A few years ago I proposed Chadwick’s laws of Government:
- if an ambitious right wing politician tells you they have a clever plan to save money it always ends up costing more, and
- if an ambitious right wing politician tells you they have a clever plan to reduce bureaucracy it always ends up with more paperwork
I was working for a DWP quango in 2010, and sitting on the commercial board of DWP. It was the age of austerity, and Osborne proposed huge cuts to Government to free up money for tax. I would be the first to admit that Government does things badly, both at a local and national level. And I would love to see a smaller state. But that doesn’t mean that getting cash out is easy.
Over a decade on Reform and the Conservative parties are starting to make the same claims again.
PIP/DLA
In June 2010, George Osborne’s first Budget announced plans to reform Disability Living Allowance. For working-age claimants, DLA would eventually be replaced by Personal Independence Payment, or PIP.
Part of the rationale was to reduce expenditure. The government expected the new system to reduce the cost of working-age disability benefits by 20%. By 2015–16, it expected to be saving around £1.4 billion a year.
It didn’t. By 2015–16, the estimated saving was around £100 million. The Office for Budget Responsibility subsequently looked at what had happened. By 2017–18, rather than costing 20% less than the system it replaced, PIP was costing around 15–20% more than DLA would have cost.
The principal reason was fairly straightforward. The government had substantially underestimated the number of people who would qualify.
That is worth remembering as politicians once again turn their attention to the cost of disability benefits.
Universal Credit
Universal Credit was another major welfare reform of the same period. Again, it was trying to address a real problem.
The benefits system was complicated. Different benefits were withdrawn at different rates as people entered employment or increased their earnings. In some circumstances the interaction between benefits and taxation meant people retained less than. 30% of the additional money they earned. This was the benefits trap.
Universal Credit was intended to simplify the system by combining six benefits into one and ensuring that work paid. I was working in government during the early implementation of Universal Credit and was interviewed by the National Audit Office as part of its investigation into the programme.
Its 2013 report wasn’t encouraging. The NAO found “weak management, ineffective control and poor governance” and concluded that DWP had not achieved value for money. Perhaps more remarkably, DWP couldn’t assess the value of IT systems on which it had already spent more than £300 million.
Universal Credit survived that beginning and eventually became the principal working-age benefit. The benefit trap did not disappear.
Universal Credit is currently withdrawn at 55p for every additional £1 of relevant earnings. Once income tax, National Insurance and other deductions are included, some recipients can still lose 6.9% majority of an additional pound earned. All of that spending for 1p in the Pound.
Nor did administering welfare become inexpensive. DWP’s departmental expenditure in 2024–25 was £9.9 billion. This isn’t an argument that Universal Credit caused that expenditure, or even that the old benefits system was preferable. It is simply a comparison between what governments expect large structural reforms to achieve with what subsequently happens.
DOGE
The same idea appeared in a rather different form in the United States.
Elon Musk’s Department of Government Efficiency was established after Donald Trump’s return to the White House. Initially Musk talked about finding as much as $2 trillion of savings from federal government expenditure. That number gradually became smaller.
DOGE eventually claimed savings of around $215 billion, although independent attempts to verify those savings found significant problems with some of the calculations.
But there is a broader way of judging what happened.
The United States federal government is still running a deficit of around 6% of GDP. That would be less surprising during a recession. But the US economy is growing at around 2–2½% and at the same time experiencing an extraordinary private-sector investment boom associated with artificial intelligence. Hundreds of billions of dollars are being invested in chips, data centres, power generation and other AI infrastructure.
As of August 2026, US gross federal debt has just passed $40 trillion, while interest payments alone are running at roughly 3% of GDP. Treasury yields have been rising and Reuters reports increasing investor concern about America’s fiscal trajectory. Whatever DOGE achieved, it has not changed the underlying fiscal position of the United States, which has deteriorated consistently since Trump returned to office.
There is also a more general problem with measuring government efficiency purely by expenditure. An employee at the Internal Revenue Service is an expenditure. Sacking that employee reduces expenditure. But if the employee would have collected more in tax than they cost to employ, the government deficit becomes larger rather than smaller.
Cutting tax enforcement saves government expenditure but also means collecting less tax. Before DOGE, the CBO estimated that rescinding $20bn of IRS funding would reduce tax receipts by $66bn — making the deficit about $46bn bigger, not smaller.
More recent research from Yale’s Budget Lab estimates that IRS workforce reductions associated with DOGE could reduce revenues by around $598bn over 2026–35. That’s an estimate rather than an observed outturn, but it’s an excellent illustration of the basic problem: a government employee isn’t necessarily a cost to the Exchequer. Sack the person collecting taxes and you’ve reduced payroll while making the government’s finances worse.
This distinction between reducing expenditure and improving the public finances is central to explaining why DOGE failed.
Reform tries DOGE
Reform adopted much of the same language in Britain. Its 2024 Contract claimed £91 billion of annual savings. £50 billion of this would come from requiring government departments to save £5 in every £100 through reduced bureaucracy, better procurement and greater efficiency.
