Supplement 20 Sep 26 - The Bonfire Burns Lower, Trump's Stooge Hikes, and the £107.7 Billion Receipt

"Money will not manage itself, and Lombard Street has a great deal of money to manage." - Walter Bagehot, Lombard Street, 1873

The quote pertains to the Deep Dive, and to a week in which the Bank of England quietly tore up four years of its own policy on a Thursday lunchtime and hoped nobody would ask what the first four years had cost. Bagehot wrote the book on what a central bank is for, and that line is the one that has always stuck with me, because the second half matters as much as the first. A great deal of money to manage. Nearly nine hundred billion pounds of it at the peak, managed on behalf of people who were never shown the bill. This week we are going to look at the bill.

As we run into the last quarter of a year that has had more in it than most, our next Property Business Workshop is live and the clock is now properly ticking. Thursday 1st October, the Hyatt Regency in Manchester, with myself and Rod Turner, powered by Roma Finance, who do bridging, development and term lending and who know what they are talking about. JVs and M&A is the subject, which after the last fortnight in the gilt market feels less like a topic and more like a survival skill. That is eleven days away. You can still squeeze into the VIP dinner afterwards and get some dedicated time with myself and Rod if you are quick. Book in at www.tinyurl.com/pbwoct26

Welcome back to Trumpwatch. Last week was the five thousand dollar promise. This week the President got something he has wanted for years, which is a Federal Reserve run by a man he picked himself. And on Wednesday the man he picked put interest rates up.

Let us just sit with that for a second, because it is the best piece of political theatre of the year and it deserves a moment. Trump told NBC back in February that Kevin Warsh would not have got the nomination unless he wanted lower rates. Trump has his man in the seat. And his man walked into his third meeting as chair, got all twelve voters into the same room, and came out with a unanimous quarter-point hike to a range of 3.75% to 4%. The first increase since July 2023. The July meeting had three regional presidents dissenting in favour of a hike and Warsh calling it a good family fight. Seven weeks on the family has stopped fighting and agreed with the dissenters.

Warsh's own description was that it was "a sober decision, a serious decision, a responsible decision". The projections that came with it have the median voter pencilling in one more before Christmas, four of them wanting two, and only two wanting to stop here. Warsh, for the second time running, declined to submit a forecast of his own. I rather admire that - even though I’m not sure it is wise. If you think forward guidance is a bad habit, the way to give it up is to give it up, and you lead from the front. He had tipped off the Prez that the others were voting to hike unanimously, and clearly decided to save his dissent for when he at least has a few other dissenters on board. Also - dissent in the Fed vote is not welcomed as widely as it is in the Bank of England vote…..

The White House called it an "unfortunate decision". The President went to Truth Social in capitals and to a rally in North Carolina in person, and his line was that American interest rates should be 1% or less because the country is the best credit in the world. It is not, as it happens - mostly because of their pure, unadulterated arrogance in constantly spending more than they can bring in, for ever and a day, with no clear plan or strategy to actually address the deficit. They are not fixing the roof when the sun is shining - imagine what happens when a hole appears? Several countries carry a higher credit rating, Germany and Canada among them, mostly due to their conservative attitude towards fiscal responsibility, but I doubt that detail troubled the crowd. What interested me more is who he blamed. Not Warsh. He went out of his way to say he has confidence in Warsh, and pinned it on a "hostile board" his chairman could not overpower. That rhetoric works today, but how long will he entertain it for? He isn’t known for his patience or staying power, let’s face it…..

Now, is this loyalty, or is it a man who has worked out that he cannot sack two Fed chairs in one term without the bond market taking a view? I suspect the second, although I would not rule out a bit of the first. Volte face will come - it is a matter of when, with Trump, not if. I would imagine in the phone call Warsh was told he’s got 3 months to show real “results” with a committee that still believes it is independent, and actually packed with sensible, economically literate operators who are no doubt not exactly fans of Mr Trump. Either way it is the most important thing that happened in American policy this week, and I would argue the most encouraging thing that has happened to the institution in eighteen months. A twelve-nil vote, from a Trump appointee, seven weeks before the midterms, with the President shouting for 1%. If you were worried that the Fed's independence had gone, and I have worried about it out loud in this section many times, this was about as clear an answer as you could have asked for. Although some is theatre here, I have no doubt - but the current “plan” - or lack of it - won’t work, so what will be changed?

Why did the other 11 do it? Because the numbers left them nowhere to hide. American inflation printed 3.4% for August with wages growing at 3.1%, which means the average American worker is getting poorer in real terms in an election year, and petrol (gas) is up something like 45% since the war started. Republican-leaning podcasts are talking of a crash if diesel hits $8, let alone $10, for a gallon - Warsh was blunt about the limits of his own power, which I rather liked: the Fed can put rates up, he said in so many words, but it cannot reopen the Strait of Hormuz.

Which brings us to the oil. Over the weekend before last an Iranian cargo vessel was attacked near the Strait and at least one crewman was killed. Asked whether Washington had a hand in it, the President said "I don't want to say", which in my experience of people who do not want to say tends to narrow the possibilities somewhat. Either that or he knows that’s the best sort of answer, a “maybe” and a side-eye with a wry smile half-hidden. A meeting that Oman had been due to host was postponed. The Saudi east-west pipeline, which is the main route round the Strait, has been shut since the 11th. Brent went back through $108 on Monday, and the Bank of England's own minutes have it at $106 at the close on the 14th, up 36% since the run-up to the July forecast, with UK wholesale gas up 78%. $104 as I write this isn’t much different. The President told the rally that oil will come down very quickly once the war ends. I am sure it will. It is the "once the war ends" part of that sentence that is doing the heavy lifting, because no-one sees the way out at the moment unless Iran truly is far weaker than it looks.

Standard Fare on the diplomacy, then, which is to say none of it is going anywhere fast and the market has stopped pricing in a quick resolution. The Bank's minutes say the same thing in politer language: most members now put more weight on energy prices staying higher for longer.

So why does any of this matter to somebody with six houses in Derby? Three reasons, and they all land in the same place.

First, this was a week in which three of the big central banks moved or nearly moved in the same direction. The Fed hiked. The Bank of Japan hiked. The Bank of England held, but with three of nine voting to hike and the Governor saying in his own words that policy may have to tighten if the conflict drags on. When the world's central banks tighten together, nobody's bond market gets a free ride, and ours less than most.

Second, there is a small but symbolic thing that happened on Wednesday night which I have not seen anybody remark on. The top of the Fed's range is now 4%. Our Bank Rate is 3.75%. For most of the last two years we have been the ones with the higher rate. That gap matters for sterling, sterling matters for the price of everything we import, which is priced in dollars and includes the oil, and so it feeds straight back into the inflation number that the three hawks on our Committee are already worried about. I am not convinced it moves the dial much on its own. It does make November's meeting more interesting than it already was.

Third, and this is the bridge to the rest of this week's Supplement: Warsh has set up five task forces at the Fed to go back over how the institution does its job, including one on the balance sheet. An American central bank chair, appointed by this President of all presidents, is asking out loud whether the way the Fed has used its balance sheet was clever. In the same week, our own central bank answered the same question about itself by ripping up its plan. I said in this section back in August that it might well be time to dust off the yield curve control explainer. This is the week. It is in the Deep Dive, and I have gone to town.

Treasury yields rose on the decision, by the way, with the ten-year pushing on towards 5%. By the time a move like that has finished being a US move it has already arrived here. We will see what it did to the gilts in a moment. And the five thousand dollar cheque? Still promised, still uncosted by the people promising it, still conditional on how you vote. Likely outcome between now and November - nothing at all, except more promising.

Phew. Back we go to the comfort blanket of the real time UK property market, where at least the people behaving irrationally are only doing it with asking prices. As is customary, Chris Watkin has been relentlessly crunching the portal numbers and publishing them at Property Industry Eye, and he is back from his break with a full stats show for Week 36 of 2026, this time with Bryan Mansell in the other chair. After last week's Bank Holiday regional special it is good to have the national series back, because I left a prediction hanging last Sunday and I would like to see how it got on.

Supply first. 36.7k new listings in Week 36, up from 31.7k in the Bank Holiday week, against a 2026 weekly average of 36k and a ten-year Week 36 average of 35k. So a touch above normal, which is what early September is supposed to look like. Year to date that is 1.299 million homes brought to market, 0.5% below last year, 2.6% ahead of 2024, and 9.7% higher than the 2017 to 2019 average. My ready reckoner of "10% more stock than a normal market", which I have declared broken at least twice this year and parked altogether last week, turns out to be sitting almost exactly on its number. I am not sure whether to be pleased or embarrassed. Probably both.

