How have the size and shape of the UK state changed?

Executive summary

1. Introduction

For a remarkably long time, the UK state remained at around the same size. Government spending outside of economic crises hovered around 40% of national income between the mid 1950s and the COVID-19 pandemic. This has been in spite of large, epochal changes to what the state does. Persistent rises in spending on social security benefits, state pensions and health and social care through this period were largely offset by a large reduction in spending on defence and a more recent fall in debt interest spending.

Since 2019, the size of the state has grown considerably, not only reaching historic heights during the pandemic but lingering at a higher level thereafter. It seems unlikely – though not impossible – that we will return to the size of the state that we had been used to seeing pre-pandemic, at least without cutting the scope of what the state provides.

In this piece, we will seek to answer three key questions. First, how did the size of the state remain so constant for so long between the Second World War and the COVID-19 pandemic? Second, what has happened to the size of the state over the 2019–24 parliament, and why? Third, what can these developments tell us about the future, and the choices faced by both main parties?

2. Pre-pandemic trends: 1955–56 to 2019–20

Figure 1 shows the size of the state, measured by government spending as a share of national income. While spending as a share of national income has fluctuated considerably between the mid 1950s and 2022–23, this fluctuation has been around the long-run average of 40%. Notably, the state was almost exactly the same size in 2007–08 as in 2019–20, on the eve of the pandemic.

To some extent, the fluctuations in the size of the state over time reflect fluctuations of the economic cycle. During recessions – in the mid 1970s, early 1980s, early 1990s, the late 2000s or during the COVID-19 crisis – national income falls, and so a government spending a given amount of money will account for a larger fraction of the overall economy. Some components of spending also tend to rise during recessions – for example, spending on unemployment benefits. When the economy booms, as in the late 1980s, spending as a share of national income will similarly mechanically fall. To some extent, too, fluctuations in the size of the state are a matter of different governments’ political priorities. For example, the steady rise in the size of the state between 1999–2000 and 2007–08 can largely be explained by different policy priorities and ambitions under the Labour governments at the time.

But the key point here is that spending as a share of national income between the mid 1950s and the COVID-19 pandemic has hovered around 40%, only reaching levels significantly higher than this during economic crises. This is a remarkable fact in light of huge changes to government, the economy and society. Over the same time, between 1952 and 2022, infant mortality fell from 29.1 per 1,000 live births to 3.4, the proportion of energy generated from coal fell from 88.5% to 6.5%, and the number of women at university rose from 23,000 to 1,569,000 (Turner, 2022).

Changes in the composition of spending

While spending in the post-war period has stayed relatively constant, there have been much more significant changes in the composition of spending. It is therefore useful to look at the factors that allowed the size of the UK state to remain so remarkably stable over many decades.

We can decompose total spending by function. The largest of these areas include health (18.5% of total spending in 2019–20, immediately before the pandemic), pensioner social security benefits and state pensions (14.4% of total spending), social security benefits for children and working-age people (11.4% of total spending), education (10.2% of total spending) and defence spending (4.7% of total spending).

Panel A of Figure 2 splits spending into various functions at five points in the last 40 years: 1979–80, 1989–90, 1999–2000, 2007–08 and 2019–20. We use 2007–08 rather than 2009–10 as the aim is to illustrate the broader change in the composition of spending, and 2009–10 was a historically unusual year, mid financial crisis (as is clear from Figure 1). The size of the state is similar in these five years, but the shape is not.

Health spending has risen over this period, representing 3.9% of national income in 1979–80 and 7.3% in 2019–20. The rise has been by about 3.5% of national income, or £96 billion in today’s terms.¹  Social security spending – both for pensioners and for younger people – has also increased, from 8.3% of national income in 1979–80 to 10.2% in 2019–20, or by around £52 billion in this year’s terms. Helping to offset this has been a fall in debt interest spending, from 3.3% of national income in 1979–80 to 1.7% in 2019–20, around £43 billion, and a fall in defence spending from 4.1% of national income to 1.9%, around £61 billion. Education spending has also fallen slightly, as well as ‘other’ spending (mostly accounted for by a fall in housing and amenities spending of 2.1% of national income).

