Stop the freeze: permanently re-link housing benefits to private rents

The Budget is the perfect opportunity for the Chancellor to prevent renters from being pushed deeper into poverty by permanently re-linking frozen housing benefits to private rents.

Unless the Chancellor explicitly chooses to unfreeze Local Housing Allowance (LHA) and re-link it to local rents, it will remain frozen in cash terms for 2025 and beyond, because that was the policy of the previous Government. The Autumn 2024 Budget is the perfect opportunity to commit to permanently re-link housing benefits to local rents.

Our new research calculates the impact of this freeze alone: if LHA remains frozen over this parliament, on average, private renters on housing benefits will be around £700 worse off per year. Fifty thousand renters will be pulled into poverty, 60,000 will be pushed into deep poverty and 80,000 (including 30,000 children) will be pushed into very deep poverty.

If the Government is serious about ending the need for emergency food parcels and tackling child poverty and homelessness, the right thing to do is to permanently stop the freeze and re-link housing costs to housing support.

Recommendation

• The Government must unfreeze Local Housing Allowance and permanently re-link it to cover the bottom 30th percentile of local rents.

Around half of all private renters in receipt of housing benefits are in poverty, but for much of the last decade, the benefit system has failed to protect them from the rising cost of rent.

New research conducted by Manchester Metropolitan University and commissioned by the JRF shows that the last 14 years of changes to the LHA (2011 to 2025) have made private renters £684 per year on average worse off.

We have also found that as a result of these changes, 60,000 more private renters are in after-housing costs poverty (that is, an income below 60% of the median) than would be had the changes not happened. This is a clear indicator that private rents are too high relative to incomes, compounding an already dire affordability crisis. A further 90,000 have also been pushed into deep poverty (an income below 50% of the median) and 60,000 into very deep poverty (an income below 40% of the median). More people are therefore entering more severe forms of poverty due to their housing costs.

LHA (the rate used to determine how much housing benefit private renters can receive in their local area) has seen a raft of changes since 2011, including being frozen in cash terms for 7 out of the last 14 years. These have reduced its adequacy. What is more, the previous Government re-froze LHA from April 2024, and this is now the presumed policy in Office for Budget Responsibility forecasts. As such, unless the new government explicitly chooses to unfreeze LHA in the next Budget and re-link it to the actual cost of local rents, it will remain frozen from April 2025 and beyond.

Our new research calculates the impact of this freeze. It finds that if LHA remains frozen over the course of this parliament, on average, private renters on housing benefits will be around £700 worse off per year by 2029, 50,000 renters will be pulled into poverty, 60,000 will be pushed into deep poverty and 80,000 (including 30,000 children) will be pushed into very deep poverty.

The Government has made a welcome commitment to ending the need for emergency food parcels and to tackling child poverty and homelessness. It cannot achieve this without action on the cost of housing. The Chancellor must choose to end the freeze in the next Budget.

Housing costs are a major driver of poverty and hardship.

As shown in Figure 1, around half (52%) of all people in private renting households that receive housing benefits are in after-housing costs poverty, compared to 21% of everyone in the UK (Department for Work and Pensions, 2024). However, 25% of all private renters receiving housing benefits are only in poverty due to their housing costs, much higher than the 14% of social renters on housing benefits. This emphasises just how significant housing costs and the support that private renters get to help pay for them matter (Department for Work and Pensions, 2024). Private renters in receipt of housing benefits are disproportionately likely to be in poverty, making it all the more important that the benefit system works as it was originally designed to – by aligning housing support with housing costs.

LHA is the rate used to determine the maximum amount of support for housing that private renters in receipt of Universal Credit (UC) or Housing Benefit can receive.

Introduced in 2008, it has been reformed significantly since, as shown in Figure 2. In 2011, it was reduced from covering the median amount of rent in a local area to the 30th percentile. It has been frozen for 7 out of the last 14 years, and for 3 of the years it was not frozen, it was capped by the Consumer Price Index (CPI) or at 1%, rather than at the level of local rents.

We have calculated the impact of keeping LHA frozen, looking at both the poverty impact and the additional costs that private renting families in receipt of housing benefits will be forced to cover.