Demand. 23k homes sold subject to contract, up from 18.7k in the holiday week. The ten-year Week 36 average is 24.5k and the 2026 weekly average is 24.3k, so we are running at about 94% of a normal week. Year to date, 874k sales agreed: 7.3% down on last year, 1.6% down on 2024, 11% up on 2023 and 5.3% above the pre-Covid years. Chris made the point himself in the comments and I would not improve on it: not booming, not collapsing, just a market that continues to transact. Net sales came in at 17.7k against a ten-year average of 18.3k, with 676k for the year, which is 6.1% below 2025 and 3.4% above the pre-pandemic norm. The Five Ds do not take a year off. Death, debt, divorce, downsizing and diddy ones turn up in the numbers whatever the five-year swap is doing.

Now the bit I was waiting for. I said last Sunday that a four-day week had piled on stock the market would have to work off, that I thought it would do so by reducing, and that September's reduction count would be a big one. Week 36: 28.1k price reductions. For context the August weeks ran at 20.9k, 21.2k and 22.5k, and the first week of July was 25.8k. So that is the biggest weekly number in a good while, by a distance, and I will allow myself a small nod before adding the obvious health warning that it is one week, that early September always sees a catch-up, and that August as a whole saw only 11.1% of stock reduced against 13.7% in July. The year-to-date average is 12.8% against a long-run 10.9%. I would want to see two or three more weeks north of 25k before I called it a trend rather than a clearing of the throat - this is simply a yardstick of exactly what is expected to happen, happening - we just see by how much.

And then the intergalactic number. The gap between the average asking price of what came to market in Week 36 and the average asking price of what went under offer: 29.2%. £456,000 against £353,000. The ten-year norm is 16% to 17%. Same two health warnings as last week, and I will keep giving them until I have settled the definition properly, which I still have not. These are different baskets of houses, not a discount on the same house. And this is not the old intergalactic series, so do not line it up against the numbers from the spring. With that said, what does a national 29.2% tell you? It tells you what comes to market in the first proper week of September: the big stuff. The family houses that were held back over the summer, brought on at the prices their owners were promised in July. London vendors, with continually overinflated valuations from the worst-performing agents in the country by percentage, without enough conviction or gumption to go to market at a realistic price rather than a fantasy one - and without the expertise or research to realise that their one chance to make a first impression is being wasted. Meanwhile what is actually selling is cheaper and further north, as it has been all year. The gap is part mix and part fantasy. How much of each? I honestly could not split it for you, but I would be surprised if fantasy was less than half, this week.

Here is this week's friction arithmetic. Hold tight, because we get nerdy (I know, regular readers are in no way surprised). The sell-through rate for August was 13.3%: of every hundred homes on agents' books at the start of the month, thirteen went under offer. July was 14.2%, June 13.8%, and the pre-Covid average was 15.5%. Compound that across a typical twelve-week sole agency and you get one minus 0.867 cubed, which is a 35% chance of finding a buyer before the agreement runs out. At the pre-Covid rate it would be 40%. At July's rate, 37%. So the odds for a seller walking into an agent's office this week have drifted from two in five to about one in three, and that is before fall-throughs, which ran at 25.1% in Week 36 against a decade average of 24.5%. Multiply the two and the chance of listing today and actually completing off the back of that first agreement is roughly one in four. Rough working, as ever, and it treats every house as average when none of them are. But if you are a vendor, that is the number to have in your head when your agent suggests testing the market at the higher figure. And if you are a buyer with your finance lined up, it is the reason a sensible offer in month three gets a very different reception from the same offer in week one.

August's exchanges and withdrawals bear that out. 77.1k exchanges, 63.2k withdrawals, so 55% of homes leaving agents' books did so because they sold and 45% because the owner gave up. That is actually better than July, when it was near enough fifty-fifty, and not far off the seven-year average of 57.6%. Stock on the market fell from 767k on the first of August to 742k on the first of September. Some of that is sales. A fair chunk is the late-summer clear-out of the no-hopers, which I flagged last week and which the agents will tell you, quietly, is the healthiest thing that happens to their window all year. It is still 742k against 736k a year ago. The sales pipeline sits at 479k against 508k last September, and that 29k gap is this year's transaction shortfall in one figure.

The number that matters most is the one Chris always saves for the end. August's agreed sales averaged £338.41 per square foot. That is 0.3% up on a year ago, which with inflation at 3.1% is a real-terms fall of getting on for 3%. More to the point it is down from £345.41 in July and from a record £350.22 in June. That is 3.4% off in two months. Chris's own read, given how closely this series tracks the Land Registry five months later, is that the official index will show a drop of around 1.8% to 2.1% between now and March (down towards zero growth, to be clear, from the 2% or so that the ONS HPI has hovered around for a few months). I think that is about right - my current forecast for house prices in 2027 (you heard it here first) is 0%-0.5% capital growth with the North and Midlands only managing to balance out continued falls in London prices. This is what every other dataset has been telling us for the best part of three years, but everyone else is too scared to say it out loud (although remember Savills did change their forecast to 2% down for 2026, which is wrong, for the reasons I’ve outlined above and at the time - but to give them their due, they have said it).

Rents, quickly. Chris has the average advertised rent in September so far at £1,766 a month against £1,791 last September. That is down 1.4%, the first year-on-year fall I can remember in his series, and yet available rental stock is 302k against 357k a year ago, a fall of more than 15%. Less stock and lower asking rents, at the same time? That is not how it is supposed to work, and I am not entirely sure what is going on. My best guess is mix again: London, where rents are highest, is where new lets have softened most, and a smaller share of London in the basket drags the average down. New rental listings in August were 120,100 against 115,584 a year earlier, so more is coming through the door while less is sitting on the shelf, which says it is letting quickly. We will come back to this in the macro section, because the official rent figures this week point the other way. A better way to look at this might be ex-London or non-London rents, and I will put this point to Chris when I speak to him in the coming week.

The £64k question, then. If reductions are rising, the asking gap is at 29%, and price per square foot has come off 3.4% since June, is this the start of something nastier? I do not think so. The withdrawal rate improved. Net sales are above pre-Covid. Fall-throughs are normal. What I see is a market doing what it should when the cost of money rises by 85 basis points in a year: adjusting on price, slowly, at the top end first, without any forced selling to speak of. That is orderly. It is also not over, and with the five-year swap where it is I would expect the square foot number to have another couple of per cent in it before Christmas. I might be wrong. I was early last time. People’s confidence - which was back from mid-2024 or so - that the market was stabilising, that rates were coming down nice and slowly - has evaporated, because people look at the evidence. Rent pressure has also eased significantly (in terms of how quickly it is going up) - however, rents are relatively high and affordability ceilings have been reached in more postcodes than in the past 20 years. 

Chris - this is my weekly appreciation paragraph. Thanks for what you do! If you want some help positioning yourself as a local market expert - as an estate agent or any form of property professional - give Chris a shout! Either way give his channel www.youtube.com/@christopherwatkin a follow and some love, please! The local focus this week is Weymouth, and the way he and Bryan pull the agents there apart is worth your time wherever you invest. 

Dust off the Macroscope, then. This is meat week, folks, and there is no choosing to be done. The labour market report on Tuesday. Inflation on Wednesday. The official house price and rent figures the same morning, which in a normal week would be the headline and this week barely got a look in. And bringing up the rear, as the contract states, the gilts and the swaps - with the small matter of a Monetary Policy Committee vote, a printed forecast of mine to own up to, and a thirty-year yield that finally went the right way.

The labour market, and as always we do it in three parts rather than one, because the unemployment rate on its own tells you about a third of the story. For May to July: the employment rate for 16 to 64 year olds was 75.1%. The unemployment rate for everybody 16 and over was 4.9%. The inactivity rate for 16 to 64s was 20.9%. If those look familiar it is because they are identical to last month's. All three. Not a tenth of a point of movement anywhere. For those new to the arc I track: we started this journey at 74.6% employment, the pre-Covid aspiration is 76.5%, and at 75.1% we are about a quarter of the way up the hill and have stopped for a sit-down. On the year, employment is down a tenth, unemployment is up two tenths from 4.7%, which is 83,000 more people to 1.78 million, and inactivity is up 12,000 to 9.12 million. And the usual reminder that the three rates add up to more than a hundred because unemployment is measured over a different age range to the other two, a quirk I have moaned about for years and will no doubt moan about again.

So nothing happened? Not quite. The underlying detail is where three different official sources tell three different stories, and I do not think I have seen them this far apart. The Labour Force Survey says employment rose by 243,000 over the year. HMRC's payroll data, which counts actual payslips, says the number of payrolled employees fell by 84,000 over the year and by 39,000 over the latest quarter. And the workforce jobs series says jobs are up 61,000. Up a quarter of a million, down eighty-odd thousand, or up a bit. Take your pick. Some of that gap is self-employment, of course, which the payroll data cannot see, and the statisticians note that part-time self-employment has been rising. But the Bank's minutes this week dropped in a detail I had missed: an operational error cut the survey's achieved sample in May and June, so more of those months than usual has been filled in by imputation. Which is a polite way of saying estimated, or more estimated than usual. I have said many times that I lean on the payroll number because a payslip is a payslip, and I have not seen anything this month to change my mind. On that measure the jobs market is still shrinking, slowly.

