The prime minister abandoned the 45p income tax rate cut in part because it was overshadowing the Tory Party conference. Unfortunately for her, its cloud was quickly replaced by an uncomfortable discussion about welfare benefits. The backbenchers Damian Green and Sir Iain Duncan Smith, alongside the cabinet ministers Penny Mordaunt and Sir Robert Buckland, used their platforms to pressure Liz Truss into raising welfare benefits to fully offset high inflation.

No government decision has yet been made on how to uprate universal credit and other benefits for next year, because the economics is tricky. Mordaunt is correct that, in theory, adjusting all benefits and tax thresholds for inflation annually “makes sense”. Policy should deliver constant levels of assistance to living standards. If today’s support is at the “correct” level, lifting the money value of benefits and thresholds by inflation each year ensures that its real purchasing power is protected.

In practice, this principle has been abandoned regularly for both taxes and benefits. Working-age benefit levels were often increased by less than inflation or even frozen between 2010 and 2019, as governments considered them too generous. The state pension triple-lock typically guarantees increasing real support for pensioners. Even that, though, was suspended in 2022 when the Covid rebound sent average earnings growth soaring.

On tax, inflation-adjustment has been corrupted further. The Institute for Fiscal Studies’ Tom Waters highlights how the frozen thresholds for income tax bands amount to a huge tax rise. The four-year freeze of the personal allowance and 40p starting threshold through 2025–26 will drag 1.4 million and 1.6 million additional people into paying the basic and higher rates, respectively. The £150,000 threshold for the 45p tax rate has been frozen since 2010 — trebling the number who pay it.

Add to these the long-term freezes in the threshold after which child benefit is tapered, and the stagnant household benefit cap, and Waters calculates households’ budgets overall will be squeezed stealthily in the coming years, with these freezes taking £2 in revenue for every £1 in personal tax cuts Truss promises.

So the Mordaunt principle is hardly sacrosanct, especially in emergencies. With working age benefits, today’s circumstances are more peculiar. Universal credit and others are supposed to be uprated each April by the consumer prices index (CPI) inflation rate recorded in the previous September. At 3.1 per cent last year, though, households struggled as inflation soared to 9 per cent by this April. The government stepped in with energy support packages that gave at least £1,200 to vulnerable households, before the new “energy price guarantee” that will freeze unit prices from October.

So now we reach Truss’s dilemma. Not lifting key working age benefits by the likely 10 per cent September inflation rate, with temporary support then expiring, would see the real value of relief to poor households fall next year. Yet simply uprating benefits with CPI again would deliver a degree of double support for energy bills in the near-term, while worsening work incentives.

The gas shortage driving inflation is a “real” shock. It reduces our economy’s capacity to produce goods and services, thus squeezing real wages. If working age benefits are inflation-protected, while real wages decline for low-paid workers, then the replacement value of benefits relative to wages rises. This discourages people from offering up (more of) their labour.

Truss is a supply-sider who thinks such incentives for work and production are important. She will want to avoid shrinking the willing workforce, particularly with 1.26 million vacancies.

Add in the potential £4 billion in fiscal savings on offer, and it’s no real surprise that some in government see linking benefits to average earnings growth (5.4 per cent), rather than the inflation rate, as a sensible compromise. Whether a government that’s announced certain headline tax cuts has the votes in parliament to do it is another matter.