Falls in these areas of spending have allowed the size of the state to remain relatively constant while spending on health and social security rose. Panel B of Figure 2 shows that while spending as a share of national income was roughly similar in 1979–80 and 2019–20, this masks larger increases in health and social security spending as a share of national income (of 3.5% and 1.9%, respectively), offset by falls in debt interest, defence and education spending.

Panel C of Figure 2 shows the change in the size and shape of the state between 2007–08 and 2019–20. Again, spending on health rose by about 1% of national income, while spending on defence and education fell. This fall in education spending during the 2010s is partially a result of cuts to capital spending after 2010, as well as changes in higher education funding (Drayton et al., 2023). Spending on social security benefits also fell (by 1.1% of national income), as a result of cuts to the generosity of working-age welfare. The overall size of the state was essentially the same.

Figure 3 focuses in particular on the paths of health, education, and defence spending since the mid 1950s, showing these trends in the longer run. Defence spending has represented a shrinking share of national income, from 7.6% in 1955–56 to 1.9% immediately before the pandemic. This fall has been termed the ‘peace dividend’, a term initially coined to reflect the situation after the Cold War, but which can be applied more broadly to the fall in defence spending after the Second World War. The peace dividend has allowed the simultaneous rise in health spending from 2.8% of national income to 7.3% to take place without increasing the overall size of the state. Education spending has remained constant, at an average of 4.4% of GDP. These patterns together have been an important reason why total spending has remained roughly constant.

As Section 3 will discuss, it seems likely that these trends may have run their course. The falls in both defence and debt interest spending have been arrested, and it looks increasingly unlikely that health or social security spending will fall to offset rises here and keep the size of the state constant.

3. The recent past: 2019–20 to 2024–25

How has the size of the state changed over this parliament?

Between 2019–20 and 2024–25, spending as a share of GDP was more volatile than at any point in the post-war period.

The size of the state spiked during the pandemic, reaching a height in 2020–21 with spending equal to 53.1% of GDP. This partly reflects the contraction of economic output: as a share of the previous year’s national income, spending in 2020–21 was 49.3%. But it also includes the very large policy response to the crisis, including £70 billion spent on the furlough scheme (HM Government, 2021). Since then, the size of the state has fallen from this peak, but remains at a much higher level than before the pandemic.

Figure 4 shows the average annual real growth rate of spending by parliament since 1951 – i.e. not growth in spending as a fraction of national income but growth in real pounds terms. Spending grew at an average real-terms rate of 2.5% a year since 1951–52, tending to grow more rapidly under Labour-led governments and more slowly under Conservative-led governments.

During the current parliament, spending has increased by 2.9% per year in real terms – above the long-run average. Much, although not all, of this has been driven by high debt interest spending, as discussed below. Spending excluding debt interest payments has increased by 2.2% a year in real terms, still higher than under any Conservative government since the 1970s, but below the long-run average for this measure, 2.6%.

Figure 5 shows the average annual growth rate of spending as a share of national income, reflecting both changes in spending and changes in national income, and indicating the average change in the size of the state under each parliament.

We can see that big changes to the size of the state have been the exception, not the rule. Most parliaments have seen very small changes to the size of the state. Under Conservative governments, the only exceptions until now were large contractions to the size of the state: under Churchill between 1951 and 1955, reflecting both relatively low spending growth and strong GDP growth, and under Thatcher between 1983–84 and 1987–88, reflecting low spending growth and strong GDP growth. The contraction under the 2009–10 to 2014–15 coalition government was primarily driven by slow spending growth rather than particularly rapid GDP growth: this parliament saw the slowest period of real-terms spending increases on average of any parliament in the post-war era, as Figure 4 shows.

The average annual increase in spending as a share of national income under this parliament has been significantly higher than under any other post-war Conservative parliament. Comparable increases to this parliament have only happened on a few occasions in the past: between 1964 and 1967, with very rapid spending growth under Wilson, and from 2001 to 2009, under New Labour governments.

Much of the increase in the size of the state during this parliament was not anticipated. Figure 6 illustrates the change in government spending as a proportion of national income, as forecast in March 2020 and in March 2024. In March 2020, the size of the state was forecast to rise slightly – by 0.9% of national income between 2019–20 and 2024–25 – and to stay close to its long-run pre-COVID average.

