Barely a week passes before the same conversation begins again. A statistic. A headline. A ministerial statement. The welfare bill is too large, the number drawing from it too high, something must be done. And then the argument moves directly to what to cut.
What almost never follows is the prior question. Where are these people supposed to go?
The bill does not shrink. It reappears elsewhere — in food banks, in NHS emergency departments, in housing crisis services, in the criminal justice system. The public sector picks up the cost at a different address.
The correct starting point is not the spending. It is the economy the spending reflects.
The Numbers, in Full
Before we can discuss what to do, we need to understand what we are dealing with. The scale of UK welfare expenditure is not well understood by the public — in part because it is routinely conflated with benefits for the unemployed, when the reality is considerably more complex.
In 2019-20, the year before the pandemic, the UK spent approximately £226 billion on social protection — around 10.3% of GDP. That figure was not especially unusual by developed-economy standards. It had been relatively stable through the latter half of the 2010s, partly because the government froze most working-age benefits between 2016 and 2020, holding real-terms spending down while the economy grew.
By 2023-24, the bill had risen to £278 billion. The OBR's forecast for 2024-25 is approximately £315 billion — 11.9% of GDP, the highest share since comparable records began.
To understand that number, it helps to break it down. The state pension alone accounts for approximately £134 billion — 43% of the total. It is the single largest item in the welfare budget by a considerable margin, and it is driven not by unemployment or dependency but by demography: an ageing population living longer, protected by a triple lock that has consistently increased pension payments faster than either earnings or prices. That spending cannot be meaningfully reduced without either changing the triple lock, raising the pension age, or asking existing pensioners to take a real-terms cut — none of which is straightforwardly achievable in a parliamentary term.
Disability and health-related benefits account for a further £68 billion in 2024-25 — nearly double the £37 billion of five years earlier. Housing support, including housing benefit and the UC housing element, runs at approximately £37 billion, up from £23 billion before the pandemic. Universal Credit payments to people in work — topping up low wages — account for a further £23 billion.
So when politicians talk about reforming welfare, they are in practice arguing about a subset of a budget that is mostly composed of items over which they have limited short-run control. The genuinely discretionary working-age element is considerably smaller than the headline figure suggests. And the largest growth areas — disability, housing, in-work support — are driven not by idleness but by structural economic conditions: a housing market that has priced millions out of affordable rents, a labour market that pays too many people too little, and a post-pandemic health crisis that has left a significant part of the working-age population unable, rather than unwilling, to work.
What the Pandemic Did
The scale of the Covid shock to the welfare system was without peacetime precedent.
Universal Credit claimants jumped from 2.9 million in January 2020 to 5.6 million by May 2020 — a near-doubling in twelve weeks. By October 2024, that number had reached 7.7 million. Some of this reflects the ongoing managed migration of people from legacy benefits onto Universal Credit, which will complete over the next year or two. But a large portion reflects a genuine and persistent expansion of the caseload.
Total welfare spending rose from £226 billion to £251 billion in 2020-21 — an increase of roughly £25 billion in a single year. The temporary £20 per week uplift to Universal Credit, the reset of Local Housing Allowance rates to reflect actual market rents, and the cost of furlough together represented the largest peacetime welfare expansion in modern British history.
What was not expected was how much of that expansion would prove permanent.
The claimant count for out-of-work benefits has come back down from its pandemic peak of 2.8 million to around 1.7 million — a level that is not dramatically elevated by historical standards. The headline unemployment rate sits below 5%.
The real legacy of the pandemic is not in those numbers. It is in the economically inactive.
The Novel Crisis: Inactivity, Not Unemployment
Economic inactivity means people who are neither employed nor looking for work. They do not appear in the unemployment figures. They are simply not in the labour market.