Then, following the 2025 local elections, Reform gained control of several English councils. This provided an opportunity to test the idea. Reform launched its own DOGE operation, initially in Kent, bringing in businesspeople and technology specialists to examine council expenditure.
It subsequently claimed that Reform councils had identified £331 million of savings. When those claims were investigated, around £71 million could initially be accounted for from the examples Reform had provided. Some related to decisions already initiated before Reform took control. Others involved expenditure that had not actually been committed.
The DOGE operation subsequently became less prominent, with Richard Tice describing DOGE more as a philosophy of finding savings.
None of this means that Reform councils cannot reduce expenditure. They probably can. But it does illustrate how much easier it is to identify a large theoretical saving than to turn it into a cashable one.
Back to welfare
Reform is now proposing a much larger experiment. Its latest welfare proposals aim to reduce annual expenditure by £50 billion by 2030.
To give that number some scale, Britain currently spends around £145 billion a year on benefits for working-age people and children. The proposed £50 billion saving is therefore equivalent to more than a third of the entire non-pensioner welfare budget.
Reform says around £22 billion of its savings would come from changes to health and disability benefits. Working-age PIP would be replaced, as would the health element of Universal Credit, with a more restricted system of support. Around 2.9 million people would be affected.
If this sounds familiar it should do. This brings us almost exactly back to where we started. This is exactly where George Osborne messed up. The last major attempt to reduce disability expenditure by replacing one disability benefit with another was the introduction of PIP.
It was supposed to reduce expenditure by 20%. Instead, expenditure eventually turned out to be 15–20% higher than it would have been under DLA.
That doesn’t prove that Reform’s proposals will fail. It does suggest that we should be cautious about assuming that £22 billion of reduced benefit entitlement automatically becomes £22 billion of savings.
Clacton
There is another aspect to these proposals which receives less attention. The constituencies most affected are not evenly distributed around the country. They tend to be places with higher levels of ill health, disability and economic inactivity.
And many of those constituencies are places where Reform performs particularly strongly.
Clacton is the obvious example. More than 10,000 people in Nigel Farage’s constituency currently receive PIP. At the average award, something approaching £77 million a year is being paid into households in his constituency through PIP.
We don’t know how those recipients vote, and there is no reason to assume that they are Reform voters. But that isn’t really the economic point.
They live in Clacton. And the money they recieve is spent in Clacton. Even if a quarter of existing PIP expenditure disappeared, that would represent something approaching £20 million a year being removed from household incomes in the constituency.
At a third it would be around £26 million. At half it would be approaching £39 million.
Those aren’t forecasts. Until the details of the replacement system are clearer, we cannot know how much PIP income would actually disappear. They simply demonstrate the scale of the local economic exposure.
For poorer areas in particular, that matters.
Moving money between government departments
There is a further problem with welfare savings. If somebody loses £5,000 of benefit, DWP has reduced its expenditure by £5,000.
That doesn’t necessarily mean government has saved £5,000. PIP exists because disability creates additional costs. If someone’s income falls, these costs remain.
There is extensive evidence of the relationship between poverty, financial insecurity and poor health. If reducing people’s incomes makes some of them less healthy, part of the DWP saving will eventually appear as additional expenditure elsewhere — particularly in the NHS, but potentially also in social care, housing and local government.
There are economic effects as well. If millions of pounds of benefit income disappear from an area such as Clacton, household spending falls. That means less revenue for local businesses, which in turn means lower profits, lower tax receipts and potentially fewer jobs.
None of these effects means that welfare expenditure can never be reduced. It means that the amount removed from the DWP budget and the amount ultimately saved by government are two different numbers.
The recurring problem
There is undoubtedly waste in government. Anyone who has worked in a large public organisation will have seen examples of it. Governments should try to eliminate it.
But the experience of the past fifteen years suggests that finding large, painless and permanent savings is considerably harder than politicians sometimes imply. We have over a decade of failed attempts to cut spending, at the end of which spending is higher than when they started.
PIP was supposed to reduce disability expenditure by 20%. It eventually cost more than the system it replaced. Universal Credit was intended to simplify welfare, improve work incentives and reduce administrative complexity. Its implementation began with the NAO reporting weak management, ineffective control, poor governance and a failure to achieve value for money.
DOGE promised enormous savings from the US federal government. The United States is still running a deficit of around 6% of GDP. Reform’s own DOGE exercise found that claimed savings were much easier to announce than to demonstrate.
Now Reform believes it can save £50 billion a year from welfare, including £22 billion from health and disability benefits.
Perhaps it can. But the relevant number isn’t how much expenditure can be stopped.
It is how much government has actually saved once the consequences of stopping that expenditure have worked their way through the benefits system, the NHS, local government and the wider economy.
George Osborne’s experience with PIP suggests that the difference between those two numbers is very large.