Jim O'Neill went on Times Radio yesterday and said what most people who advise chancellors say in private; the Budget on 28 October will need either cuts or tax rises to rebuild the headroom the gilt sell-off has eaten, and the markets would respond well to a government that took credible action on what he called the excesses of the triple lock. Nobody at the Treasury will have been surprised. The Guardian reported the night before that the sell-off could wipe out half of John Healey's room for manoeuvre against his own fiscal rules. Debt interest is running at something over £100 billion a year. Every economist with a microphone has been pointing at the state pension for a fortnight.
And yet the triple lock is not going anywhere in October, and it should not. Andy Burnham said in July, in his Reddit session before he took office, that the manifesto commitment stands. He was right to say it, and he is bound by it. Labour promised in 2024 to keep the lock for the whole parliament. A prime minister who arrived by a route as unusual as his, with no general election behind him, cannot spend his first Budget tearing up a pledge his party made to nearly thirteen million pensioners. Whatever the markets want, that is the one thing they cannot have this autumn.
The interesting question is not what happens in October. It is what Labour puts in front of the country in 2029, and whether the numbers are strong enough to carry it.
Start with what the triple lock actually is, because most of the argument about it proceeds as if it were a moral principle rather than an accounting rule. Each April the state pension rises by whichever is highest of earnings growth, inflation, or 2.5 per cent. It was a coalition invention in 2010, a Liberal Democrat idea that George Osborne adopted because it looked cheap and was popular. It was sold as a correction. Margaret Thatcher had broken the link between pensions and earnings in 1980, and thirty years of price-only uprating had left the British state pension among the meanest in Europe relative to wages. Restoring that link was a fair aim. Labour had already legislated for it, in the Pensions Act 2007, before the coalition arrived.
But the lock does not restore the earnings link. It overshoots it, deliberately and every year. If wages grow faster than prices, pensions track wages. If prices grow faster than wages, pensions track prices. If neither grows much, pensions rise 2.5 per cent anyway. Over any run of years the pension therefore rises faster than either wages or prices, because it takes the best of the two in each period and never gives anything back. It is a ratchet. That is not an accident of the design; it is the design.
The Treasury's own estimate in 2010 was that the lock would lift the pension by about 0.2 percentage points a year faster than earnings, on average. On that basis the OBR calculates it would have cost about £5.2 billion a year by the end of this decade. What has actually happened is that the two non-earnings elements have been triggered in eight of the thirteen years the lock has operated, because inflation turned out to be more volatile and earnings growth weaker than anyone assumed. The consequence is that in the current financial year the state pension costs about £12 billion a year more than it would have done had it simply tracked earnings since 2011. By 2029-30 the OBR puts that figure at £15.5 billion a year, or half a per cent of GDP. Against price uprating the gap is wider still, at £22.9 billion. The lock has cost three times what it was designed to cost, and it has done so despite being suspended for a year during the pandemic.
For the individual pensioner the same arithmetic runs the other way. The full new state pension rose in April to £241 a week. The IFS calculates that is £30 a week, or 14 per cent, more than it would be had it tracked earnings since 2011. That is roughly £1,560 a year, every year, for anyone on the full rate, and the gap does not close. It compounds.
Those are the costs already banked. The larger argument is about the costs still to come, and here the honest answer is that nobody knows, which is the whole point.
The OBR's central projection is that state pension spending rises from about 5 per cent of GDP now, around £138 billion, to 7.7 per cent by the early 2070s. In today's money that is an increase of roughly £80 billion a year. Demographics account for 1.1 percentage points of the rise; the country is getting older and that is nobody's fault. The triple lock accounts for 1.6 points, which is more than half of the total and something like £45 billion a year in today's terms. That is the central case. If the economy is more volatile than the central case assumes, and the last fifteen years suggest it might be, the OBR says the lock could add a further 1.5 per cent of GDP on top, which the IFS translates as another £44 billion a year at 2025-26 prices. If things turn out calmer, the bill could be £40 billion lower. The range between those two outcomes is £84 billion a year. No other line in the public finances carries an uncertainty of that size for a reason that has nothing to do with need and everything to do with a formula.
It is worth pausing on what those numbers are not. They are not a forecast of how much the country will spend on pensioners; that will rise whatever happens, and should. They are the cost of one uprating rule over the alternative that Labour itself legislated for in 2007. Every pound of it is the difference between a pension that tracks wages and a pension that beats them.
The fairness case follows from the numbers rather than from anyone's feelings about anyone's parents. Two million or so pensioners live below the poverty line, many of them women with incomplete contribution records, and nothing in this argument is aimed at them. The problem is not that the old are rich. It is that the state has given one group of citizens a permanent guarantee that its income will rise faster than the economy paying for it, and has never offered the same to anyone of working age. Working-age benefits were frozen in cash terms from 2016 to 2020, losing something like a tenth of their real value. Child benefit lost far more than that over the decade. None of those people had a lock of any description. The generation now paying for the ratchet will retire with fewer defined benefit pensions, lower rates of home ownership, and a state pension age that the OBR would like to see hit 68 from 2037. The IFS estimates that someone in their early forties today, retiring in 2050, might need to save an additional £50,000 to £60,000 in today's money to hedge against the uncertainty the lock introduces into what their state pension will actually be worth, because the same formula that makes it expensive makes it unpredictable.