Before the pandemic, approximately 8.7 million working-age people in the UK were economically inactive. By 2024, that number had risen to approximately 9.4 million — around 800,000 more than the pre-pandemic baseline, and stubbornly resistant to recovery. The UK's inactivity rate of around 21.5% is now meaningfully above comparable economies: Germany sits at roughly 18%, the Netherlands at 18–19%, Denmark at 17–18%.
The single largest reason for that gap is long-term sickness. The number of working-age people citing long-term ill health as the reason for their inactivity rose from approximately 2.0 million in 2019 to approximately 2.8 million by 2023-24. Every one of those additional 800,000 people represents both a human cost and a fiscal one: most are drawing Personal Independence Payment or the limited capability for work element of Universal Credit, or both.
PIP claimants grew from 2.4 million in January 2020 to 3.6 million by early 2024 — an increase of 50% in four years. PIP spending rose from approximately £14 billion in 2019-20 to a projected £26 billion in 2024-25. The fastest-growing diagnostic category is not physical disability. It is mental health — anxiety, depression, and related conditions now account for approximately 30% of PIP claimants, with the number claiming for psychiatric disorders having roughly doubled since 2019.
This is not principally a story about fraud or malingering. It is a story about a society in which the experience of the pandemic — isolation, bereavement, job loss, the disruption of routine — left a significant and lasting mark on mental health. It is also a story about a healthcare system so under strain that accessing treatment takes months or years, leaving people in a limbo of ill health from which the route back to work is neither obvious nor well-supported.
The welfare bill reflects all of that. Cutting the benefit does not cure the condition. It moves the person somewhere else in the system.
The People We Are Actually Talking About
Before designing solutions, it is worth being precise about who is on welfare and why. The picture is considerably more varied than the political shorthand suggests.
The long-term unemployed — those out of work for twelve months or more — number approximately 417,000. That is above the pre-pandemic figure of around 280,000 but not dramatically so. These are people for whom the conventional job market has not provided a route back. Many have skills that are no longer in demand. Some have gaps in their employment histories that make them unattractive to employers operating in a risk-averse way. Getting them back into work requires not just job vacancies but active placement support, skills conversion, and — crucially — employers willing to take a chance.
Youth unemployment presents a different challenge. The unemployment rate for 16 to 24-year-olds stands at approximately 14.3% — well above the pre-pandemic figure of around 11%. Approximately 874,000 young people are currently classified as NEET: not in education, employment, or training. For many of them, the issue is not motivation but opportunity — entry-level jobs have become harder to access as automation has reduced the number of routine roles that traditionally served as the first rung of the employment ladder.
Older workers who have left the labour market represent a third and often overlooked group. Many left during the pandemic, some by choice and some by circumstance, and have not returned. They carry decades of experience and institutional knowledge. Ageism — explicit and implicit — is a significant barrier to re-entry. An employer looking at a 58-year-old candidate with a two-year employment gap sees risk. They should see expertise. Addressing this requires both a cultural shift in how employers think about experience and fiscal incentives that make re-hiring experienced older workers less uncertain for businesses.
And then there are the disabled — people for whom full-time conventional employment is not possible, but for whom complete inactivity is neither inevitable nor, as the research consistently suggests, desirable. The evidence on work and mental health is clear: meaningful activity, contribution, structure, and the social connection of being a respected team member are themselves therapeutic. Designing a system that creates genuinely accessible opportunities for people working within the limitations of their condition — whether through flexible hours, supported employment schemes, remote work, or part-time roles — serves both the individual and the public finances. It is not about forcing people into unsuitable situations. It is about not allowing a binary choice between full-time work and complete inactivity to be the only option available.
The Core Argument: The Economy Has to Go First
Here is the central point, and it is one that political debate consistently skips.
Welfare reform without economic growth is a zero-sum redistribution of difficulty. Germany's Hartz reforms — the most cited example of successful welfare-to-work restructuring in the developed world — reduced unemployment from 11.3% in 2005 to 3.2% by 2019, and drove long-term unemployment down by roughly two-thirds. Employment rose by 2.5 million in the three years following reform alone.