The most recent forecasts (March 2024) show the size of the state remaining relatively high even after the COVID shock has passed, rather than falling back to pre-COVID forecasts. Spending is now expected to rise from 39.6% of national income in 2019–20 to 44.0% in 2024–25, a rise of 4.5% of national income.

What has driven this change in the size of the state?

In March 2020, cash-terms spending was intended to increase by 4.0% a year on average, and cash-terms national income by 3.6%. Given expected inflation at the time, this meant intended real annual increases of 1.9% in total spending and 1.4% in national income. Day-to-day and investment spending were planned to increase by 3.1% and 7.4%, respectively. Social security spending was expected to increase by 1.4%, debt interest spending to fall by 2.8%, and other annually managed expenditure to stay roughly constant in real terms. With then-forecast national income, this would have meant a rise in spending from 39.8% to 40.7% of national income.

As shown in Figure 7, both cash-terms spending and cash-terms national income in fact grew faster than was expected in March 2020. Cash-terms spending in fact grew by 6.7% on average each year, and cash-terms national income by 4.4%. But both growth rates overshot their forecasts by very different extents: spending has grown 1.7 times as fast as was forecast in March 2020, while national income has grown only 1.2 times as fast. While real-terms spending growth has been higher than was forecast in March 2020, real-terms GDP growth has been lower. According to March 2024 forecasts, this has meant a rise in spending from 39.6% to 44.0% of national income, an increase approximately 5 times higher than what was expected in March 2020.

Both the rise in forecast real spending growth and the fall in forecast real GDP growth drive the higher-than-expected size of the state. One way to illustrate this is by considering two hypothetical scenarios. Imagine real GDP had grown at the faster rate forecast in March 2020. Then, the same increase in cash spending would have amounted to an increase in the size of the state of only 2.8% of national income, compared with the 4.5% of national income rise that actually took place.

On the other hand, had spending grown at the slower rate forecast in 2020, lower growth in the real economy alone would have meant that the size of the state would have risen by 2.4% of national income. Clearly both factors – lower GDP growth and higher spending growth – were important in driving up the size of the state.

The increase in spending as a share of GDP was not evenly spread between different areas of spending. In Figure 8, we contrast the change in spending as a share of GDP over this parliament that was already anticipated before the pandemic with the change that eventually occurred, splitting by component of spending. The originally planned increase of 0.9% of GDP, and the increase of 4.5% of GDP that occurred, are shown as the right-most bars.

Most of the growth in the size of the state expected in March 2020 was expected to come from central government day-to-day and investment spending on public services, forecasts for 2024–25 which did not change much between March 2020 and March 2024. Day-to-day spending was expected to increase by 1.2% of national income in both March 2020 and March 2024. Investment spending was expected to increase by 0.9% of national income in March 2024, and has in fact increased by 1.0%.

The change has instead come from the other three components: social security spending, debt interest, and other annually managed expenditure (AME). All three have turned out higher than expected. Spending on social security benefits (including state pensions) was expected to remain roughly constant as a share of GDP over this parliament in March 2020, but in fact grew by 1.2% of national income. Spending on debt interest payments was expected to fall by 0.3% of national income over the parliament, but has instead risen by 1.5% of national income. And spending on other AME was expected to fall by 0.8% of national income, but has fallen by only 0.4%.

Debt interest spending

The forecast change in debt interest spending as a proportion of GDP between 2019–20 and 2024–25 is the area where there has been most change between March 2020 and March 2024. This reflects both lots more additional borrowing – to a large part because of the COVID-19 pandemic – and a much higher-than-expected cost of servicing debt, in the form of higher interest rates.

Figure 9 shows how forecasts for spending on debt interest as a percentage of GDP have changed between March 2020 and March 2024. Over the first two decades of this century, the share of GDP represented by debt interest payments was 2.0% on average. This share jumped to 4.4% in 2022–23 and is expected to be 3.2% in 2024–25, more than 1% of national income above its average since 2000 (although this level is less unusual over a longer span of time).

Spending on debt interest broadly depends on three factors: the stock of debt, interest rates, and the rate of inflation. Increases in all three of these factors over this parliament have driven up expected debt interest payments.

