The new anti-inflation toolkit of transfers, price interventions, buffer stocks, structural resilience is credibility-intensive by construction. Its proposal for an explicit division of labour between monetary and fiscal authorities can only work if that credibility is defended jointly rather than separately. We ask what defending it jointly actually requires, institutionally, and why artificial intelligence makes the answer urgent rather than optional.
In September 2022, the Bank of England was forced to intervene in the gilt market to stop a fire sale. The trigger was not a Bank of England decision; it was a budget. For a week, the independent central bank did emergency repair work on a mess made entirely by fiscal policy, while formally maintaining that the two institutions operate in separate rooms. That week captures something the profession has been reluctant to say out loud: the wall between monetary and fiscal policy, foundational to the modern central banking canon, was never quite as solid as it looked. It held in calm conditions. It cracked the moment markets needed to know whether government and central bank were on the same page.
The doctrine most central bankers were trained on has a clear lineage – Volcker’s disinflation, the rise of inflation targeting and the consensus that independence, understood as formal distance from elected government, is what makes credibility possible. It answered a specific problem: governments leaning on central banks for cheap financing. What it was never built to answer is what happens when fiscal and monetary credibility move together, priced by the same investors, in the same market, at the same time. The BoE gained operational independence only in 1997; the ECB was built on the Maastricht separation as a founding choice. A generation raised inside that consensus tends to treat the separation as natural law rather than what it is – a deliberate fix for a 1970s problem, transplanted into a financial system it was never tested against.
The pattern is not unique to the UK or the euro area. The Federal Reserve’s own independence was itself a post-war construction, formalised through the 1951 Treasury–Fed accord after a decade of wartime bond-price pegging and revisited every time the Fed’s balance-sheet operations have needed a Treasury backstop. The specifics differ by jurisdiction and by era; the underlying tension between formal separation and practical interdependence does not.
Co-operation before the next crisis
None of this is an argument for letting governments dictate interest rates and the objection writes itself the moment ‘co-operation’ appears next to ‘central bank.’ Monetisation-by-another-name is a real risk. But the wall as currently built doesn’t prevent that risk. It relocates it from a transparent, rules-based channel markets can price, to an invisible one that reveals itself only in crisis, when it is too late to design anything.
A co-operative framework that avoids capture needs three features the old system never had to specify, because it was never meant to allow co-operation at all. First, pre-commitment: the terms of engagement are set out in advance, in normal times, not negotiated mid-crisis when one side holds all the leverage. Second, transparency: markets can observe the rule being followed, which is what lets them price it rather than panic at its absence. Finally, a sunset clause: any joint operation has defined triggers for activation and winding down, so emergency coordination cannot quietly become permanent monetary financing.
A framework along these lines has  been proposed for the BoE specifically: a standing council, headed by the finance ministry with senior central bank participation, reviewing debt issuance, inflation risk and fiscal space on a standing cadence – not a vote on interest rates, but an explicit, accountable version of a coordination that currently happens informally, through market pressure and after-the-fact improvisation. The Bank of England’s Asset Purchase Facility, indemnified by HM Treasury since inception, is a working example of the fiscal authority underwriting monetary operations rather than pretending not to notice them. Proposals for central banks to undertake conditional, rules-based debt operations extend the same logic further.
A central bank surveillance role over systemically significant prices and buffer stocks designed to target speculation rather than scarcity, both need a standing venue where fiscal and monetary authorities already share a situational picture – otherwise each new shock re-litigates who acts first.
The obvious objection is  any arrangement that brings finance ministry and central bank officials into the same room, however well designed, is one step away from becoming exactly the capture mechanism the wall was meant to prevent. That risk is real but also manageable through instruments central banks already use to police their own independence. Published minutes already let markets scrutinise the reasoning behind a rate decision; extending the same discipline to a fiscal-monetary council is a matter of institutional will, not new technology. External audit of whether pre-committed triggers were respected, conducted by an independent fiscal watchdog or parliamentary committee, would reactively perform the same function that market pricing performs in real time.
The gap is not for lack of technical capacity. It is rather a gap in mandate: nobody has been asked to apply that procedural discipline to the fiscal-monetary relationship, only to the interest-rate decision itself.
AI as the forcing function
AI turns this solution from desirable to urgent, for two reasons. First, markets: central banks are already using AI for nowcasting and reserve management, and private markets for trading and liquidity routing. Policy space itself has become more dynamic than the old separation assumed, increasingly repriced in real time by markets reshaped by algorithmic trading and AI-driven information processing. A central bank whose actual coordination with government is invisible has less time than ever to explain itself before an algorithm has already moved the exchange rate. Foreign-exchange and government-bond markets, in particular, are among the most heavily automated in the world, which means a credibility gap between fiscal and monetary authorities is now priced within minutes rather than the weeks or months it might have taken a generation ago.
Second, the real economy: AI-driven capital intensity risks widening the gap between productivity gains and wage growth. Regions with higher AI intensity  have already seen labour’s income share decline, driven mainly by wage compression among medium- and high-skilled workers even as productivity rose – a distributional problem monetary policy alone cannot fix and fiscal policy alone can rarely fund sustainably. Addressing it requires transfers, taxation and monetary stance moving together, not three institutions each treating the transition as someone else’s mandate.
A new canon for central banking would keep the independence that has served the profession well and replace the silence around co-operation with a rulebook: pre-committed, transparent and time-bound. None of this requires a single template, as the shape of the council, its triggers and its reporting lines will differ across advanced and emerging economies alike. What should not differ is the willingness to design the relationship in daylight, before markets are forced to price its absence. The wall was a solution to the last crisis. The bridge – built now on these terms – is what the next one requires.
Biagio Bossone is an adviser to international financial institutions and national central banks.
Join OMFIF on 21 October to examine Central bank independence in an era of financial instability.
Interested in this topic? Subscribe to OMFIF’s newsletter for more.