Vacancies fell again, to 702,000 for June to August, down on the quarter and the year and well below where we were before the pandemic. There are now 2.5 unemployed people for every vacancy. And pay. The Bank's minutes have private sector regular pay growth at 2.9% for the three months to July. Hold that 2.9% against the inflation figure we are about to come to, which is 3.1%. On the headline measure private sector pay is no longer keeping up with prices. Similar to the US situation outlined earlier. The Bank reckons that figure is flattered downwards by who is in the sample and that underlying growth is nearer 3.5%, and they may well be right. But it is a long way from the 5% and 6% numbers that were keeping the Committee awake a couple of years back. Will that translate into anything in the rental numbers? It should, eventually. Rents cannot outrun wages forever, and I have been saying so since 2023. Whether eventually means this winter, I am less sure. With a horrific energy price cap looming for January (pending Burnham and Healey intervening).....it looks difficult for tenants.

OK. Inflation. The Consumer Prices Index rose 3.1% in the year to August, up from 2.9% in July. Prices rose 0.5% in the month alone, against 0.3% in the same month last year. That takes us back over 3%, which means the Governor has had to write to the Chancellor explaining himself, and the letter went out with Thursday's decision. Was it a surprise? Not to anybody who has bought petrol lately. The statisticians have unleaded averaging 161.3p a litre and diesel 181.8p, with motor fuels as a whole up 23% on a year ago. That’s not real time as you will see when you fill up the car or drive past the petrol station, smugly, in your electric vehicle - £1.72 and £1.95 are the prices according to RAC Fuel Watch as I write this now. Transport made the largest upward contribution to the inflation number, clothing and footwear went the other way, and according to the Bank about 0.7 of the 1.1 points by which we are over target is energy, most of that at the pump.

Now the part that matters more than the headline. Core inflation, which strips out energy, food, alcohol and tobacco, was unchanged at 2.6%. Services inflation was unchanged at 3.4%, and it was 4.5% in March. Goods inflation jumped from 2.2% to 2.7%. So this is an oil story sitting on top of an economy where the domestic pressure is still easing. That is the good news. This is the difference between now and the inflation period of 2021-23 that I spent so much time and effort talking about. The core is refusing to be broken (I know 2.6% is above target, but it is an acceptable inflation rate and not dissimilar to the average rate of inflation in the 2010s, for example). The less good news is where the Bank thinks it goes next: around 3.75% by the end of the year, against the 3.2% it expected as recently as July, and a touch over 4% in the first quarter of next year. The energy cap for October to December has been set at £1,723 and the Bank expects it to go up substantially again in January (24% is the current number as we plough through the forecast period). 

Is there a world in which that simply washes through, as the doves on the Committee hope? There is, and the evidence so far is on their side: the Bank's Agents now expect food inflation to end the year around 4% where in April they feared 6% or 7%, and firms do not seem to have the pricing power to pass much on. But 4% inflation in January is 4% inflation arriving at the exact moment next year's pay rounds are being settled, and that is the hawks' whole case in one sentence. For what it is worth the index measure that includes housing costs was 3.3%, with owner-occupiers' housing costs alone contributing 0.68 of a point. A year ago this month the headline was 3.8%. We have come a long way down. We are now going some of the way back up, and I remain as unconvinced as I was years ago that the Bank and the forecasters have a proper handle on it. 

Next up, the official house prices and rents. Average UK house prices rose 1.4% in the year to July, to £273,000, down from a revised 1.5% in June and the third month in a row that the annual rate has slowed. England was up 1.1% to £293,000, Wales 2.6% to £215,000, Scotland 2.3% to £196,000. The North East led the English regions at 4.9%, up from 4.0%. London fell 3.3%, and has now fallen on the year for eleven months in a row. The statisticians put this month's slowdown mainly down to a sharp slowing in the South West, with London and the West Midlands chipping in. That last one made me sit up, being local, and it is one to watch rather than to panic about. Remember too what this index is: completions, registered, from sales mostly agreed in the spring. It is the market of five months ago. Set it alongside Chris's per square foot number above and you know the direction of the next five prints already. With inflation at 3.1%, a 1.4% rise is a real-terms fall of 1.7%. I have said for a long while that this cycle's correction would come through inflation doing the work rather than through falling nominal prices, and outside London that is still what is happening. I still bang my head against a brick wall occasionally on LinkedIn with the permabears who are calling for a “crash” - those who don’t even bother to look at inflation adjusted, or wage adjusted house prices. Those who believe the nominal prices forced down our neck in the Guardian are the yardsticks we should be using. They aren’t. 

Rents are the more interesting half. Average UK private rents rose 3.8% in the year to August, to £1,400 a month, up from 3.7% in July and 3.3% in June. England is at 4.0% and £1,459, Wales 4.3% and £846, Scotland 1.1% and £1,013. The North East and the North West are the strongest at 5.8% each, and the South East the weakest at 3.0%. So after two years of slowing, official rent inflation has turned back up for the third month running. How do I square that with Chris's advertised rents being 1.4% down on the year? Different things. The official series measures the whole stock of tenancies, including every renewal and every long-standing tenant whose rent is only now catching up with the market (about 77% is existing rents, 23% is new rents). The portals measure new lets only. New-let rents stopped rising a while ago, the stock is still catching up to them, and with available stock down 15% I would not bet on new-let rents staying soft for long. Zoopla's September report, for a third opinion, has new lets up 2.6%, supply 3% lower than a year ago and 5.3 enquiries per property, the most in nearly two years, and expects 4% to 5% by Christmas. The thing I would hold on to: rents at 3.8% are running ahead of inflation at 3.1% and well ahead of private sector pay at 2.9%. That ordering is not sustainable for long. It is also not the landlord's doing. It is what happens when supply leaves.

Moving on. Gilty, or not Gilty? For once, slightly less Gilty, and the reason why is the subject of the whole Deep Dive.

The Bank first. Bank Rate held at 3.75%, where it has been since December, on a vote of 6-3. Greene, Mann and Pill voted to raise it to 4%. For the record, last week I put a 5-4 hold as the most likely outcome at 55%, a 5-4 hike second at 35%, and a 6-3 hold as the saver at 10%. The hold was right. The split was the saver. If you followed this, and had bet hard money on it (please don’t do that, by the way) then you got your money back. This is as wrong as I’ve been on the “correct score” market for years. It’s a reminder to me just how slowly central bankers like to do things. 

Do not mistake 6-3 for comfort, though. Read the six. Bailey wrote that if the conflict persists, as appears to be the case, it is likely that policy may have to tighten. Lombardelli said the case for a rise is building. Breeden and Ramsden said much the same in more words. That is four of the six holders telling you, in writing, which way they are leaning. Only Dhingra and Taylor sound settled, and Taylor's paragraph had the line of the day: the curve is now higher at every maturity from two years out than it was at the last hike of the 2022 to 2023 cycle, when inflation was 6.8%. His point is that the market has already done the tightening for them. He is not wrong - but dislocating the Base Rate from SONIA is not a functional way to approach a market either. The Bank's own figure is that a two-year fixed mortgage is quoted around 95 basis points higher than before the war, with Bank Rate not having moved at all. The market curve has Bank Rate peaking at around 4.9% by the end of next year (that’s too bearish for me, but Trump makes everything so much harder to predict). I don’t see a need to move above 4.5% in the next 12 months - but let’s see. The next meeting is 5th November, a week after the Budget, with a fresh set of forecasts. If I had to write it down today I would have November as a genuine coin toss. I will firm that up nearer the time.

The gilts. The five-year closed at 4.914% yield, which is more or less where it started at 4.902%. The ten-year closed at 5.2967% on Friday, having dropped more than ten basis points on Thursday and given some of it back after retail sales came in at plus 0.5% for August when a fall was expected. That is 57 basis points higher than a year ago. And the thirty-year, which closed last week at 5.910% having touched a 28-year high and within an intra-day whisper of 6%, closed at 5.7561%. The five-year did not move and the thirty-year fell fifteen. That is the curve flattening from the long end, and the cause is no mystery: the Bank announced at noon on Thursday that it will never sell another long gilt, and the long gilts went up. Reuters had it as the biggest one-day fall in the thirty-year yield since April.

Why did the short end not join in? Because everything else this week pointed the other way. Three votes for a hike here, an actual hike in Washington, another in Tokyo, inflation back over 3% and the shops busier than expected. The five-year is priced off where Bank Rate is going, and the message on that was not friendly - it’s much more sensitive to the proxy for the 3-month cost of money, the Base Rate, than the 30-year - because so much more of it (as a percentage) is determined by that 3 months.

