- The researchers have developed new and more effective ways of measuring and monitoring homelessness
- They have harnessed a sub-regional housing market model to develop medium-longer term projections of housing markets and homelessness outputs and quantify the impacts of a wide range of potential policies on outcomes
- The research has created impact on key stages of the evolution of homelessness policy in England, Scotland and Wales including on legislation, statistics and targeted programmes
Homelessness has presented a threefold challenge to policymakers in the last decade.
Firstly, despite ostensible goals to reduce or even end some forms of homelessness, the actual scale of the problem has continued to increase (particularly in England) or to remain stubbornly high.
Secondly, the conventional ways of measuring homelessness have continued to have major flaws, increasingly recognised by bodies such as the UK Statistics Authority (UKSA) and the National Audit Office (NAO), although some improvements have been made (partly in response to this team’s research).
Thirdly, there is uncertainty and contention about the most effective means of reducing and responding to homelessness, so that it is difficult to make rational decisions about the most appropriate policy choices.
This research programme, by providing a continuing monitoring and analysis input, by developing a new and more consistent measurement framework, and especially by demonstrating through simulations of future outcomes what the likely impact of different policies would be, has engaged decisively with all three challenges.
Within the overall Homelessness Monitor Programme from 2011 there was a regular reviewing of the general economic, demographic and policy factors affecting homelessness, with a particular scrutiny of the impacts of the Financial Crisis/Great Recession and its aftermath and the implementation of welfare reforms and cuts.
Interviews and a regular survey of local authority homelessness services collected systematic and qualitative insight into trends in actual homelessness pressures and responses. A strand of work on ‘hidden homelessness’ interrogated key datasets from the UK Data Service (Labour Force Survey – LFS, English Housing Survey – EHS, Understanding Society – the UK Household Longitudinal Study – UKHLS) to track trends in concealed households, overcrowding, sharing and household headship rates.
From 2017 the concept of core homelessness was developed jointly with Crisis, and this entailed investigating a range of sources and approaches, building on the above as well as on an earlier key study (‘Estimating Housing Need’) for the then Department for Communities and Local Government (2010) and a separate strand of work on modelling the housing market at sub-regional level.
This new sub-regional economic model of housing markets built on previous work (notably Meen 2011; Leishman et al 2008, ODPM 2005, Bramley & Leishman 2005; as reviewed in Bramley 2013) but went beyond it in terms of using a more appropriate geographical framework (sub-regional ‘housing market areas’), explicit modelling of the supply process as a function of planning, economic modelling of demographic change, and linking component models in an integrated simulation approach which takes account of spatial interaction between markets.
The fullest account of the model in a peer-reviewed monograph length article is in Bramley & Watkins (2016), but the model has continued to evolve to the version currently in use for projecting homelessness.
Key components of the model are:
- econometric functions to predict housebuilding output
- house prices, rents, private rental lettings and tenure share
- housing vacancies, social housing relets
- poverty measures
Some of these are panel models fitted to annual data at market area of local authority level, while others are micro-level models (typically logistic regressions) with area-level market variables attached.
Key inputs include macro-economic assumptions and recent output and productivity trends and sub-regional levels, planning targets, and overall national population trends, but internal migration and household formation are endogenous, as are the emergent tenure shares and affordability conditions.
The homeless projections extension of the model adds functions to predict total homeless applications, temporary accommodation, evictions, rough sleeping, ‘unsuitable’ temporary accommodation, and sofa surfing. Ways have been found to simulate a range of specific or more general policy changes, which feed into typically 10 variant scenarios around a baseline prediction.
In 2020-21, the effects of Covid-19 and special measures had to be worked into and through this system.
Tailored variant models have been developed within this framework to cover Scotland and Wales, drawing on some specific micro data for these countries (e.g. Scottish Household Survey) as well as comprehensive local authority level panel data. For all three GB countries, models were substantially recalibrated on recent data for 2021 Monitors.
Data used from the UK Data Service collection
Key data from the UK Data Service has included a range of large scale repeated or panel surveys including
- British Household Panel Survey Special Licence Access (with LA codes) 1991-2009.
