Welfare Cuts Can Move Costs From Government Budgets to Households and Services

A government can count a welfare saving quickly.

It appears in the forecast.

A budget line goes down.

A minister says the welfare system is becoming more sustainable.

But that does not prove the country has saved money overall.

It may only prove that the cost has moved from one place to another.

That is the point of this article.

A welfare cut can save one budget while creating costs for households, public services and other parts of government.

What this is not saying

This is not a claim that public spending should never be reviewed.

It is not a claim that every benefit award is perfect.

It is not a claim that fraud and error are imaginary.

It is not a claim that work support is pointless.

It is not a claim that assessment systems should have no rules.

It is not a claim that government budgets are unlimited.

Those would be weak claims.

The point is narrower.

When government removes support from one budget line, the public should be shown where the pressure goes next. If the cost moves into NHS appointments, tribunal appeals, rent arrears, unpaid care, school breakdown, homelessness support, lost work or family crisis, the saving has not disappeared into clean efficiency.

It has been moved onto other people or other services.

The government counts the budget saving

The official Spring Statement 2025 impact document put a clear number on the health and disability benefit reforms.

The reforms were expected to make welfare savings of £4.8bn by 2029/30.

£4.5bn of that was expected to come from working-age sickness and disability benefits.

That figure is simple.

It is countable.

It fits a headline.

It lets government say the welfare system is being made sustainable.

But the same document also shows people losing money.

Households also lose income

The official impact document estimated that 3.2 million families would lose financially as a result of the package.

The average loss was estimated at £1,720 per year.

That gives a rough gross household-loss scale of £5.504bn per year.

3.2 million families.

£1,720 each.

£5.504bn.

That is not a perfect Treasury comparison, because the same package also had gains, overlaps, protections, modelling limits and later political changes.

But it shows the scale of the transfer.

The saving is not just money found in the system.

It is money removed from households.

And the official document says the vast majority of families losing financially have someone with a disability in the household.

That is where public language can make the real effect sound softer.

The state says reform.

The household has less money.

Fraud, error and fiscal pressure are real

A serious welfare article has to say this clearly.

Fraud and error exist.

Bad decisions can go both ways.

Some people need better work support, not passive abandonment.

Some rules become outdated.

Some systems become too expensive, too slow, too adversarial or too poorly targeted.

Governments do have to make fiscal choices.

But those facts do not settle the argument.

They only define the test.

A reform that prevents fraud, reduces error, supports people into suitable work and protects people who cannot work is different from a reform that simply shifts hardship out of the welfare budget and into households, services and crisis systems.

The public should be shown which one is happening.

The assessment system also costs money

The state also spends money building the system that decides who qualifies.

That system is expensive.

The National Audit Office reported that DWP’s Health Transformation Programme was expected to run until 2029 and cost more than £1bn.

The Public Accounts Committee put the position more sharply. It said the programme had already spent £168m by March 2023 and depended on an interim Functional Assessment Service expected to cost around £2bn from 2024 to 2029.

That is the assessment system.

Forms.

Digital systems.

Contractors.

Reviews.

Medical evidence.

Decision-making.

Reconsiderations.

Appeals.

Management.

Governance.

The public is often told that support is too expensive.

Less often is it told how expensive it is to police, restrict and re-test support.

Contractor spending is still public spending

This is not only an argument about bureaucracy.

It is an argument about where public money goes.

When money is paid directly to a disabled person, it is visible as welfare spending.

When money is paid to run assessment systems, digital programmes, management structures and outsourced services, it can be described as reform.

But both are public spending.

The question is what the spending does.

Does it stabilise a person’s life?

Or does it build a more complex system for deciding whether that person should receive less support?

A country can spend heavily on restricting support while calling the reduced support a saving.

That is the risk.

Wrong decisions create extra costs

There is another cost: getting decisions wrong.

A 2024 Administrative Justice Council report looked at the cost of not getting Personal Independence Payment decisions right first time.

It found that in 2019/20, out of 780,000 PIP claims, 92,200 awards were changed after Mandatory Reconsideration and 59,400 were successful at appeal.

The report estimated DWP administrative costs for appeal-overturned cases alone at £23m to £29m.

That figure does not include all Mandatory Reconsideration costs.

It does not include the full cost to His Majesty’s Courts and Tribunals Service.

It does not include the claimant’s time.

It does not include health deterioration.

It does not include debt.

It does not include food-bank use.

It does not include GP appointments, emergency care, homelessness support or social care.

It is only one counted part of the cost.

That is the point.

The system counts the appeal administration before it counts the damage caused by the wrong decision.

The work promise is uncertain

The public argument for benefit restriction often rests on work.

Cut the wrong incentives.

Move people toward work.

Reduce long-term dependence.

Save money.

Some people do want work and could work with the right support, the right employer, the right hours, the right health care, the right adjustments and the right security.

That matters.

But the Office for Budget Responsibility was careful about the Spring Statement welfare measures.

It said it had not made a comprehensive assessment of the labour-supply impacts of the incorporated measures. The effects were complex and interacting. The government had not provided comprehensive and robust analysis. The OBR also said it did not have enough time to develop its own net-impact analysis.

It also warned that people in the limited capability for work and work-related activity group, and people receiving disability benefits, often have restricted capacity to work and may have been out of the labour market for some time.

That makes standard work-incentive assumptions less useful.