The swaps, which is what your lender prices off. Last week's 4.67% was up 31 basis points in a week and 85 points in a year. We had a 12bp haircut back to 4.55%, which is still outside of a comfort range for me, and my forecast from last week stands: limited company five-year money at 6.5% in six months, as a coin toss with a lean. Nothing this week moves me off that. The relief was all at the long end, and nobody reading this borrows for thirty years. It matters to the Chancellor, to pension funds, and to anybody trying to make a build-to-rent appraisal stack. It matters to the rest of us only indirectly, through what it does to the Budget arithmetic on 28th October. But indirectly is not the same as not at all. If the long end of the curve has been carrying a penalty for four years because of a decision made in Threadneedle Street rather than anything to do with the economy, then every one of us has been paying a slice of it. How big a slice? Well. Let’s see - and I’ll try to stay objective rather than get worked up, because this subject really does work me up into a near-frenzy.

Here comes the Deep Dive. One subject this week, four ways in, and regular readers will have seen it coming from the moment the Bank's announcement landed on Thursday lunchtime. I wrote "The QT Vanity Bonfire" in August last year, came back for a second go in October, and first got properly stuck into it the year before that. My position has not moved an inch: the way the Bank of England has gone about quantitative tightening is an intergalactic mistake, it has cost us a fortune, and almost nobody outside a few hundred people in the City could tell you what it is. This week the Bank rewrote the whole plan. Unanimously. Now it nearly looks like the one I laid out last year (although not as aggressive as I would like, nor will it ever be). So we have the Bank's own package; then the receipt, which is the Asset Purchase Facility's own accounts; then the hidden cost, which is the bit the Bank has always marked its own homework on, and where I have had a proper go at building a model; and last, a piece published nine days ago by a fund manager called Christopher Mahon which is the best single thing I have read on this subject. There is also a late addition, with thanks to John Newton, who regular readers will remember from last October's piece on reserve interest and who pointed me at a blog the National Institute for Economic and Social Research put out on Friday afternoon. It has gone into the third theme below, and John's one-line review of it is better than anything I could have written. Buckle up. It gets technical, and I will translate as we go.

First up, the Bank's package itself: the QT section of Thursday's minutes, and the Market Notice that went with them.

Theme 1: The New Plan, in Numbers

The Summary: At its meeting ending 16th September the Monetary Policy Committee voted 9-0 to reduce the stock of gilts held for monetary policy purposes to zero through a multi-year plan, replacing the annual votes it has held each September since 2022. The Asset Purchase Facility peaked at £895 billion in February 2022 and stood at £488 billion on 16th September 2026. Over the last twelve months it fell by £70 billion, of which £21 billion was active sales. Under the new plan £120 billion of the longest-dated gilts is set aside permanently, leaving £368 billion to unwind. Of that, £222 billion matures by 2034 and will be held to maturity. The remaining £146 billion, maturing between 2035 and 2049, will be sold at £20 billion a year, giving an average annual reduction of £46 billion and completion in September 2034. No gilt maturing after 2049 will be sold at all. Bank auctions are paused in the meantime, and the Bank will set out by April 2027 how the sales will be carried out. Reuters noted that the Bank has been unique among major central banks in conducting outright sales of government debt.

The Propenomix Perspective: Let me do the arithmetic the press release did not. Of the £488 billion left, £222 billion will now be held to maturity and £120 billion will be kept for good. That is £342 billion, or 70% of what remains, that will never be sold into the market. Thirty per cent still will be. So is this a U-turn? Seventy per cent of one, I would say, and I will have to take it, while noting that it arrives four years and one series of very large bills too late. Last year's pace was £70 billion, the year before £100 billion. The new number is £46 billion, of which £26 billion a year is simply gilts reaching their redemption date, which is the "no profit, no loss" route I have been banging on about since this started. Bailey told the broadcasters this was not a reaction to the market and that the work began before the war. “Yeah right” was the first thing that sprang to mind but we might as well believe him because that makes it more damning rather than less. If the Bank had concluded by last winter that the old approach needed rewriting, then it spent this spring and summer selling into the weakest gilt market in a generation under a plan it had already privately decided to bin. More billions on the bonfire of vanities. I might be wrong about the timing. I know I am not wrong that a policy which needed rewriting in 2026 was the same policy in 2022 and should never have been enacted in the first place.

Theme 2: The £120 Billion That Is Never Coming Back

The Summary: The minutes record that reducing monetary policy gilt holdings to zero obliged the Bank to return to the question of what should back the banknotes it issues. In 2006 the Bank concluded that gilts were an appropriate backing asset given the enduring demand for notes. Accordingly the Bank, in consultation with the Treasury, will retain £120 billion of the longest-dated gilts in the facility to back current and future banknote issuance, via what the Market Notice calls the Issue Deposit. The holdings affected include £7.8 billion of the 1.75% 2049 gilt and every gilt maturing after it, among them £17.2 billion of the 3.75% 2052, £15.1 billion of the 4% 2060, £8.2 billion of the 2.5% 2065, £6.2 billion of the 3.5% 2068 and £5.7 billion of the 1.625% 2071. These will be held to maturity. Separately, the quarterly report notes that banknote-related cash flows from the Bank to the Treasury have totalled £20.5 billion since 2009. Bloomberg reported that the proposals have yet to be finalised.

The Propenomix Perspective: This is elegant, and I mean that as a compliment with a raised eyebrow attached. The ultra-longs are the gilts that have lost the most. Mahon, who we will come to, reckons some are down 70% or more. Sell them and you crystallise a loss that would make your eyes water, straight through the indemnity and onto the taxpayer. Relabel them as the backing for the fivers in your wallet, hold them to 2071, and the loss never crystallises at all. Is that an accounting sleight of hand? A bit. Is it also exactly what I have been asking for, which is to stop selling the things at the bottom? Yes. I’ve gone further and said to intervene in the long dated market, although my solution is more elegant than that and involves fiscal co-operation - but I’m convinced it would work. I am not going to complain about the label on the tin - one way or another, they are getting it done. What it does do, of course, is contradict and indeed quietly demolish the argument we were given for four years, which was that these holdings had to go because QE assets were never meant to be permanent. A hundred and twenty billion of them just became permanent on a Thursday lunchtime, nobody fainted, and the thirty-year yield fell. One detail I enjoyed more than I should have. The longest gilt in the facility matures in 2071. That is 49 years after QT began. Hold that thought for the last theme of the day, because I have been arguing for a fifty-year repair period for the Covid tranche and it turns out the bonds had one built in all along. Missed opportunity of the decade, if not the millennium so far.

Theme 3: Selling to Yourself

The Summary: For the £146 billion still to be sold, the Bank has engaged with the Treasury and the Debt Management Office on a model in which the sales are made to the Government rather than to the market, at market prices, on a pre-announced schedule and at the same £20 billion annual pace, with a decision before April 2027 so that it can be built into the DMO's financing remit. The minutes record that the institutional questions this raised led to what they call "a robust discussion" before all nine members judged that independence was preserved. On Friday 18th September David Aikman of the National Institute of Economic and Social Research published an analysis of the package through the consolidated public sector balance sheet. Under the DMO route the gilts would be passed to the National Loans Fund and cancelled, funded through ordinary issuance, so the quantity of debt the private sector holds does not change. What changes is its composition: a fifteen-year gilt is replaced by whatever the DMO issues, which will be shorter, so interest rate risk that would have passed to investors stays with the state, which Aikman gives as the reason the long end rallied. Assuming a term premium of 100 basis points, he estimates a saving of about £130 million in the first year building to roughly £1 billion a year by 2034, at the price of an interest bill more sensitive to Bank Rate and higher refinancing risk. He notes that slower sales could reduce headroom against the current budget rule while improving it against the supplementary rule, that the cost will in future appear as ordinary debt interest rather than as a separately reported indemnity transfer, and he describes the decision as "sensible given its remit".

The Propenomix Perspective: FINALLY. This is a massively watered down version of what I suggested a couple of years back, and is less aggressive and less confident than I would like - but it is something, and not to be sniffed at. Credit where it is due. John Newton sent me this over the last 48 hours with the comment that it "was interesting in what it didn't cover as well as what it did", and having read it twice I think that is about the most accurate review of a piece of economics I have seen this year. What it covers, it covers properly. The left hand of the state sells a gilt, the right hand buys it and cancels it, and the market never sees the bond. Aikman's point, which is the catch I would have flagged anyway, is that the DMO still has to raise the money, so the supply still arrives - just at the five or ten year point rather than the fifteen or thirty. That is a maturity swap, and it moves the pressure towards the part of the curve our mortgages price off. I would watch that carefully. A lot of me loves this maturity swap, and let me be very clear about how I would address it while the long dated gilts are as high as they are. This is where fiscal co-operation is needed.