- UK Household Longitudinal Survey (‘Understanding Society’) Special Licence Access with Local authority codes, Waves 1-10
- Survey of English Housing 1997-2007
- English Housing Survey 2008-18
- Labour Force Survey 1992-2020
- Scottish Household Survey 2000-2018
- CORE (Continuous Recording of Social housing lettings data)
- British Cohort Study (1970 birth cohort, waves up to 2012)
While the research has also used a substantial range of other data sources, including demographic and labour market data from NOMIS, statutory homelessness statistics compiled from local authority returns via Government departments, and market data from HM Land Registry and Valuation Office Agency, the above micro individual/household level datasets have been very valuable for modelling key housing decisions and outcomes.
This has included models to predict household formation, tenure choice, risk of experiencing particular housing needs (e.g. concealed household, overcrowding), particular forms of homelessness (especially sofa surfing), and retrospective experiences of homelessness.
The ability to attach housing and labour market information at local or (sub-)regional levels has been crucial to developing effective models of this kind, facilitating more effective and relevant model structures.
While contributing to significant and influential academic publications (Bramley 2016, Bramley & Fitzpatrick 2017) this analysis has also enabled predictive functions to be developed for the largest category of core homelessness (sofa surfing), while innovative joining of Understanding Society data with the ‘Destitution in the UK’ survey of users of crisis services has enabled new predictive models for rough sleeping and core homelessness as a whole.
In addition, in related research a micro-simulation approach to destitution was developed, based on Understanding Society which was then used to feed back some key response functions into the homeless projection model, for example relating to changes in benefit rates (Bramley 2020).
Official measures of homeless remain inadequate: not all core homeless people apply to councils for assistance, and official measures of rough sleeping used in England are only a fraction of reasonable estimates based on triangulation of different sources and methods of estimation.
Highlights from the Chapter 5 of the Homelessness Monitor England (Crisis, early 2022) show that more than 200,000 households experienced ‘core homelessness’ in England in 2020. This was a modest reduction on 2019, following a gradual rise over the previous 7 years, thanks to the ‘Everyone In’ initiative responding to Covid-19.
Rates of core homelessness in England are markedly above those for Scotland and Wales, which particularly reflects the situation in London as well as policy differences. Core homelessness is forecast to rise significantly in England and remain one-third above recent levels on current policies, due to expected changes in the labour and housing markets and existing welfare policies, and further increases are anticipated in the longer term, particularly in London. The government has espoused a target of ending rough sleeping by 2024, but this model suggests numbers will rise under current policy settings.
These outcomes are not inevitable, but proactive policy changes would be needed to bring numbers down. The modelling enables the most effective policies to be identified, in both the shorter and longer term.
For example in the shorter term, direct social rehousing quotas for core homeless households, significant rises in welfare benefit levels and changes to avoid the destitution-inducing aspects of the current system, including ensuring that the Local Housing Allowance covers private market rent levels, and maximising the effectiveness of local authority prevention activities would be the most effective ways to bring homelessness down.
In the longer term, in addition to the above measures, consistent and larger scale application of the Housing First model accompanied by appropriate rehab provision and a scaling down of traditional hostels would be very helpful in reducing complex need and repeat homelessness. The enhancement of social housing supply, targeted on areas with greater need, would also help the overall package, although the effects are less clear-cut than might be expected, because of induced household formation and population movements.
The model can also be used to explore the impact of wider policy strategies, such as the current Government’s ‘Levelling Up’ aspirations for regional economic performance. Our models suggests that this scenario could increase homelessness pressures in the shorter term, while offering modest benefits in the medium to longer term. We found the effects more positive in Wales and less evident overall in Scotland.
Recommendations for policy
The Government’s stated objective of ending rough sleeping by 2024 is not achievable under any of the policy scenarios or combinations tested, although rough sleeping can be substantially reduced by that date through a combination of measures, of which much the most important relate to the welfare benefits system.
One very tangible reason why this target is not achievable is the government’s adherence to policies of ‘No Recourse to Public Funds’ for certain groups of migrants, who can be shown to represent a significant share of rough sleepers, particularly in London. Therefore, there is some focus on this group in current recommendations coming from the research
Some of the policies required to achieve substantial reductions in homelessness would carry substantial cost, for example raising benefit rates or large-scale social housebuilding.
On the other hand, some measures shown to be potentially very effective would not necessarily require large extra outlays, for example effective prevention work and rehousing quotas.
The initial set of homeless projections produced in 2017 were used as a key input to a costing exercise by PWC which Crisis used to argue that their plan to end homelessness published in 2018 was feasible.