But the Hartz reforms did not work in isolation. They worked because the German economy was growing, because export demand was strong, because German industry could absorb the additional labour that tighter benefit conditionality pushed towards the market. The reforms created pressure. The economy provided the release valve.
The UK's productivity problem — a gap of approximately 20% against the United States and persistent underperformance against France and Germany — means the economy is generating fewer well-paid jobs relative to its size than its competitors. You cannot solve a welfare dependency problem when the economy systematically produces an insufficient number of jobs at wages that make work genuinely preferable to not working. You can shift the caseload around. You cannot reduce it.
This means that the welfare reform argument cannot be separated from the growth argument, the productivity argument, the infrastructure argument, and the energy argument. They are the same argument.
Where the Jobs Come From
If the work has to exist before the welfare can be reformed, the question is: what creates the work?
The answer is not complicated, though it is demanding. It requires government and business to act together across a sustained period on three fronts that are not normally discussed in welfare policy: physical infrastructure, industrial policy, and skills.
Physical infrastructure spending creates immediate labour demand across a range of trades that do not require degrees — and that pay wages which make work clearly preferable to benefits. Bricklaying, plumbing, electrical installation, groundwork, carpentry, steel fixing: these are skilled trades for which demand has consistently outstripped supply in the UK for decades, and for which the entry barriers are vocational training rather than academic qualification. A sustained infrastructure programme — in housing, in transport, in energy — does not just improve the productive capacity of the economy over the long run. It creates immediate, geographically distributed employment at skill levels accessible to a significant proportion of the welfare caseload. The apprenticeship model, combined with on-site training embedded in large public contracts, has produced results in other countries and can do so here.
The industrial policy argument centres on the sectors where the structural investment is already happening and where job creation is a foreseeable consequence. AI and data centre infrastructure is a rapidly expanding source of capital expenditure. Every facility built requires construction workers, electrical engineers, cooling systems engineers, and a permanent maintenance workforce. The optical fibre networks being laid across the country to support that infrastructure require skilled installers. The grid upgrades necessary to power it all require electricians, engineers, and project managers at scale.
Nuclear energy is perhaps the most compelling single job-creation programme available to the UK government. A credible nuclear new-build programme — not just Hinkley Point C but a rolling programme of smaller reactors at multiple sites — would generate sustained construction employment over a generation, in trades that are precisely those where the welfare-to-work pipeline makes most sense. Nuclear construction requires large numbers of qualified tradespeople. It creates communities of skilled work. It addresses simultaneously the energy competitiveness problem and the employment problem.
None of this happens spontaneously. It requires government to commit to long-term capital programmes and hold to them across parliamentary terms. The UK's historic failure to do this — to start infrastructure programmes and then cut or redesign them every few years — is itself a significant barrier to the investment in training that would otherwise follow. Employers do not train people for jobs that may not exist in three years.
Business as Partner, Not Spectator
Government can create the conditions. It cannot create the jobs itself, at least not at the scale required. Business has to be an active partner.
The relationship between welfare reform and the business community is usually framed in terms of conditionality — making benefits contingent on activity, requiring job searches, imposing sanctions. These are not unreasonable tools. But they address the supply side only. They do not create demand.
The other half of the equation requires employers to change behaviour. To take chances on people who have been out of work for a year. To invest in training rather than hiring only those who already have skills. To offer genuinely flexible and part-time roles that can serve as entry points for people returning from long-term sickness. To stop filtering out older candidates before interview.
Tax policy has a role here. Targeted reductions in employer National Insurance for businesses hiring and training the long-term unemployed, young NEETs, or older workers returning from inactivity — time-limited, with clear performance conditions — reduce the risk of hiring someone unfamiliar and make the economic case for doing so. They are not subsidies for doing something easy. They are compensation for taking a genuine chance.