Firstly - prepare policy to address the big 3 problems in the OBR’s long dated forecasts. Productivity, long-term health, and climate change and the economic impact. Announce the long-dated gilt purchases at a yield above x, with x to remain a) a secret (it would be obvious as soon as the purchases happen) and b) to be variable day-to-day. Buy aggressively in the markets longer than 10-years (leave the 10-year market alone). THEN announce the policies. 

A fix - I hear you cry. Yes. A bit. But the market would present challenges and this light touch yield curve control would remain under scrutiny. But what we would be doing is arbitraging away the long-dated premium by forcing it down - we would also then issue more longs if they were in demand via the DMO at a profit compared to our market-buying activities. And then we would stop. Give it a year or so - a bit like a “windfall premium” - and back ourselves to sort out the long-term problems. And have the temerity to make a profit, rather than lose tens of billions.

Back to the knitting. The presentational point above from the blog is the sharp one: once these bonds are cancelled, the quarterly indemnity payment that select committees have been asking awkward questions about simply melts into the general debt interest line. Same cost, no label. Too cynical? Perhaps, but I notice these things. And the headroom point matters for 28th October, because holding rather than selling moves the loss from one fiscal rule to the other. Isn’t that handy? 

Now, what it did not cover, and this is my list rather than John's. Not a word on whether the active sales of the last four years were worth doing (let me add those words. They weren’t).. Nothing on what they added to the cost of every gilt issued alongside them (tough to model, but it’s a lot, and it’s been borne by every borrower, or every tenant in a leveraged property). Nothing on paying full Bank Rate on every pound of reserves, which is John's specialist subject and was the whole of last October's piece (and also desperately needs changing, although with a solution less extreme than Farage’s that looks more like Gordon Brown’s). And one line I would gently query: that the Bank has done what it was asked to do, because its instruction is to minimise cost and risk. If that is the instruction, Mahon's account of the Hunt correspondence suggests the Bank read it rather narrowly. They most definitely have not treated cost with as much respect as they should have done, or have badly mispriced the risk. It is a very good analysis of Thursday. It is silent on the four years before Thursday, which is where the money went - and perhaps you can say “don’t cry over spilt milk” but when you have been shouting into the void about it for years, you can understand my frustration…and let me be absolutely clear about where I lay the blame. Government. It started with Truss who at the time (and still today) understood nothing about any of this. That was just a timing thing and nothing to do with her nonsense approach to governance or lack of it. Sunak failed here, Hunt failed, and let’s face it if they didn’t get it or give it airtime, Starmer and Reeves were never going to get it. Burnham and Healey? Well, the fix is in and is nothing to do with them, which is a good job, because I don’t see this lot being Burnham’s cup of tea. It’s found a few billion, which shouldn’t be sniffed at, but likewise - won’t go far. We need to go further and faster - to borrow a soundbyte that’s been retired….

Theme 4: A Fixed Pace, Two Escape Hatches, and What Has Actually Been Conceded

The Summary: The Committee judged that a fixed multi-year pace would make QT more gradual and predictable, allow the path to be embedded in market prices shortly after announcement, and limit the effect of the remainder of QT on yields beyond that initial move. The plan will be reopened in only two circumstances: if the Committee judges that movements in Bank Rate alone are insufficient to meet the inflation target, or if the Bank judges markets to be very distressed, in which case the Financial Policy Committee has a role. The minutes restate Bank staff's estimate that term premia on long gilts have risen by around 200 basis points since February 2022 and that QT accounts for 20 to 30 basis points of that. They also state that, because of the tightening effect of QT, the path for Bank Rate would have been slightly lower than otherwise. Long gilts rallied on the announcement: Reuters reported the thirty-year on course for its largest one-day fall in yield since April, to a three-week low, and other reports had yields down around ten basis points.

The Propenomix Perspective: So what has the Bank actually conceded? Not the principle. It still says it wants the monetary policy portfolio at zero, and it still says pace changes timing rather than cost. What it has conceded is everything practical: no more long sales, most of the book held to maturity, a chunk kept forever, and the rest possibly sold to its own government. And the market's verdict was instant. Call it ten basis points off the long end inside an afternoon, for the price of a press release. I will come to what ten basis points is worth in the third section, but the short version is around two billion pounds of lifetime interest for every single year of issuance at that lower rate. Saved. Immediately. By stopping. If announcing less of something takes ten basis points off, what did four years of doing more of it put on? The Bank says 20 to 30. I would note only that the first tenth came off rather easily. One fair point in the Bank's favour, and I will return to it: if QT tightens conditions, the Committee sets Bank Rate a touch lower than it otherwise would, so floating-rate borrowers got a little back. I am not convinced that offset is as big as the cost. But it exists, and I would rather you heard it from me.

Second, the receipt. The Asset Purchase Facility publishes a quarterly report and a set of annual accounts, and the Office for Budget Responsibility keeps a running estimate of the lifetime cost. Nobody reads any of them. We are going to.

Theme 1: £123.9 Billion In, £107.7 Billion Back Out

The Summary: The facility's quarterly report for the second quarter of 2026, published on 4th August, records that between 2009 and 2022 it generated net cash transfers to the Treasury peaking at a cumulative £123.9 billion at the end of September 2022. The first payment in the other direction was made in October 2022 and payments have been made every quarter since. By the end of June 2026 the cumulative net position had fallen to £16.2 billion in the Treasury's favour, meaning the Treasury has paid £107.7 billion to the facility in fifteen quarters. The accounts to 28th February 2026 show gilts at a fair value of £382.1 billion against a loan from the Bank of £554.5 billion, with £150.7 billion shown as due from the Treasury under the indemnity. The OBR's central estimate of the net lifetime loss was £104.2 billion in March 2024, within a scenario range of £46.6 billion to £156.9 billion; Mahon cites the OBR's November 2025 forecast as putting it at £164 billion.

The Propenomix Perspective: £107.7 billion in fifteen quarters is a little over £7 billion a quarter. Every quarter. Since Truss. To put that in the only currency Westminster understands, it is more than the whole of the employer National Insurance rise was ever supposed to raise, leaving the Treasury every three months, with no vote, no debate and no headline. I want to be careful here, because I have used the phrase "over a hundred billion we did not need to lose" and I owe you the honest breakdown rather than the angry one. Not every pound of that £107.7 billion was avoidable. A big slice is what the statisticians call interest losses: the Bank pays Bank Rate on the reserves it created, and collects coupons of one or two per cent on gilts it bought at the top. As I have said before - some of that could be changed by a solution between the one proposed by Gordon Brown (something like what the European Central Bank does - have different rates) and the Farage solution (we pay the commercial banks nothing on their deposits with the central bank). That gap opens up the moment rates rise, whether you sell a single bond or not. What was avoidable is the rest, and the rest has three parts: the losses crystallised by selling, the higher cost of every new gilt the Government has issued while the Bank was competing with it, and the decision to keep paying full Bank Rate on every pound of reserves. I will price each one before we are done. My hedge, up front: the number with three figures in the billions is reachable, but only on the less charitable assumptions. On the Bank's own assumptions it is smaller. It is still not small. Their errors have cost us tens of billions - and the Government’s failure to task them properly, and to hold them to account properly, are ultimately where the buck stops. 

Theme 2: "The Pace Does Not Matter"

The Summary: The quarterly report sets out four illustrative scenarios for future cash flows. Where Bank Rate follows the market path, the net present value of lifetime cash flows is approximately minus £120 billion whether the facility unwinds at £70 billion a year or at a constant £32 billion of sales plus maturities. Where Bank Rate falls back over three years to the Bank's estimate of its equilibrium level, the figure is approximately minus £60 billion under either pace. The report concludes that different unwind paces need not change lifetime cash flows on a present-value basis, that the path of interest rates matters more, and that holding bonds for longer increases the interest rate risk borne by the public sector, so that a faster pace has the benefit of reducing that risk. Reuters reported the Bank repeating on Thursday that changing the pace of sales mostly affects the timing of losses rather than their total size. The OBR's March 2026 forecast had assumed £32 billion of active sales a year.

The Propenomix Perspective: This is the Bank's best argument, and it deserves a straight answer rather than a sneer. The logic runs like this. A gilt's price today is just its future cashflows discounted at today's rates. Sell it and you take the loss now; hold it and you pay the same loss in instalments, as the gap between Bank Rate and the coupon. Same number, different timing. On a whiteboard that is correct, and I have done more than enough “classroom” economics to know it. It holds on two conditions. One, that today's yield curve is a fair guess at where rates go. Two, that your own selling does not move the price you sell at. Look at the Bank's own table. Market path, minus £120 billion. Rates drift back to neutral, minus £60 billion. The gap between those two outcomes is sixty billion pounds, and it depends entirely on whether rates fall. The Bank's own stated view is that they fall in the end. Alan Taylor put neutral at 3% in Thursday's minutes. So the Bank has been selling long gilts at yields its own rate-setters think are too high to last, which is the definition of selling at the bottom. And on the second condition, the Bank itself says its sales lifted yields by 20 to 30 basis points. I (alongside other, better researched bodies) have said consistently that that is an underestimate, although it is an increase from the Bank’s first stab at 10-15 basis points (which then drifted to become double that, of course). You cannot claim to be a price-taker and a price-mover in the same report. I suspect that is why the "robust discussion" was robust. But who will call this out? Me, and a very small group of people with a very niche audience. 