Figure 10 shows the change in forecasts for debt interest payments in 2022–23, 2023–24 and 2024–25, between March 2020 and March 2024. The March 2024 forecasts for debt interest payments were much higher in each of the three years. Fully disentangling the complex web of economic effects that have driven these changes would be beyond the scope of this piece. However, we will illustrate the role of two factors in a stylised fashion: higher inflation increasing the inflation uplift on index-linked debt, and changes in interest rates and financing. This second factor includes the impact of changes in the types of debt issued and held by different investors, changing the effective interest rate that applies, which has been an important feature of borrowing during this parliament.

Higher inflation than was expected in March 2020 has been important in driving up debt interest spending. Inflation affects debt interest spending via the portion (roughly one-quarter) of debt that is index-linked, which means that the cost of servicing this debt is linked to inflation as measured by the Retail Prices Index (RPI). Higher inflation – if temporary – largely has a transitory effect, so drives up debt interest spending in the period in which inflation is higher than expected but not thereafter.²  Figure 10 shows that higher-than-anticipated RPI inflation accounted for almost three-quarters of the increase in spending in 2022–23 – driving the spike in that year – but a much smaller proportion in 2023–24, and in fact reduced expected spending on debt interest in 2024–25, when RPI inflation is expected to be very low.

Higher interest rates – including both the effect of higher gilt and bank rates, and the effect of changes in the composition of debt held by different investors – and an increase in the expected stock of debt are expected to have a more important permanent impact in driving up debt interest spending. This is shown in Figure 10.

We can think specifically about the effect of the additional money borrowed to pay for the policy response to COVID, which represented around £300 billion between 2019–20 and 2021–22 (Office for Budget Responsibility, 2022). By increasing the stock of debt directly, this clearly added to debt interest payments. A back-of-the-envelope calculation, taking the additional amount borrowed and multiplying by the expected gilt rate in 2024–25 in March 2020, implies that this additional borrowing could account for around £3.4 billion of extra debt interest payments in 2024–25, if interest rates and inflation had remained the same. This is a small part of the overall change in forecast debt interest payments. The relative lack of importance of COVID borrowing in driving up debt interest payments can also be seen from Figure 9. The March 2022 forecast – after COVID borrowing had been undertaken – does spike but returns to pre-COVID expectations rapidly: by 2024–25, debt interest spending was expected to return to 1.7% of national income, broadly its pre-pandemic level, even though debt was expected to remain 20% of national income higher than in 2019–20.

More important has been the impact of changes to interest rates and the composition of debt during the pandemic. The increase in expected Bank Rate since March 2020, and the growing sensitivity to Bank Rate, with the Bank of England continuing to buy gilts by expanding its reserves during the pandemic, have both been important here in driving up debt interest payments, with the UK increasingly sensitive to higher-than-forecast rates.

In some ways, the more remarkable feature of Figure 9 is not the increase since 2020, but the stability of debt interest payments as a share of GDP until the pandemic. The composition of debt, as well as the size of the debt stock, changed from the financial crisis onwards, in a way that made debt interest spending particularly sensitive to changes in interest rates. The Bank of England began its quantitative easing programme in the wake of the financial crisis, which effectively entails substituting gilts for Bank of England reserves. The contemporaneous Bank Rate is paid on these reserves, and so this substitution makes debt interest payments increasingly sensitive to short-term interest rate changes. In 2020, IFS work warned of the much increased sensitivity of debt interest payments to short-term interest rate changes, as a result of this substitution between gilts and Bank of England reserves (Emmerson, Miles and Stockton, 2020). So to some extent, the sharp increase in debt interest payments since the pandemic reflects this longer-term increased vulnerability to interest rate changes, as a result of changes in the structure and financing of debt.

Social security spending

Another area where spending growth has been considerably revised, as shown in Figure 8, is spending on social security benefits and state pensions. This was not expected to rise as a proportion of GDP between 2019–20 and 2024–25 in March 2020; it is now expected to have risen by 1.2% of GDP over this parliament, reflecting both higher spending and lower-than-forecast GDP growth. This growth is shown in Figure 11. Real spending on social security benefits was expected to grow by 1.4% on average each year during this parliament in March 2020, and in fact has grown by 3.0% on average each year.

