Policy-makers in London, Washington and Frankfurt will confront a question with large consequences for public finances: who should receive the income generated by central bank money? On 28th October, John Healey delivers his first budget as UK chancellor, while Reform UK and the Greens build part of their fiscal plans on cutting what the Bank of England pays on reserves. The same day the Federal Open Market Committee meets under Kevin Warsh, who has questioned paying interest on excess reserves. On 29th October, the European Central Bank’s governing council meets. A proposal to double the share of reserves banks hold without remuneration remains under discussion.
The Eurosystem pays its 2.25% deposit rate on around €2.16tn of excess liquidity,  €48.7bn a year on Reuters’ estimate. In 2025 the Federal Reserve’s interest expenses were $167.4bn against interest income of $155.3bn, and its deferred asset reached $243.5bn.
The case for paying
The most rigorous defence of present arrangements comes from John Cochrane. He argues that interest-bearing reserves are a money-market fund invested in Treasuries with a good payment system attached. The Fed pays banks out of what its securities earn, so there is no subsidy. Banks are indifferent between a reserve and a Treasury bill, so the Fed is not paying them to sit on money they would otherwise lend. And a system awash with safe, interest-bearing reserves is a more stable one. BoE Governor Andrew Bailey has offered a central banker’s version of the same argument: banks must fund the reserves they hold, so the interest is not free money. The Federal Reserve insists that interest on reserves is not a subsidy, and in the euro area Robert McCauley and Julien Pinter have argued that ending remuneration would act as an unexpected levy on banks.
The argument is right on monetary control but wrong on who gets paid.
Reserve interest, Cochrane says, comes out of the central bank’s earnings. Yet those earnings are not the central bank’s to distribute as it pleases. They arise from the public monetary franchise and, net of the costs of operating it, belong to the public, whether the central bank retains them or transfers them to the Treasury. Paying them to banks beyond what monetary control requires is a transfer, whatever the ledger says. In work with Massimo Costa for OMFIF and The International Economy I have argued that base money is better understood as equity in the state’s monetary institution than as debt, which makes that ownership explicit. But the question of who gets what arises under conventional accounting too. The claim also needs the yield curve to co-operate. It has not since 2022 and for three years the Fed paid banks more than its portfolio earned.
Indifference has a price
Banks are indifferent between reserves and bills, but only because remuneration makes them so. The usual reply is that the banking system cannot shed reserves: the central bank determines the aggregate stock, and what one bank pays away another receives. That is true and precisely why the reply misses the point. Reserves are not extinguished when they move, but each bank decides what to do with its balance. A bank that judges its balance excess uses it to lend or buy assets, passing reserves to other banks, which face the same choice. As activity expands across the system, the ratio between activity and reserve balances moves towards the point where, for each bank, the marginal benefit of the last reserve held equals its marginal cost. The central bank sets the aggregate; each bank chooses its position within it. The need to influence those marginal decisions does not establish a right to remuneration on the entire stock.
The money-market fund analogy proves too much. If reserves deserve the market rate because they are backed by Treasuries, then so do banknotes, which are backed by the same balance sheet. Both are central bank money: riskless in nominal terms, maximally liquid and final in settlement. Yet nobody regards the absence of interest on banknotes as a tax. Reserves are an input into banking, like labour or capital, and the unremunerated portion is the price of that input. The return on the assets backing central bank money belongs to the issuer’s ultimate owners, not automatically to its holders. Paying it to those with an account at the central bank, and nothing to everyone else, is not the Friedman rule applied. It is a privilege with the Friedman rule as its alibi.
Bailey’s funding point concedes more than it intends. If banks funded their reserves at Bank Rate, remuneration would be a pass-through. The issue is the spread between what banks pay depositors and what the central bank pays them. On Paul Tucker’s reasoning, whatever competition does not pass on to depositors or borrowers is better regarded as a transfer to banks.
Cochrane’s strongest argument is that abundant safe reserves make the system less fragile. But the safety comes from the state, which stands ready to supply more reserves when markets seize. That is liquidity insurance and insurance is something the insured pays for. Paying the policy rate on every reserve works like a fire insurer that covers the house and, instead of collecting the premium, pays it to the owner. No private insurer could survive on those terms. A central bank can, because the bill falls on the public.
A better rule
None of this requires abolishing interest on reserves, as Reform UK and bills in the US Congress propose. Control of short-term rates can be kept by remunerating marginal balances without paying the policy rate on the whole stock, provided the tiering structure preserves arbitrage and the floor under market rates. Tiering has precedents in the ECB’s two-tier system of 2019–22 and the Swiss National Bank’s differentiated remuneration. Their designs and objectives differed from the one proposed here, but they show that uniform remuneration is not indispensable.
The rule follows. Give each bank an unremunerated allowance and pay the policy rate on balances above it. Paying nothing at all is not an option while reserves remain this abundant: banks would lend out their surplus at any rate above zero and the policy rate would cease to bind. The allowance works if it is large. It should cover most of each bank’s reserves, so that most of the interest returns to the public, but stop short of the total, so that the marginal reserve  still earns the policy rate. Set that way, the saving survives reserves moving between banks. The allowance should be recalibrated as balance sheets shrink, with the gap between market rates and the policy rate as the test of whether the floor holds. Surveys such as the BoE’s preferred minimum range of reserves can inform that calibration, though they measure demand for liquidity, not the allowance itself. Central banks should also publish each year how much seigniorage their balance sheets produced and who received it.
Banks will call reduced remuneration a tax. Ending a transfer is not the same as imposing a tax, and the cost of any unremunerated balance must be weighed against the monetary and public benefits it delivers. The question for London, Washington and Frankfurt is not whether central banks can afford to pay banks on their reserves. It is why the public should.
Biagio Bossone is an adviser to international financial institutions and national central banks.
Interested in this topic? Subscribe to OMFIF’s newsletter for more.