Theme 3: The Defence That QE Paid for Itself

The Summary: The report argues that cash transfers under the indemnity capture only part of the facility's fiscal effect. By lowering gilt yields, QE reduced the rate at which the Government issued debt, a saving locked in at issuance and, given the long maturity of UK debt, still largely to come. Combining projected cash flows with estimated issuance savings, the Bank puts the net fiscal impact between minus £70 billion and plus £65 billion, and states that under one set of plausible assumptions the savings "significantly, or fully" offset lifetime transfers. The same analysis nets off the cost of QT, assuming an uplift to issuance yields of 12.5 basis points in the first year of QT, 15 in the second and 20 in the third, and states that a further year at 20 basis points at the past year's issuance would add about £4 billion to the cost of debt issuance. The report adds that the wider macroeconomic support from QE, which it calls its most significant effect, is not included.

The Propenomix Perspective: Notice what has happened here. The defence of QT has quietly become a defence of QE. Nobody sensible, and I include myself on a good day, argues that buying gilts in March 2020 with the wheels coming off was wrong. I have made the case for the printer in a genuine emergency many times. I wanted us to buy more and reinvest the money in a Keynesian way in infrastructure, which was the original Johnson/Sunak plan (and it was a good one). The question on the table is not whether to buy them. It is how to get rid of them, and "we saved you money in 2020" is not an answer to "why did you sell at a loss in 2024". It is a bit like defending a terrible exit from a property deal by pointing out that the purchase price was good. Both can be true. Only one of them was this year's decision. But there is a gift buried in that footnote, and I am going to take it. The Bank has told us its own price for the damage: twenty basis points, for one year, at last year's issuance, costs about four billion pounds over the life of the debt. That is two hundred million pounds per basis point per year of issuance. Keep that figure in your pocket. It is the Bank's number rather than mine, and in the next section I am going to use it against them.

Theme 4: Buy High, Sell Low, and the One Trade That Made Money

The Summary: The report's timeline records that on 21st September 2022 (you may remember it - a few days before the lettuce wreaked havoc) the Committee voted to begin gilt sales, which started on 1st November 2022 at a pace of £80 billion for the first year, rising to £100 billion in each of the following two years and £70 billion in the year just ended. It also records that on 28th September 2022 the Bank began emergency purchases of gilts under its financial stability mandate following the mini-Budget, concluding them on 14th October, and that the resulting portfolio was sold between 29th November 2022 and 12th January 2023. The gilts in the monetary policy portfolio were bought between 2009 and 2021, the majority at yields between close to zero and around 2%. The Market Notice shows holdings such as £9.6 billion of the 0.625% 2035, £8.9 billion of the 1.25% 2041 and £5.2 billion of the 0.875% 2046 among those still scheduled for sale. Ten-year gilt yields closed the week near 5.3% and thirty-year yields near 5.8%.

The Propenomix Perspective: Strip out the jargon and it is this. We bought bonds at the highest prices in three hundred years of gilt history and sold them at the lowest in a generation. That is the reverse of what any decent investor, or any decent custodian of somebody else's money, is supposed to do, and no amount of present-value algebra makes it look clever on a bank statement. When I last totted this up, in October, I had the Bank at about £45 billion of losses crystallised on sales against about £23 billion of interest saved by no longer holding those bonds. Call it £22 billion net, gone, by choice. There is now a cleaner source for the first half of that. A footnote in the National Institute blog that John Newton sent me has the Bank's sales to date at £111.9 billion of nominal, bought at an average price of about 115 and sold for £84.9 billion of cash. So we paid roughly £129 billion and got £85 billion back. About £44 billion, or a third of the purchase price, gone. And here is the bit that still gets me. In the same autumn it started selling, the Bank was forced by Truss and Kwarteng to do the opposite for a fortnight: it bought long gilts when nobody else would, held them for a few weeks, and sold them into a calmer market. On that trade it made the Treasury money, around £3.8 billion as I recall it. Buy when there is panic, sell when there is not. It can be done. They have done it. It is the only profitable gilt trade the Bank has made this decade and it was the one they did not want to do. What could we learn from that? Quite a lot, I would suggest, and nobody ever asks. Third, the hidden cost. The cash going from the Treasury to the Bank is at least visible, if you know where to look. What is not visible anywhere is the extra interest on every new gilt issued since 2022 because the Bank was in the market selling at the same time as the Government. The Bank has always produced its own estimate of that. Others have produced very different ones. And as far as I can tell nobody has pulled it into one place with the workings shown, so I have had a go. Herewith!

Theme 1: The Creeping Estimate

The Summary: The Bank's published estimate of QT's effect on gilt yields has risen at every review. According to Mahon's account, a Bank survey of specialist investors in September 2022 produced an expectation that gilt sales would lift yields by 25 basis points, and the Bank's first assessment of its own, eight months later, was 0 to 10 basis points. The August 2023 Monetary Policy Report put the effect at 10 to 15 basis points on the ten-year yield, out of a term premium rise then estimated at around 40. The August 2024 Report said 10 to 20, the August 2025 Report 15 to 25, and the July 2026 Report, as restated in Thursday's minutes, 20 to 30. The Treasury Committee's February 2024 report on QT recorded the Bank's position that the effect was small, and that decisions on the scale, pace and composition of QT "were not and will not be affected by value-for-money considerations". Written evidence to that inquiry from Handelsbanken put the effect at the equivalent of 50 basis points on Bank Rate by mid-2025.

The Propenomix Perspective: Nought to ten. Ten to fifteen. Ten to twenty. Fifteen to twenty-five. Twenty to thirty. If a surveyor valued your house five times in four years and it went up every time by this sort of margin, you would start to wonder about the first valuation. Some of the rise is fair enough: the more you sell, the bigger the cumulative effect, and the Bank says so. But the midpoint today is 25 basis points, which is precisely what the people who trade gilts for a living told the Bank to expect in September 2022, before it had sold a single bond. It has taken four years and a few hundred billion pounds for the institution to agree with its own survey. I have called this "marking your own homework" since I first wrote about it, and I am pleased to see I am no longer the only one using the phrase. Does it matter who was right? Only because the pace was set every September on the basis that the effect was small. If the Committee had believed 25 in 2022, would it have voted for £80 billion and then £100 billion twice? I doubt it, and I suspect a couple of the people in that room now doubt it too. The regularity with which the Bank sidestep particularly difficult decisions and use a market forecast - if only they’d done that here, many billions would have been saved. Instead, they’ve had to revise until they ended up what the real experts said in the first place. 

Theme 2: What the Independent Numbers Say

The Summary: A 2024 National Bureau of Economic Research working paper by Wenxin Du, Kristin Forbes and Matthew Luzzetti, "Quantitative Tightening Around the Globe", compared QT across seven central banks. As summarised by the National Institute of Economic and Social Research, it found the largest cumulative effects in the United Kingdom, with government bond yields 44 to 70 basis points higher at horizons of one year and longer, and attributed this to the UK's heavier reliance on active sales, which the authors found to have a larger market impact than passive roll-off. Press reports of the paper put the effect of passive roll-off at the Federal Reserve and the European Central Bank at around 20 basis points. Forbes is a former external member of the Monetary Policy Committee. NIESR noted that event studies tend to understate the effect because QT decisions are well flagged in advance, and that most evidence comes from the deeper US Treasury market. Separately, Morgan Stanley has been reported as estimating an effect nearer 70 basis points at the thirty-year point. In October 2025 Reuters quoted Artemis's head of fixed income, Stephen Snowden, describing active QT as "madness".

The Propenomix Perspective: This is the paper I leant on last August, so I will not pretend to be neutral about it. But think about who wrote it. Kristin Forbes sat on the Monetary Policy Committee. This is not a think tank with an axe to grind or some rent-a-gob with a newsletter (ahem). It is a former insider, with two serious co-authors, comparing seven central banks and finding that ours stands out, and the reason it stands out is the thing ours did that nobody else did: sell. The Fed let its bonds roll off. The ECB let its bonds roll off. About 20 basis points each. We sold, and got 44 to 70. I want to hedge this properly. These are estimates, from a method with known weaknesses, in a period when every gilt yield on earth was rising for a dozen other reasons, and disentangling one cause from that lot is closer to art than science. I would not go to the stake for 70. But I would say this. When the institution doing the selling says 25, a former member of its own committee says 44 to 70, and an investment bank says 70 at the long end, the honest position is a range, and the honest midpoint of that range is not 25. For the model that follows I have used three cases: the Bank's own, a middle case at 40, and the midpoint of the academic range at 57.