The benefit cut can be scored.

The promised work return is less certain.

Work support is not the same as benefit removal

There is a good version of the work argument.

It says people who can work should not be written off.

It says employers should make adjustments.

It says health care, transport, skills, flexible hours, occupational support and confidence matter.

It says paid work can improve income, connection and dignity when it is suitable and secure.

That argument is real.

But it is not the same as removing money and hoping pressure produces employment.

A person pushed into poverty is not automatically closer to suitable work.

A disabled person made more anxious, unstable, indebted or unwell may become further from work, not nearer to it.

So the serious test is not whether a reform contains work language.

The test is whether it funds the conditions that make work possible without punishing people whose capacity is genuinely limited.

Unsupported ADHD shows why delayed support can cost more later

ADHD gives a clear false-economy warning.

The political debate can easily become about overdiagnosis, rising demand, benefit claims, school adjustments and prescription costs.

Those questions are not all fake.

But the NHS England ADHD Taskforce makes the opposite cost visible.

It says unsupported ADHD is linked to educational failure, long-term unemployment, crime, substance misuse, suicide, mental and physical illness, and avoidable economic costs of at least £17bn to individuals and government.

That changes the question.

If unsupported ADHD already costs billions, then delaying support to control diagnosis numbers may not save money.

It may only move the cost.

To schools.

To families.

To employers.

To welfare.

To the NHS.

To the justice system.

To the person’s future.

A diagnosis can be misused.

But absence of support is not free.

The same pattern appears in other public services

This is the same pattern seen in dentistry, mental health and digital NHS access.

Dentistry shows the public-name, private-necessity model.

Mental health shows the public-distress, private-burden model.

Digital triage shows the easier-request, rationed-treatment model.

Welfare shows the budget-saving, cost-transfer model.

In each case, the public promise stays visible.

Care still exists.

Support still exists.

Assessment still exists.

Referral still exists.

The form still exists.

But the route tightens.

People wait.

People prove.

People appeal.

People self-manage.

People get worse.

People pay privately.

Families absorb the gap.

Then government counts the saving where the support used to be.

What a full cost test would ask

A proper cost test would ask where the pressure goes.

If someone loses PIP, does their health worsen?

Does a carer lose income?

Does a parent leave work?

Does the household fall into rent arrears?

Does debt rise?

Does food-bank use rise?

Does the GP see them more often?

Does A&E become the backstop?

Does the council pick up homelessness risk?

Does social care take over later, at greater cost?

Does the tribunal system pay to correct decisions that should have been right first time?

Does a school carry unsupported ADHD until the child breaks?

Does the justice system later meet the person the health system did not support?

Those costs are real.

They are just spread across too many budgets to make a neat headline.

Public language can hide the cost transfer

Public language makes the transfer sound responsible.

Sustainability.

Targeting.

Reform.

Incentives.

Work-readiness.

Fraud and error.

Higher needs.

Value for money.

Those phrases can describe real issues.

They can also hide a transfer.

Money is removed from people.

Responsibility is moved to households.

Risk is moved to families.

Pressure is moved to services.

Wrong decisions are moved to appeals.

Untreated need is moved into crisis.

Then the government points back to the first budget line and calls it a saving.

The TWIS point

The question is not whether public spending should ever be reviewed.

It should.

The question is not whether every diagnosis, award or service pathway is perfect.

They are not.

The question is whether the country is being shown the whole cost.

A benefit cut is easy to count.

A disabled person becoming poorer is harder to count.

A child waiting unsupported is harder to count.

A carer breaking is harder to count.

A tribunal correcting a bad decision is harder to count.

A delayed crisis admission is harder to count.

A life narrowed by unsupported ADHD is harder to count.

That is why welfare cuts should be tested against costs moved elsewhere.

What is fact and what is interpretation

Fact: The official Spring Statement 2025 impact document scored welfare savings of £4.8bn by 2029/30, including £4.5bn from working-age sickness and disability benefits.

Fact: The same official document estimated that 3.2 million families would lose financially, with an average loss of £1,720 per year.

Fact: NAO and the Public Accounts Committee reported major costs and risks in the health-assessment transformation and interim assessment-service system.

Fact: The OBR said it had not made a comprehensive assessment of the labour-supply impacts of the incorporated welfare measures because the evidence and interactions were complex.

Fact: The Administrative Justice Council estimated DWP administrative costs for PIP appeal-overturned cases alone at £23m to £29m.

Fact: NHS England’s ADHD Taskforce said unsupported ADHD carries avoidable economic costs of at least £17bn to individuals and government.

Limit: These figures do not prove every reform is wrong, every benefit award is correct, or every pound of support is well spent. They show that savings should be checked against transferred costs, implementation costs, decision error and downstream harm.

Interpretation: The state can count the removed support more clearly than it can count the harm moved into households, services, appeals and crisis systems.

TWIS frame: A saving in one department can become a bill somewhere else. The country should see the whole cost.

The question underneath

The public is told the country cannot afford support.

The clearer question is whether the country can afford the damage caused by removing it.

Until that question is answered honestly, welfare savings should be treated carefully.

Not because every pound of support is automatically well spent.

But because a saving in one department can become a bill in another.

And when the bill lands in a home, a school, a hospital, a tribunal, a food bank, or a person’s body, it may no longer be labelled as public spending.

It is still a cost.