On average, private renters in receipt of housing benefits will be £243 per year worse off in 2025/26 if LHA remains frozen. By the end of this parliament (2029), if LHA is not unfrozen, this will rise to £703 per year (as shown in Table 1). For some households, such as a family with a working-age couple and children, the impact is greater, with them being £340 worse off in 2025/26 alone and rising to £882 by the end of the parliament.

In 2025/26, 20,000 more private renters will be pulled into poverty if LHA remains frozen, including 10,000 children. But more will be pulled further into more severe forms of poverty, with 40,000 more being in deep poverty and a further 10,000 in very deep poverty.

By the end of the parliament, in 2029, 50,000 more private renters would be in poverty, including 30,000 children. Sixty thousand more would be pulled into deep poverty (including 10,000 children) and a huge 80,000 into very deep poverty (including 30,000 children).

This briefing is a warning to the Government that if it does not stop the freezing of LHA and fails to permanently re-link housing benefits with local rents, it will be actively choosing to push further costs onto vulnerable private renters and could drive 80,000 private renters into the most severe form of poverty over the course of this parliament.

This Budget is the perfect opportunity for the Chancellor to set themselves apart from the previous Government’s approach to housing benefits and ensure that vulnerable private renters have the best chance of being able to pay their rent, avoid hardship and avoid homelessness.

Secure data

The UKDS has recently made available a secure version of the FRS through its secure data services (UK Data Service, n.d.). In the secure version of the FRS, geographical information at a more detailed level, including BRMA, is provided in the dataset. However, there are strict conditions on access to the data.

Researchers have to undergo accredited training on data handling, the research project has to satisfy a public benefit test and must be approved by the data owner, the DWP.

Furthermore, users cannot download the data to their own computers: instead, they must use a secure connection to the UKDS servers, carry out analysis remotely on UKDS machines, and all outputs from the analysis (tables, charts, and so on) have to be approved individually by the UKDS to ensure that none of them contain any risk of individuals being identified.

IPPR Tax-Benefit Model

For this project, we sought permission from the DWP and the UKDS to install the Institute for Public Policy Research (IPPR) Tax-Benefit Model within the UKDS secure servers and to use this model with the secure version of the FRS to carry out analysis of the effects of changes to LHA rules since 2010.

The IPPR Tax-Benefit Model is a tax-benefit microsimulation model used widely outside government. Its current users include the IPPR, Resolution Foundation, New Economics Foundation, Joseph Rowntree Foundation, Legatum Institute, Centre for Social Justice, and the Fraser of Allander Institute. It is used by these organisations to analyse the effects of government policy, design their own policy proposals, and to forecast the distribution of disposable income. The model is maintained and developed by the Policy Evaluation Research Unit at Manchester Metropolitan University (Policy Evaluation Research Unit, n.d.).

Running the model for 2 scenarios – for example 2 sets of policy rules – allows estimation of the change in disposable income for each family arising from that policy change. By looking across the representative sample, we can estimate total savings/costs to government and the distributional effects of policy – who gains and who loses. We can also estimate the number of households, people, children and pensioners in poverty. Note that estimates of poverty are calibrated to be consistent with official statistics on poverty. Notable technical features of the simulation are discussed below.

Uprating

The latest year of FRS data is for 2022/23 whereas typically users want to analyse the effects of policy change that may take place next year (2025/26) or in subsequent financial years. To enable this to happen, the 2022/23 FRS data is ‘uprated’ to April of the chosen policy year by increasing financial values in the data in line with official forecasts of inflation and earnings growth. For example, all earnings figures are uprated to the chosen policy year in line with Office for Budget Responsibility (OBR) forecasts for average earnings growth whilst rents are uprated in line with OBR rent forecasts (see Office for Budget Responsibility, 2024). Where analysis is taking place for a historical year for which FRS data exists, the FRS of that year is used, but uprated to April of the relevant financial year to control for the effects of within-year inflation.

Employment adjustment

For years from 2023/24 onwards, age band-specific employment rates within the sample are adjusted to equal those seen in 2019/20. The underlying assumption is that 2022/23 employment rates reflect a post-Covid situation that is still unwinding. This adjustment is carried out using a dynamic microsimulation approach. People who are out of work are randomly chosen to move into work so as to hit the desired age band-specific employment rate, randomly assigned specific hours of work based on the observed distribution of hours of work. They are then assigned an hourly pay rate using a wage equation with a random term to account for unobserved variation in wages.