Theme 3: The Propenomix Model - What It Cost the Exchequer

The Summary: The method takes the Bank's own published rule of thumb - that a 20 basis point uplift applied to one year's issuance at recent levels adds about £4 billion to lifetime debt costs, or roughly £200 million per basis point per year of issuance - and applies it to gross gilt issuance in each financial year since QT began, approximately £170 billion in 2022-23, £237 billion in 2023-24, £297 billion in 2024-25, £304 billion in 2025-26 and an assumed £140 billion for the first half of 2026-27, around £1.15 trillion in all. Using the Bank's own assumed uplifts, rising from 12.5 to 25 basis points, the lifetime cost of debt already issued is approximately £15 billion. At a constant 40 basis points from the start of active sales in November 2022 it is approximately £28 billion, and at 57 basis points approximately £40 billion. A cross-check using Mahon's approach, which applies the uplift to around £600 billion of medium and long issuance over a longer average life, gives £25 billion at 25 basis points, £40 billion at 40 and £57 billion at 57. Each further year of active sales at current issuance adds £5 billion at 25 basis points and around £11 billion at 57. The effect of the Bank's selling on the price of its own sales, on roughly £129 billion of gilts sold to date at purchase value, adds an estimated £3 billion to £7 billion.

The Propenomix Perspective: I have tried to build this the way I would want somebody else to build it if I were going to pick holes in it. The key input is the Bank's own, so nobody can say I cooked the conversion rate. Two different methods land in the same postcode, which is normally a good sign. And the answer is a range rather than a number, because anybody giving you one number on this is selling something. So: fifteen billion pounds if you believe the Bank. Somewhere between twenty-eight and forty if you believe the middle of the independent work. North of fifty if Forbes and her co-authors are right at the top end. My own lean, and it is a lean, is that the truth almost certainly has a floor at thirty rather than fifteen, because the Bank's estimate has never once been revised down. That is on top of the £107.7 billion receipt, not inside it, and it is locked in: a gilt sold at auction in 2024 with an extra 30 basis points on the coupon pays that extra every year until it matures. Now run Thursday through the same machine. Ten basis points off, at £200 million per basis point per year of issuance, is about two billion pounds of lifetime interest saved for every year the lower rate holds. For a press release. Health warnings: the issuance figures are my rounded ones and want checking against the DMO, the rule of thumb is linear when the world is not, and I have ignored index-linked entirely. Rough working. I would not put it in front of a statistician. I would put it in front of a Chancellor.

Theme 4: Households, Businesses, and the Offsets I Owe You

The Summary: Higher gilt term premia feed into swap rates and from there into fixed-rate borrowing. The regulators' second-quarter lending statistics put outstanding residential mortgage balances at £1,760.6 billion, the large majority on fixed rates, nearly all of which have been repriced at least once since November 2022. Assuming the effect at the five-year point is half that at ten years - an assumption, since QT's influence is concentrated at longer maturities - the running cost to mortgage borrowers is approximately £1.9 billion a year on the Bank's estimate, £3.0 billion at 40 basis points and £4.3 billion at 57, before any effect on business or commercial property lending. Against this there are offsets. The additional interest paid by the Exchequer and by borrowers is income to gilt holders and savers, some of it taxed, although pension funds are largely tax-exempt and around 30% of gilts are held overseas. The minutes state that QT led to a slightly lower path for Bank Rate; an illustrative 15 basis points on average reserves of around £650 billion would have saved the Exchequer about £1 billion a year, roughly £4 billion to date, and would have benefited floating-rate borrowers at savers' expense.

The Propenomix Perspective: The Exchequer number is the big one, but the household number is the one that lands on your desk and mine. On these assumptions every fixed-rate mortgage and every limited-company five-year fix written since late 2022 has carried a small surcharge, somewhere between an eighth and a quarter of a per cent, that nobody voted for and nobody itemised. On a £200,000 loan that is £250 to £500 a year. Not ruinous. Not nothing either, multiplied across £1.76 trillion. Now the offsets, because I said you would get both sides and because the argument is stronger for it. Yes, somebody receives the extra interest. If it is a UK pension fund, that is somebody's retirement, and I am not going to describe that as a loss. But pension funds largely do not pay tax on it, around three in every ten gilts are owned abroad so that slice simply leaves the country, and the "savers gain" line needs an asterisk: QT lifts long yields, not the rate on your easy-access account. If anything the Bank's own logic says deposit rates ended up a touch lower. Which is the real offset: a slightly lower Bank Rate, worth perhaps £4 billion to the Treasury so far on my rough numbers. Set against fifteen to forty, it does not cover it. And the thing no model captures is what lower borrowing costs would have done for an economy that has spent four years trying to find one per cent of growth down the back of the sofa. I cannot price that. I am fairly sure it has the same sign.

Last up, the best thing written on this subject this year. On 11th September, six days before the Bank moved, the New Economics Foundation published a guest piece by Christopher Mahon - a chartered financial analyst and fund manager with nearly thirty years in markets, writing in a personal capacity - called "How the Bank of England has wasted billions, and why Australia shows the way forward". I do not share a great deal of NEF's politics, which I mention because it is the point: the people cross about this now run from Farage to Gordon Brown to NEF to a property bloke in Solihull (the Sage, to his fans, and the Sausage to his detractors), and that is not a coalition you see very often.

Theme 1: Why Our QE Lost More Than Everybody Else's

The Summary: Mahon argues that the Bank's losses are large by international standards because of how QE was designed rather than whether it was done. On his calculation, counting mark-to-market valuation losses and interest paid on reserves from 2009 to 2024, the Bank lost more than 5% of GDP on its government bond programme, roughly twice the eurozone figure and four times that of the United States. He identifies four differences from the Federal Reserve. At its peak the Bank owned 37% of all gilts against the Fed's 19% of Treasuries. Its holdings had an average maturity of around 14 years against six to nine. It bought no index-linked gilts despite the UK having one of the largest index-linked markets in the world. And it bought passively across maturities where the Fed targeted the parts of the curve that looked dislocated. He notes that UK mortgages and business lending price off short and medium rates, unlike US mortgages which are anchored to thirty-year rates, and that the Reserve Bank of Australia confined its purchases to maturities of around five to ten years. Ultra-long gilts, he writes, have lost 70% or more of their value.

The Propenomix Perspective: This shifted my own thinking, which does not happen as often as it should - but I’m always very happy when it does happen.. I have always pointed the finger at the exit. Mahon points it at the entrance as well, and he has a point. If the purpose of QE is to get borrowing costs down for households and firms, and ours borrow for two and five years, why on earth were we buying fifty-year gilts? You get less stimulus per pound and far more risk per pound. The American mortgage is a thirty-year instrument, so the Fed buying long made sense for them. We copied the homework without checking if the question was the same. And the index-linked point is a belter, and is one I’ve made before. I said last August that if the Bank was going to hold anything it should hold linkers, to cut the Government's exposure to the inflation it cannot control. I’ve admonished and advised the Government against issuing so many linkers since early 2021 when I started warning of inflation. We hold none. About a quarter of our debt is index-linked, and we chose to buy none of it at the exact moment inflation was about to go to eleven per cent. I would hedge one thing. Hindsight makes everybody a genius, and in March 2020 the job was to stop the gilt market seizing up, not to optimise a portfolio. Fair. But by November 2020, the fifth round, with a vaccine announced four days later, there was time to think. The final £150 billion is the tranche I would most like to see the minutes behind.

Theme 2: The Incurious Committee

The Summary: Mahon describes the Committee as incurious about evidence that conflicted with its own estimate. After independent work, including the Forbes paper, pointed to effects up to three times the Bank's figure, the Bank did not re-poll investors to test whether those estimates reflected a market consensus; it stopped asking the survey question. He notes that the Bank defended its estimate by citing research on the United States, where the central bank was not selling bonds. He cites Deputy Governor Dave Ramsden's July 2023 statement that the Committee "does not take into account financial risk or profit" in decisions on the gilt portfolio, and the Governor's April 2024 letter to the Treasury Committee stating that such decisions are taken solely to meet the inflation target. He records that Jeremy Hunt, as Chancellor, reminded the Bank that the facility remained subject to the requirement to protect value for money, and that the Governor's reply confined that requirement to operational matters such as coordinating auction dates with the DMO.