So why has nobody touched it? Because it was built not to be touched. Rishi Sunak suspended the earnings element for a single year in 2022, when furlough distortions threatened an 8.3 per cent rise, and even that required a manifesto-breaking vote and a promise to restore it the following April. Liz Truss's Downing Street floated dropping it in October 2022 and reversed within a day. Every party has now pledged it at every election since 2010. Mel Stride's line this summer, that Burnham must stand by pensioners and that the Conservatives remain committed to the lock, is the whole strategy in a sentence. Reform says the same. Any Labour government that changes the lock without a mandate will face both of them running the same attack in every seat with a bowling green.
That is precisely why the only sensible route is the manifesto, and why Burnham is the one Labour leader in a generation who could plausibly take it.
He has kept the promise. A prime minister who honoured the 2024 pledge to the last day of the parliament, while the markets and half the economics profession told him to break it, can go to the electorate and say that he keeps his word and this is the next word he is asking them to trust. He also has the right past. He was the health secretary who proposed a National Care Service in 2010, and social care has been the unfinished business of his career ever since. That gives him something to offer in exchange, and the exchange is the whole game.
Here the numbers matter again, because the replacement has to be costed honestly, and the honest costing is less dramatic in the short run than the headlines would suggest. Switching to an earnings link does not recover the £12 billion a year already built into the pension; that money is spent, and clawing it back would mean cutting the pension in real terms, which nobody serious proposes. The saving from switching is prospective. The OBR expects the non-earnings elements of the lock to trigger in three of the next five years, which means the ratchet keeps turning through this parliament. In the first years after a switch the saving would be low single-digit billions annually, growing with each year the ratchet would otherwise have turned. A useful rule of thumb is that each percentage point on the state pension costs about £1.4 billion a year at current spending, so a year in which the lock would have paid 2.5 per cent against earnings growth of 1.5 per cent saves roughly that much, permanently.
The long-run saving is a different order of magnitude. The OBR's own alternative modelling has an earnings link easing the pressure on the public finances by around 1.6 per cent of GDP over the projection period, which is the £45 billion a year in today's money mentioned earlier. More important than the central figure is the tail. A replacement framework removes the £44 billion downside scenario entirely, because a pension that tracks earnings with an inflation floor cannot ratchet. What the taxpayer buys is not mainly a saving; it is insurance against a bill nobody can currently size.
The design itself is already done. The IFS has proposed a smoothed earnings link of the kind Australia uses. Set a target for the state pension as a share of median full-time earnings, say a third, which is roughly where the new state pension now sits. Once it is there, uprate by earnings. In years when inflation outruns wages, uprate by inflation instead, and let the earnings link resume as real wages recover, so the pension never falls in real terms in a recession. No means testing. Write to everyone at fifty telling them their pension age, and fix it ten years out. The IFS models the pension staying within a band of 31 to 37 per cent of median earnings by 2050 under the triple lock, depending on luck; under a target, it would be wherever the country chose to put it. Labour can say, truthfully, that this is the earnings link the party legislated for in 2007, delivered at last, with a floor the Conservatives never offered. It can say that the 2.5 per cent element, the one that has no economic rationale and that nobody has ever defended on the merits, is going.
The politics are not as frightening as the folklore suggests. Labour is not the party of pensioners and has not been for some time; the over-seventies gave it around a fifth of their votes in 2024. The Conservatives and Reform will fight each other for that vote regardless, and they will both promise to keep the lock, and they will both be promising something whose cost the OBR has put at anywhere between £5 billion and £80 billion a year depending on the weather. Labour cannot win that auction. It can only decline to enter it, and put a different offer on the table that the other two cannot match because they have spent their money on the lock.
The risk is real and should not be waved away. Reform took Wigan in May on a swing that surprised people who thought they understood the north-west, and "Labour is coming for your pension" is a leaflet that writes itself. But the alternative is not safety. The alternative is another manifesto promising something the markets have already priced as unaffordable, a second term spent being told by the government's own forecaster that its flagship pension policy is the reason it cannot fund care or defence or anything else, and a reform imposed eventually by whichever chancellor runs out of road, without a mandate and in the worst circumstances. That is what the bond markets are asking for now. It is not what Burnham should give them.
The triple lock was designed to be too popular to break. It has also turned out to be too expensive to keep, by a margin its own designers never imagined. The only honest way to change it is to ask. Labour should put the question on the ballot paper in 2029, with the numbers beside it and a care settlement beside the numbers, and find out whether the country is as attached to a ratchet as its politicians assume.