UC vs legacy benefits

Initially, each family’s entitlement is simulated for the legacy benefits that UC replaced. Rates of transition to UC for each legacy benefit and for new claims are used as probabilities which are compared against a random number drawn for each family to decide whether they should be assessed against UC entitlement conditions or legacy benefits.

Tax and benefit simulation

Income tax, national insurance, local taxes, and a wide range of benefits are simulated by applying the rules of the tax and benefit system to each person and family using actual rates and thresholds up to 2024/25 and forecast rates/thresholds beyond that. By default, the assumptions in the Treasury’s baseline forecasts, as set out in the appendix of the Policy Costings document published at the time of each budget (HM Treasury, 2024), are used as the basis for uprating rates and thresholds. The exception is where the government has announced a departure from these baseline assumptions such as the freeze in the Income Tax Personal Allowance and Higher Rate Tax Threshold.

Benefit take-up

Means-tested benefits are subject to a take-up simulation. The latest published take-up statistics for each means-tested benefit are used as aggregate take-up rates and an aligned dynamic microsimulation is used to decide take-up for each family. A logistic take-up equation along with a suitable random term to account for unexplained variation is used to determine the underlying take-up propensity.

Entitled families are ranked by propensity and the correct number is chosen so that the weighted total of recipients equals the total implied by the aggregate take-up rate. Note that take-up statistics for working-age means-tested benefits have not been published by the DWP for several years and so the aggregate take-up rates used in the model may be different from the reality of current take-up rates but this is the best information available. Also, the DWP has never published UC take-up rates. So, we use the highest of the take-up rates of the legacy benefits that UC replaced as the aggregate take-up rate for UC. This was the assumption used by the DWP in its own modelling in the early days of UC.

Poverty estimates

By default, estimates of poverty numbers are calibrated to be consistent with official statistics on poverty as published by the DWP in its Households Below Average Income publication (Department for Work and Pensions, 2024b).

The approach taken is a micro-calibration in which modelled income for each family in the year of FRS data using a baseline policy scenario is compared to actual income as reported in the Households Below Average Income dataset.

The difference between the two is used as a calibration factor that is used to adjust income before calculating poverty lines based on 40%, 50% and 60% of median calibrated income. Calibrated income is then used to decide whether households are above or below the various poverty lines.

This method ensures that, when running the baseline policy scenario, poverty numbers will be the same as in official poverty statistics and so that changes to poverty arise from policy change rather than from the data and features of the modelling approach.

Counterfactual LHA scenarios

The heart of the modelling approach being used is to compare modelled income under 2 policy scenarios, calculating the change in income for each sample family and therefore the aggregate effects (on government finances, incomes and poverty numbers) across the population.

This means that, if, for example, one is comparing a scenario in which LHA rates are set to the 30th percentile of rents in each BRMA to one in which rates are set to median rents, one needs to know what the LHA rates in every BRMA and for every size group would be in both scenarios. If in the year in question, actual LHA rates were set to the 30th percentile, we can use data on actual rates but we would need to estimate what the rates would be if they had been set at the 50th percentile, and vice versa.

The broader point is that, for this modelling, we need actual data on LHA rates, data on the 30th percentile of rents, and median rents, for every size group for all 200 BRMAs for every year of analysis, including forecasts up to 2030.

Read the full technical report

Read the full technical report here, it documents the modelling approach used to assess the fiscal, distributional and poverty effects of changes to Local Housing Allowance policy between 2010 and 2025 and sets out detailed results tables.

What is Local Housing Allowance (LHA)?

The Local Housing Allowance (LHA) was introduced in 2008 as a major reform of the method for setting the maximum amount of rent that could be covered through means-tested benefits such as Housing Benefit and more recently the Housing Element of Universal Credit (UC).

In means-tested benefits, claimants renting in the Private Rented Sector with no other income receive Housing Benefit or, more recently, UC Housing Element, equal to the lower of their actual rent, or the LHA in their area for the size of accommodation to which they are entitled.

If a family rents a property that has a higher rent, their claim is assessed as if their rent were at the local LHA level. Effectively, the family has to meet any shortfall between their actual rent and the local LHA. Thus, the LHA operates as a maximum rent that acts as a cap on the amount of support that private renters can receive through means-tested benefits.