The Propenomix Perspective: They stopped asking the question. I have read a good few central bank documents in my time, and I am not sure I have come across a better summary of how not to run one. It matches my own experience, for what that is worth. I have put this to Bank people more than once, at the regional briefings and in conversation, and the reply has always had a certain coldness to it: we do not need the gilts any more, so retiring the position makes sense. What, even if it costs the country FIFTY billion pounds? That is not our remit. And here is where I want to be fair to them, because they are not wrong about that. It is not their remit. The remit says hit 2% inflation, and nothing in it says do so at the lowest cost to the taxpayer. These are clever people doing exactly what they were asked to do, inside a framework that never asked the right question. Which is why I have always said - and have repeated today - that the blame sits in Downing Street rather than Threadneedle Street. Hunt had a go, far too politely, and was told where the door was. Reeves did not try - I suspect because she did not understand; this was not on her radar. Chancellors come and go, the indemnity keeps paying out, and the bill turns up in the public finances under a line nobody reads. Can we blame the Bank for not volunteering to be marked on something it was never set? Not really. We can - and we must - blame the people who set the exam.

Theme 3: The Australian Answer

The Summary: Mahon points to Australia, where a wide-ranging review followed the disorderly end of the Reserve Bank's yield curve control policy in 2021. The Reserve Bank and the federal government now publish a joint Statement on the Conduct of Monetary Policy under which the Monetary Policy Board must communicate a framework for additional monetary tools covering "the benefits, costs and risks" of each, drawing on international experience, independent expert assessment and the lessons of the Bank's own past use, and setting out how monetary policy will work alongside other arms of policy and how such tools would be exited under different scenarios. A skills matrix was introduced to broaden the board's markets expertise. Mahon proposes equivalent wording in the Chancellor's next remit letter, supplemented by requirements that the Bank's balance sheet be used efficiently, proportionately and with frequent review. He argues that scrapping the Treasury indemnity would help but is not sufficient, and calls on the Chancellor to commission an independent review led by a respected former Committee member, pausing active sales while it reports.

The Propenomix Perspective: Benefits, costs and risks. Independent assessments. Lessons. Exit scenarios. Nothing in that list touches who sets Bank Rate, and nothing in it would trouble a competent finance director for five minutes, because it is what any business does before it puts a nine-figure sum to work, never mind a twelve-figure one. This is more or less what I was fumbling towards last October when I wrote that the Bank should sell a bond only where doing so is net-present-value positive for the taxpayer, and publish the framework so everybody can see the working. Mahon's version is better drafted than mine, and it has the advantage of already existing in a serious country with an independent central bank that has not collapsed. I would make one observation on timing. He asked the Chancellor to pause sales and hold a review. Six days later the Bank paused sales and announced a review of its own, with the Treasury in the room. Draw your own conclusions about whether the Bank saw which way the wind was blowing. Healey has said nothing on it, as far as I can find. But the remit letter goes out with every Budget, and there is a Budget in five and a half weeks. If I could put one paragraph in front of him before 28th October, it would be that Australian one. This way gets it in the forecast, with much more certainty than launching yet another review, consultation, or other “non-action-taking” manana solution.

Theme 4: Thirty-Seven and a Half Years, Fifty Years, and the Case I Have Been Making All Along

The Summary: Mahon describes the Bank as having gone from running the world's most aggressive QE programme to running its most aggressive QT programme, at around twice the pace of the Federal Reserve and the European Central Bank, and argues that neither extreme was obviously necessary. The programme was built in stages: £375 billion between 2009 and 2012 in response to the financial crisis, extended to £445 billion after the 2016 referendum, and a further £450 billion in 2020 in response to the pandemic, reaching £895 billion. Between February 2022 and September 2026 the stock fell by £407 billion, an average of roughly £88 billion a year. Under the new plan it falls by £46 billion a year to 2034, with £120 billion retained. Internationally, the Federal Reserve and the ECB have reduced their holdings through maturities alone, and the Bank of Japan has for decades operated at a far larger balance sheet relative to GDP and for several years capped ten-year yields directly. During the 2022 gilt market emergency the Bank itself bought long gilts to restore order.

The Propenomix Perspective: Here is my case, as plainly as I can put it. The financial crisis was described at the time as a once in seventy-five year event. The pandemic was a once in a hundred. We responded to both with the biggest monetary intervention in British history, and then decided to reverse it inside a decade. Why? If something happens once every seventy-five years, you have seventy-five years before you need the ammunition again, and taking half of that to reload strikes me as prudent rather than reckless. Thirty-seven and a half years for the crisis tranche. Fifty for the Covid tranche. On those numbers the book runs down at about £21 billion a year. The Bank ran it down at £88 billion, more than four times as fast, and sold at the bottom to do it. The mad part is that the gilts themselves would have done nearly all of it for us. The last one matures in 2071. Forty-nine years. We could have done nothing at all and met my timetable with a year to spare. I would have gone further, and I still would, as I’ve laid out above. Where the long end offers nearly 6% for thirty years, a Bank with a brief to protect the taxpayer would be a buyer at the margin, not a seller, replacing a slice of what rolls off and leaning on the curve where it is obviously out of line. Yes, that is a cousin of yield curve control. No, it is not risk-free. Australia's attempt ended messily and markets do not enjoy a big player with a view, which I accept. You need to be opaque-transparent about the process to stop one of the hedgies taking you apart. We know it works in an emergency because we watched it work in October 2022, and we know stopping works because we watched ten basis points vanish on Thursday. So, do it only when fear indices reach a certain level, if you like. It is genuinely a win-win situation, and means we are greedy only when others are fearful, and fearful when they are greedy. That one could take off, you know (and yes, this is my hat-tip to the GOAT, who has progressed to “Chairman Emeritus” at Berkshire, at the grand old age of 96. Warren - I know that when you are gone, we will all miss you, but we will never forget your wisdom).

I have argued this for years. The Bank has not said I was right. It just had a different take based on a different remit - and, in my opinion, a total void of commercial nous. Instead of being the greatest arbitrage trader of them all, it has fed billions to the better informed private sector. Now, it has just started doing most of what I’ve been asking for all along. Back in the Oxford days, we would say “Quod Erat Demonstrandum” in situations like that (yes, we were a right laugh at the Eagle and Child, back in the day). 

The Week Ahead. A lighter diary after a week like that, which is just as well. The public sector finances for August are due, and that is the one I would mark: it is the last full borrowing print before the forecasts for the Budget are locked, and it will show what a summer of 5%-plus gilt yields has done to the debt interest line. Look for the payments to the Asset Purchase Facility while you are in there, because after this week you will know what they are. The flash purchasing managers' indices, my darling PMIs - the real time economy, will land midweek: watch whether services employment, which has been falling for two years, shows the stabilisation the Bank claims to be seeing, and watch the prices charged balance, because that is where the second-round effects would show up first. Consumer confidence from GfK usually closes the month's penultimate week, and with petrol at 161p last month and in the 170s now, I would not expect fireworks. The Bank's money and credit figures and the Nationwide index are the week after.

Off-calendar, the same three as last week with one addition. Oil, and whether the Oman meeting gets rearranged. The Budget countdown, now at five and a half weeks, with every Sunday paper a market-moving event until the red box. Bank speakers, because after a 6-3 with four of the six sounding twitchy, any of them could shift November on their own. And the new one: whether the Treasury says a single word about the Bank's proposal to sell its gilts to the Debt Management Office. If the Chancellor wants to own that idea he has about five weeks to say so, and I will be listening for it.

As we get towards the end for this week - I can't wait for our next workshop at the Hyatt Regency in Manchester on Thursday 1st October, with Rod Turner, powered by Roma Finance. JVs and M&A is the topic, and I said last week that the conversation in that room would be a more useful one than it would have been a month ago. It has got more useful again since. When the cost of debt is set by a strait in the Gulf and a committee that is 6-3 and wobbling, the ability to bring in a partner with cash, or to buy the portfolio of somebody whose refinance has just landed at 6.5%, is what separates the people who grow over the next three years from the people who sit tight and hope. Get the last dinner ticket, for goodness sakes. Eleven days to go. Book in at www.tinyurl.com/pbwoct26

Above all - please remember to Keep Calm, ALWAYS listen to or read the Supplement, and Carry On. I have spent a lot of this week's edition being cross about something that has already happened, and I make no apology for it, because the bill is real and it is ours. But take the good news where you find it. The Bank has stopped selling long gilts. It wasn’t because they read the Supplement, sadly - although if you know any of them, I really believe they should. Most of what is left will now be held to maturity, the rest may never see the open market, and the thirty-year yield came off its 28-year high inside an afternoon as a result. The Federal Reserve showed it still has a spine. Inflation is an oil story rather than a wages story, for now. None of that gets you a cheaper mortgage by Christmas, and I am not going to pretend it does. What it does is take one self-inflicted weight off a gilt market that had plenty of others to carry. The fundamentals have not changed: there remains a shortage of two and three bed terraces and semis, official rents are rising faster than prices and faster than pay, and supply is still leaving the sector the country has admitted it depends on. Investing in property is not a guaranteed win, and the next year will sort the people who priced their debt honestly from the people who priced it hopefully. Most of you reading this are in the first group. Enjoy the rest of your Sunday.

KCCO!