This increase in spending can broadly be attributed to two main factors. First, there is the increase in benefit caseloads, largely those related to health and disability, which has been documented in recent work at IFS (Ray-Chaudhuri and Waters, 2024). This work concludes that the rise in caseloads is unlikely to represent an entirely transitory pandemic-related phenomenon, and the OBR also expects that a higher disability benefit caseload will persist through its forecast period (until 2028–29).

Second, there is the fact that most benefits are uprated with CPI inflation. This has grown faster over this parliament than the measure of inflation used in general to think about real-terms changes in public spending, the GDP deflator. This is because the GDP deflator measures domestic inflation only, and a large part of the increase in inflation after 2022 was a result of price rises for imported energy (and related goods). So this higher-than-expected benefit uprating relative to the GDP deflator will also have pushed up expected spending on social security.

When excluding the state pension (which is uprated in a different way), spending on social security benefits in 2024–25 is now forecast to be around £35 billion higher than was forecast in March 2020. Part of this is because economy-wide inflation is higher. Of the roughly £19 billion increase after adjusting for this, higher uprating explains around 30%. A large part of the remainder will reflect rising caseloads, particularly for health- and disability-related benefits.

Thinking separately about the state pension, spending is roughly £13 billion higher in cash terms in 2024–25 than was forecast in March 2020. Again, partly this is to do with higher economy-wide inflation. The increase in expected spending after adjusting for this is essentially entirely explained by higher-than-expected uprating (via the triple lock mechanism). Other factors, such as changes in caseloads, contributed very little to the rise.

Figure 12 splits the rise in the share of national income spent on social security benefits in 2024–25 into the contribution of different areas of benefit spending. As shown, higher spending on universal credit and legacy equivalents accounts for the largest part of the increase in expected social security spending in 2024–25, contributing around two-fifths of the overall 1.3% of GDP rise in social security spending. This increase will, as explained above, be driven both by rising caseloads for health-related incapacity benefits since the pandemic and by the higher-than-expected CPI relative to domestic inflation meaning benefits are uprated by more, both in cash terms and relative to the GDP deflator.

Higher expected spending on disability benefits has contributed the second-largest part, contributing 0.36% of GDP to the rise. As with universal credit, this will have been increased both by higher CPI relative to the GDP deflator and by higher caseloads, with higher caseloads likely to be an especially important part of the rise here.

Pensioner spending, including the state pension, contributes 0.32% of national income. The rise here, as described above, is explained almost fully by higher-than-expected uprating (under the triple lock) relative to the GDP deflator. There are small contributions from other benefits.

Summing up, around a fifth of the rise in spending as a share of GDP during this parliament was pre-planned. Of the further four-fifths, increases in debt interest spending and welfare spending have been most important in driving up the size of the state. Higher debt interest spending, as we have seen, was initially largely a result of inflation, but is now higher than its previously forecast levels as a result of elevated interest rates. With debt set to continue at elevated levels for the foreseeable future, persistently higher interest rates – or even just the risk of them rising sharply once again – will continue to be a concern for planning government spending. Higher welfare spending is largely a result of higher CPI inflation relative to the GDP deflator and higher health and disability benefit caseloads, the second of which seems unlikely to be a short-lived phenomenon.

4. Looking to the future: 2024–25 to 2028–29

Over the rest of this decade, the size of the state is set to fall slightly from its level in 2024–25. Spending is forecast to fall by 1.6% of GDP from 44.0% in 2024–25 to 42.5% in 2028–29. This still leaves the state larger by 2.9% of GDP (£80 billion in today’s terms) at the end of the forecast period than its size before the pandemic, when spending was 39.6% of GDP. The growth is despite tight public spending plans currently pencilled in after this year.

To some extent, as Figure 13 shows, the increased size of the state is a result of consistently higher debt interest payments post-pandemic. While spending minus debt interest payments as a share of GDP is still set to be at a higher level in 2028–29 than its pre-pandemic level, the difference is much smaller, with spending outside of debt interest set to be 1.2% of national income (or £33 billion in today’s terms) higher in 2028–29 than its pre-pandemic level.

As well as debt spending remaining higher than its pre-pandemic level, spending on social security benefits and state pensions is set to remain elevated as a share of GDP, as described in the previous section. Departmental spending is also set to remain higher than pre-pandemic levels, although fall slightly as a share of GDP between 2024–25 and 2028–29.