The number of people in the family determines the size of accommodation (measured by number of bedrooms) to which each family is entitled. Under the original 2008 Local Housing Allowance scheme, each claimant’s LHA (or maximum rent) was set to the median rent in the claimant’s local area for that number of bedrooms. This means that half of the properties of that size in that area were available to claimants without facing a shortfall.

Whilst those with no other income receive the lower of their LHA or actual rent as their Housing Benefit or UC Housing Element, those with some income have their benefit reduced (‘tapered away’) as their income rises.

Changes to LHA rules

However, almost as soon as it was introduced, the rules for setting LHA rates were subject to a series of reforms designed to reduce Housing Benefit expenditure. For example, from 2011, the LHA in each area for each size group was reduced from the median rent to the 30th percentile. Further changes since then have in many years reduced LHA rates below this 30th percentile level. See Table 1 for a listing of changes to rules for setting the LHA (also see Hobson, 2023).

Modelling the effects of changes to LHA rules: data issues

The standard tool for assessing the impact of social security changes is tax-benefit microsimulation modelling. In this approach, a tax-benefit model is used to simulate the tax paid and benefit received by every family in a representative sample of the population in order to estimate their disposable income.

By comparing simulated disposable income under 2 different sets of policy rules it is possible to estimate the effect of a policy change on each family in the sample. By looking across the representative sample, it is possible to identify the total costs/savings to government of a policy change, the distributional effects of the change, which groups gain or lose more or less, and the effects on poverty.

In the UK, the best data source for such modelling is the Family Resources Survey (FRS), run by the Department for Work and Pensions (DWP) (Department for Work and Pensions, 2024a), which was introduced in the early 1990s partly for the very purpose of improving the government’s tax/benefit modelling capability.

Anonymised data from this survey is made available to researchers outside government through the UK Data Service (UKDS). If it were possible to simulate benefit entitlement under 2 different rules for setting LHA rates, it would be possible to estimate the change in income for each family, and therefore to estimate the fiscal, distributional and poverty effects of the change.

However, for setting LHA rates, the UK is divided into 200 Broad Rental Market Areas (BRMAs). Separate LHA rates are set for accommodation of each size group within each BRMA.

Therefore, to be able to model the effects of changes to LHA policy using a tax-benefit model, one needs to be able to identify any shortfall between the rent each family actually pays and the LHA rate applicable to that family. For the latter, it is necessary to know the BRMA in which each sample family lives. However, for confidentiality reasons, the version of the FRS that is routinely released to researchers outside government through the UKDS has geographical information on each family limited only to the region or country in which they live rather than more detailed information on the BRMA in which they live.

The consequence is that, with this data, it is not possible to provide an accurate simulation of the effect on each sample household of a change to LHA policy. The result is that changes to the LHA have been subject to significantly less scrutiny from outside government than other major reforms to the social security system.

One welcome aspect of the 2008 LHA reforms was that, for the first time, maximum rents, LHA rates, for each size group in each BRMA were published.

This has enabled studies that have looked at local rent data to assess the extent to which LHA rates have diverged from prevailing rents (for example Berry, 2023 and Buckle, 2017). Whereas in the original LHA scheme, setting LHA rates to median rents in each BRMA meant that half of the private rental market was available to Housing Benefit claimants, the switch to the 30th percentile reduced this to 30% of the market. Further reductions in many years reduced this proportion further, which is likely to have led many more families to experience shortfalls between their Housing Benefit and the rent actually paid.

Other studies (including Berry, 2023) have used survey data to assess the extent to which there are differences between actual rent paid and the amount of Housing Benefit or UC Housing Element received. However, for people with some income, this could either be because of shortfalls between the LHA and rent paid or because their income is sufficient for some of the Housing Benefit or UC Housing Element to have been tapered away. Given this uncertainty, this data cannot be used to estimate aggregate statistics such as the effects of these shortfalls on numbers of people in poverty.

Department for Work and Pensions (2024) Households below average income: for financial years ending 1995 to 2023

Earwaker, R. and Johnson Hunter, M. (2024) The scale of the challenge: JRF’s pre-election cost of living tracker

Office for Budget Responsibility (2023) Economic and Fiscal Output November 2023, page 66

Schmuecker, K. (2023) What protects people from very deep poverty, and what makes it more likely?

Valuation Office Agency (2024) Local Housing Allowance (LHA) rates applicable from April 2024 to March 2025