2026. Minimum reserves create stable demand for central bank money, help steer short-term rates and may matter more as excess liquidity contracts. The 1% ratio and zero remuneration were retained pending that experience.

The second is fiscal. A higher ratio would directly reduce the Eurosystem’s interest bill and losses at some national central banks. Reuters reports that this motive divides governors: to some, using a monetary instrument to repair accounts looks more like a fiscal choice.

The financial need is not operationally urgent. The ECB says it can fulfil its mandate regardless of its results and reported €71.4 billion of capital and revaluation accounts at the end of 2025. The Bundesbank points to €363 billion of net equity, largely supported by the revaluation of its gold. A central bank does not fail like a commercial bank in the currency it issues.

The losses are already fading. The ECB expects to return to profit in 2026 or 2027, depending on interest rates, exchange rates and the composition of its balance sheet. APP and PEPP holdings shrink as bonds mature. The remunerated liability declines with them.

That does not make the interest paid to banks irrelevant. The ECB says it and the national central banks earned about €300 billion of profits between 2012 and 2021. Some strengthened their buffers, while the distributable balance of national central bank profits generally supports public budgets. Cutting an unnecessary expense can protect central bank independence and limit future calls on taxpayers. But the efficiency argument should remain separate from the desire to erase an embarrassing loss more quickly.

The test that settles the debate

One question provides a useful test: would the ECB still choose 2% if the Eurosystem were reporting record profits?

If the answer rests on better-calibrated reserve demand, more robust rate transmission and a framework adapted to scarcer liquidity, the increase can be a coherent monetary reform. The ECB should then publish the effects by country and business model, the timetable, the expected money-market impact and the review conditions.

If the main justification is saving nearly €4 billion, the mechanism looks more like a levy on banks. It may still be defensible, particularly after several years of strong bank profitability, but it should be presented as a distributional choice rather than a plumbing adjustment.

The QE bill did not vanish when bond purchases ended. It remained in balance sheets as slow-moving bonds and reserves that had become expensive. Moving from 1% to 2% would not erase it. At today’s rate, it would shift roughly €4 billion per year from the commercial-bank column to the Eurosystem column.

The final payer would still be unknown.

---

Sources

• ECB, press conference of 23 July 2026, question on minimum reserves: Christine Lagarde confirms a future discussion and says the topic was not considered at that meeting.
• ECB, minimum reserve statistics, maintenance period from 6 May to 16 June 2026: €173.561 billion average requirement, remunerated at 0%.
• ECB, monetary policy decision of 23 July 2026: 2.25% deposit facility rate and passive run-off of APP and PEPP.
• ECB, March 2024 operational framework: 1% reserve ratio, 0% remuneration and review of parameters in 2026.
• ECB, decision of 27 July 2023: minimum reserve remuneration reduced to 0% from September 2023.
• ECB, Annual Accounts 2025, sections 1.2 and 1.3.3: portfolios, rate-risk mechanism, loss, net interest expense, buffers and expected return to profit.
• Bundesbank, Annual Accounts 2025: 2024 and 2025 losses, net interest income, accumulated losses and net equity.
• ECB, how banks are adjusting to declining reserves, 2 April 2026: projected pace of decline and banks’ preferred reserve levels.
• ECB, Financial Stability Review, May 2026: bank profitability, margins and differences across deposit markets.
• ECB, explainer on central bank profits and losses, updated 19 May 2023: budget transmission, buffers and operational capacity.
• Reuters, the 2% option and nearly €4 billion calculation, 30 June 2026, and alternative options and the monetary-versus-fiscal debate, 23 July 2026.
• Bank for International Settlements, “Reserve requirements as a financial stability instrument”, Working Paper 1182, April 2024: cost of unremunerated reserves and their economic treatment as a tax.

Method and limitations

• Information cut-off: 13 August 2026. No increase to 2% had been decided. The scenario was reported by Reuters, while Christine Lagarde confirmed that the issue would be discussed.
• The €3.905 billion calculation is a static order of magnitude: €173.561 billion × 2.25%. The reserve base, policy rate, volume and distribution of reserves would change after any decision.
• The €173.561 billion and roughly €2.16 trillion excess-liquidity figures do not describe exactly the same date. The second amount and nearly €4 billion cost are Reuters calculations; the visual is not a consolidated balance sheet.
• Lower interest expense for the Eurosystem does not immediately become profit or a budget transfer. The ECB and national central banks have separate accounts.
• The article identifies possible channels to deposits, loans, fees or shareholders. It does not estimate their future distribution or claim a causal effect on credit supply.
• Banks hold very different reserve volumes and funding structures. An aggregate result does not describe the effect on each institution or country.
• For the full context on the balance sheet, APP and PEPP portfolios and TARGET, see l0g’s guide to reading the ECB balance sheet and TARGET2.