There is reason to think that even this elevated level of spending as a proportion of GDP may be unrealistically low. Spending plans pencilled in for departments after 2024–25 are tight, with day-to-day funding set to grow in real terms by 1% each year, and investment funding set to be held constant in cash terms (and so cut in real terms).

Plans for day-to-day spending imply cuts to some departments, since the 1% real-terms growth will not be spread evenly across all areas. Assuming that NHS spending grows in line with the long-term workforce plan, that schools spending is held flat in real terms, that defence and aid spending grow with GDP, and that new childcare commitments made last year are fully funded, other areas of spending are set to face real-terms cuts of around 1.9–3.5% per year, as shown in Figure 14. To avoid cuts to day-to-day spending, a top-up of £10–20 billion in 2028–29 would be needed.

Making these cuts would not be impossible. But they come at a time when lots of departments have already faced significant real-terms cuts. Figure 15 shows the evolution of real-terms day-to-day spending for comparable departments between 2009–10 and 2024–25. The budgets for Health and Social Care, the Home Office, and Education are above their 2009–10 levels in real terms, but other departments have fared less well.

The majority of departments are still at lower real-terms day-to-day funding levels than they were before the period of austerity in 2009–10, many of them to a significant extent. Housing and Communities day-to-day funding is set to be 65% lower in real terms in 2024–25 than it was in 2009–10. The Department for Work and Pensions (which manages the administration of benefits, including the costs of running Jobcentres and employment schemes; its funding is not the actual benefit payments), the Department for Transport, the Foreign, Commonwealth and Development Office, the Department for Environment, Food and Rural Affairs, and the Law Officers’ Departments all face funding levels in 2024–25 that are more than 20% below their real-terms level in 2009–10.

We are here comparing total budgets, but real-terms cuts in spending per person have been even sharper, since the population has grown since 2009–10. For some areas of spending – for example, housing or the NHS – it seems clear that growth in population will add directly to pressures, implying that funding has been even more stretched than apparent from these figures. In areas such as defence, population growth may not feed through directly into increased demand for services. In other cases, the most sensible measure may be somewhere in between – for schools and prisons, for example, the relevant populations are pupils and prisoners, which do not necessarily increase one-for-one with the population at large – but departments will be under somewhat more pressure than the overall figures suggest.

Unprotected areas – such as criminal courts, prisons, police services, and higher education – are thus unlikely to be in a position to easily absorb further cuts. The Institute for Government’s 2023 Performance Tracker, which examines the performance of nine key public services, found that all except schools were performing worse immediately pre-pandemic than in 2009–10, and that all except children’s social care worsened during the pandemic itself. This is a challenging baseline from which to make further significant spending cuts without accepting changes to the scope or quality of services offered.

Demographic change will make things harder. The UK population is both growing and ageing. Higher population growth, even without changes to the age structure, implies more pressures on some public services, as described above. The ageing of the population puts specific pressures on spending on health, long-term care and pensioner benefits (including the state pension). The share of the adult population above 65 is projected to rise from 24% in 2020 to 31% in 2050. The share above the state pension age is projected to rise from 24% in 2023 to 27% in 2050, having not risen since 1975 as a result of state pension age increases and immigration (Cribb et al., 2023). These demographic facts imply that age-related spending is likely to be on a rising path in the medium and long run just in order to provide the same level of state support. Absent changes in the desired scope of the state, a larger state is implied.

These demographic pressures come at a time of heightened geopolitical uncertainty, with defence spending set to continue its recent upward increase as a share of GDP. As described in Section 2, the ‘peace dividend’ since the end of the Second World War, with defence spending falling as a fraction of national income, allowed an upward trend in health spending to be maintained within a stable level of overall spending. Defence spending clearly could not have continued on the same downward trend for much longer, even without change in the geopolitical environment. In 2006, NATO, of which the UK is a member, set a target for defence spending of 2% of GDP, creating an effective floor on defence spending. After Russia’s invasion of Ukraine in 2022, there was an international shift towards increased defence spending. The Conservative Party has now committed to increasing defence spending to 2.5% of GDP by 2030, and Labour have said that they will do the same when resources allow.

