At £94 billion this year, the British government’s projected spending on debt interest is as much as it spends on the Department for Education. Since 2019 a combination of £800 billion in new debt, higher index-linked debt payments due to elevated inflation, and a quadrupling of the average yield of new gilts has led to a surge in debt interest costs.

For those itching to rerun the macroeconomic debates of the 2010s, new debt being more expensive screams missed opportunity: “Why didn’t we simply borrow more when debt was cheaper, building schools, hospitals and transport infrastructure?” Even some austerity advocates are sympathetic to the argument. Nick Macpherson, former permanent secretary at the Treasury, recently said on BBC Radio 4’s Analysis: “With hindsight, we probably should have taken advantage and borrowed more.”

Given his fiscal conservative credentials, the intervention is seen as an admission that austerity’s critics were right: more investment then would have been prudent. But he misses a crucial point. Unless paid off, more debt from that era would face the harsh prospect of refinancing at today’s higher interest rates. “Cheap” borrowing from the 2010s is getting expensive.

This “rollover risk” usually gets ignored, but it’s very important. As much as 91 per cent of government gilts due for redemption before 2027 were issued when borrowing costs were lower than today. We must refinance a hefty 20 per cent of our outstanding £2.4 trillion gilt pile by then and 40 per cent by 2030, now at potentially much higher rates.

More borrowing in the 2010s would have worsened this prospect. Even if the government could have repeated today’s 15-year average debt maturity profile, gilts issued for investment would soon require refinancing. Yet the present 4.3 per cent yield for a ten-year gilt dwarfs the 2.6 per cent of a decade ago — the longer rates remain elevated, the higher payments will rise.

Champions of more investment in the 2010s would say this rollover risk could have been mitigated by borrowing longer at, say, 30 years. But traders doubt the market would have easily absorbed huge volumes of additional long-term debt.

People forget that longer-dated yields in the early 2010s hinted at more “normal” interest rates returning. Thirty-year yields ranged between 3 and 4 per cent until late 2014, plunging below 2 per cent only when the “zero for ever” interest rate hypothesis really took hold from 2017 onwards. It wasn’t as cheap to borrow between 2010 and 2015 as many seem to remember.

The UK has already borrowed longer than other big economies but even this “advantage” is misleading. The Bank of England’s quantitative easing programme involved purchasing longer-dated debt and issuing interest-paying reserves. That dramatically shortened our effective debt maturity — an effect that would have affected debt issued for more investment too.

The wisdom of more investment borrowing in the last decade hinges on whether you think it would have transformed our growth prospects. I am sceptical. New schools and hospitals are nice, but they don’t do much for growth. Even on transport, HS2 shows the government often plumps for low-return prestige projects that take so long to build they wouldn’t have affected GDP much yet. Plus, when debt was actually dirt cheap, unemployment was low, leaving government projects competing with the private sector for workers, as well as machines.

The truth, as one American budget analyst put it, is that “borrowing is never free and is rarely cheap”. State spending is projected to rise by 4.1 percentage points from the pandemic to 2027, with almost half the rise driven by debt interest costs. More borrowing in the 2010s would have compounded the problem.