The "make work pay" argument has been a policy aspiration for three decades, but the implementation has been inconsistent. The principle is right: the financial return to working must be materially and visibly better than the return to not working at comparable effort. The Universal Credit taper rate was reduced from 63 to 55 pence in the pound in 2021, at a cost of approximately £1.5 billion per year, and improved work incentives for around 2 million households. But the interaction between benefits, tax credits, childcare costs, and transport costs still creates scenarios — particularly for parents with young children in high-rent areas — where the financial case for taking a low-paid job is weaker than it should be. Fixing that is not a single measure. It requires looking at the whole system. Raising the income tax threshold — making work pay before the state takes its first penny — is part of that.
What Works Elsewhere
Three countries have achieved materially better outcomes than the UK on welfare-to-work transitions. Each took a different approach, and each has lessons.
Germany's Hartz reforms of 2002–2005 combined tighter conditionality on long-term unemployment benefits with active placement support and wage subsidy schemes for employers. Unemployment was halved over fifteen years. The critical point is that this worked because the economy was growing and could absorb the labour. The UK cannot import the German solution without the German economic conditions — which means the growth agenda must come first.
Denmark operates what is known as the "flexicurity" model — a combination of flexible hiring and dismissal rules for employers, generous unemployment benefits for individuals (up to 90% of previous wages for low earners, for up to two years), and mandatory, intensive active labour market participation. The generosity is not a concession to idleness. It is part of the design: security enables workers to take risks, accept retraining, and move between sectors without fear of catastrophe. Denmark spends approximately 2% of GDP on active labour market policies — the highest in the OECD, and roughly four times the UK proportion. Its employment rate is around 76%. Its activation rate — the share of unemployed people actually engaged in active programmes — is 40–50%, compared to roughly 10–15% in the UK. You get what you invest in.
The Netherlands has one of the highest employment rates in the world — 80.5% in 2023. It achieves this partly through a "work first" approach that requires jobseekers to accept suitable work quickly, but equally through a system that makes part-time work accessible and treated as a genuine stepping stone rather than a second-rate outcome.
Singapore's Workfare Income Supplement is worth noting as a model for incentivising older and lower-income workers specifically. Rather than punitive conditionality, it provides a financial supplement — partially in cash, partially deposited into retirement savings — for workers in low-wage employment. Employment among Singaporean workers aged 55–64 rose from approximately 55% in 2006 to 72% by 2023. The supplement treats work as something to be made attractive rather than dependency as something to be made unattractive.
The Long View
It would be dishonest to conclude by suggesting this is a short project.
The welfare bill has not grown to £315 billion in one parliamentary term. It reflects decades of demographic change, housing policy decisions taken and not taken, industrial decisions that hollowed out entire regions, a pandemic whose full mental health consequences will not be understood for years, and a healthcare system under strain that leaves people waiting too long to get well enough to work.
None of it will be reversed in five years. The structural forces — an ageing population, a disability caseload driven partly by conditions that take years to treat, a housing market that makes work financially marginal for people in high-rent areas — are not amenable to a single Budget measure or a departmental reorganisation.
What can be done is to begin the sequence correctly. Invest in growth-creating infrastructure. Create real jobs at accessible skill levels. Invest in the training infrastructure that turns welfare claimants into tradespeople. Design tax policy that makes hiring the long-term unemployed worth doing for businesses. Reform the benefit system so that work always pays visibly and materially more than not working. Address the mental health backlog — because until those 800,000 additional economically inactive people can access treatment, the welfare bill that reflects their condition will not fall.
The OBR projects welfare spending reaching 12.5% of GDP by the end of the decade without intervention. The state pension alone may reach 7% of GDP by 2070 under current demographic trajectories. These are not figures that yield to rhetoric. They yield to sustained, multi-Parliament commitment to an economy that generates enough well-paid work that fewer people need the state to compensate them for the absence of it.
Before the reform, the job.