To cut spending as a share of GDP, given these factors, would require a genuine reduction in the scope of the state. There are some options here. One option is ‘last in, first out’: we could scrap the recent introduction of a lifetime cap on social care costs and the expansion of childcare subsidies. Alternatively, we could begin to charge for things that are currently free (e.g. road-use charging or loans for post-16 education), begin to means test things that are currently universal (e.g. disability benefits or the state pension), or just pare back the boundaries of the system (e.g. stop paying child benefit for secondary-school-aged children or increase the threshold that determines whether treatments should be available free at the point of use on the NHS).

But there is no sign that either of the main parties is willing to make these kinds of changes to the scope of the state. Indeed, in the face of these pressures – public services struggling after a decade of austerity followed by a pandemic, demographic pressures on health and social care spending set to intensify, and stronger calls on defence spending – there have been a range of recent spending commitments, accepted by both parties, which would increase, rather than cut, the size of the state. The recent expansions in the welfare state, in terms of childcare and social care, exemplify this tendency.

Given this, current spending plans after 2024–25 look unrealistically tight. An alternative path for spending as a share of GDP would avoid making real-terms cuts to investment spending as well as to day-to-day spending for unprotected departments. This scenario is illustrated in Figure 16. Here we assume that investment spending is topped up in 2028–29 by £18 billion and day-to-day spending by £13 billion, our central estimate for the amount needed to protect unprotected departments from real-terms cuts. As shown, this roughly £30 billion top up does not quite prevent the size of the state from falling between 2024–25 and 2028–29, but more than halves the extent to which it is set to fall, from 1.6 percentage points to 0.6 percentage points.

This might be considered a central scenario for the amount of money needed to prevent real-terms cuts to departmental spending, but there is uncertainty around these figures. And this scenario will remain difficult for some departments, which are facing strain under current real-terms funding allocations.

Any top-ups to spending will require some combination of higher borrowing or higher taxes, absent an improvement in the outlook for the public finances. The specific trade-offs involved can be explored using the new IFS ‘Be the Chancellor’ tool.³  During the campaign so far, both main parties have heavily implied that higher taxes will not be used to top up spending totals, and have explicitly ruled out a range of tax rises. Both parties have committed to a set of fiscal rules that would preclude them from higher borrowing, at least given current fiscal forecasts, which are analysed more fully in Emmerson and Stockton (2024). Neither party has hinted at any desire to narrow the current scope of the state.

Keeping to tight spending totals, in the face of the range of pressures on public services discussed, without paring back what the state does – what we might call muddling through – leads to rationing or a deterioration in quality, as we are already seeing in many public services. This might happen via growing waiting lists, further deteriorations in the estate or increased difficulties with public sector retention.

5. Conclusion

Absent significant changes to growth, or a contraction in the desired scope of the state, higher spending seems here to stay. The rise in spending as a proportion of national income in 2020–21 to a post-war peak of 53.1% was clearly a pandemic phenomenon. But many of the conditions allowing spending to remain constant for such a prolonged period between the mid 1950s and the COVID-19 pandemic were likely to unwind even without the pandemic. And some of the impacts that the pandemic may have had on spending seem unlikely to be entirely or immediately transitory – most obviously, the higher numbers on incapacity and disability benefits do not seem likely to dissipate in the short term.

Even before the full-scale invasion of Ukraine, defence spending had stopped falling, after the UK government pledged to spend at least 2% of GDP on defence each year as part of its NATO commitments. Health spending, meanwhile, was likely to continue its upward trajectory, in the context of well-publicised pressures on the NHS, ambitious plans for health committed to by both parties, and an ageing population. An increase in the debt stock, and a shift in its structure and financing, after the financial crisis made the state more vulnerable to sharply elevated debt interest payments in the face of rising interest rates, as happened during the last two years. Parts of welfare spending – in particular, increased pensioner benefits – were affected by long-term demographic trends, and increased caseloads for health-related benefits seem increasingly unlikely to subside swiftly. After a decade of austerity, the strains of a pandemic and the pressures of a rising population, public services were increasingly struggling to deliver what was asked of them within tight funding limits.

Parties should be honest about the choice faced by the next government. Current tight plans for departmental spending are unlikely to be sufficient to keep public service performance from seriously degrading, unless the scope of public services is narrowed. Further top-ups to spending require tax rises or increases to borrowing, neither of which either main party has indicated they will accept. The trade-offs here cannot be solved by denying their existence.