The projections model was used in 2018 to assess levels of housing provision to achieve desirable and equitable outcomes in terms of housing need, affordability and homelessness across the country.
The recommendations of this study have been endorsed by a wide range of organisations and were commended by the House of Commons Select Committee on Housing and Communities in 2020 as the ‘best available’ estimates.
The Homeless Reduction Act 2017
In 2015, Suzanne Fitzpatrick chaired an independent review panel, assembled by Crisis, the national charity for people experiencing homelessness. She had previously had a key role in both Scotland and Wales’ improvements to laws and regulations regarding homelessness. The panel called for significant expansion in legal entitlements for single homeless people in England, underpinned both by her body of research work on ‘rights-based’ approaches and the substantial qualitative work on mapping and profiling the issue she had undertaken with Glen Bramley and Sarah Johnsen.
The panel’s report underpinned a private member’s bill laid before the House of Commons in 2016. With government support, the bill passed into law as the Homeless Reduction Act 2017, coming into force in April 2018.
According to official statistics, during 2018/19 English local authorities prevented homelessness for 58,290 households
Improving official homelessness statistics
In 2015, the then Secretary of State for the Department for Work and Pensions (DWP), Ian Duncan Smith referred the Homelessness Monitor 2015 to the UK Statistics Authority (UKSA). His reasoning for this referral was he disagreed with its challenge to official statistics.
The UKSA found in favour of the I-SPHERE-led research team and launched an enquiry into the Government’s homelessness statistics. It produced a report criticising the UK Government’s approach to collating statistics on homelessness. The UKSA report drew heavily on research undertaken by I-SPERE for the Homelessness Monitor series, described by the UKSA as ‘influential’.
The enquiry and report led directly to a replacement of the previous ‘headcount’ data with the implementation of an individual case record-based national monitoring system on homelessness applications and outcomes (H-CLIC). In December 2021 the UKSA assessed these new statistics and, while acknowledging the major step forward these represented, identified six further actions for the DLUHC to address in order for these to achieve the status of National Statistics
Partly reflecting this background but also continuing concern about homelessness statistics, including the above UKSA assessment, and concerns voiced through the Inclusive Data Taskforce, the Office for National Statistics (ONS) are conducting a further investigation into means to improve homelessness statistics, with particular focus on additional ways of measuring potentially hidden elements. The team conducting this review have had extensive contact with Glen Bramley and have shown particular interest in several of the methods and survey instruments used in the 2021 and 2022 core homelessness estimates published in the Homelessness Monitors.
Improving public scrutiny
The National Audit Office (NAO), heavily influenced by the work of the I-SPHERE team, produced a high-profile report on the Government’s record on homelessness. The report directly led to the DWP and the Ministry of Housing, Communities and Local Government (MHCLG) to collaborate on understanding and addressing the underlying causes of homelessness.
Influence on targets and Government research programmes
Following a commitment in their 2017 Conservative election manifesto, the Government introduced targets for a massive reduction in rough sleeping in England.
The same Government set up a research programme with the DWP with the aim of modelling the causes or drivers of homelessness. Bramley’s research was extensively quoted from. The scoping and feasibility studies for the programme also strongly indicated that the Bramley model is the most relevant example for Britain of a policy simulation tool.
Subsequently, MoHCLG analysts used Bramley model predictions to inform their baseline projections of rough sleeping, when developing their programme, and later used his homeless projections scenarios to argue within Spending Review for changes of Local Housing Allowance, as well as holding detailed discussions about Bramley’s overall needs estimates and model assumptions, for example about affordability.
Changing policy in Scotland
There has been a significant shift towards Housing First in Scotland, heavily underpinned by I-SPHERE research.
Suzanne Fitzpatrick was a member of the Scottish Government’s Homelessness and Rough Sleeping Action Group (HARSAG). The group’s recommendations supported significant investment by the then Minister for Local Government, Housing and Planning, Kevin Stewart, into ‘rapid rehousing’ for homeless people (£21m) including £4m for Housing First services in five Scottish cities.
As a direct result of her involvement in HARSAG, Fitzpatrick now chairs a Prevention Review Group with the aim of producing more legislative proposals to improve public bodies’ approaches to preventing homelessness.
These policy changes are supported by Glen Bramley’s modelling work which shows that three of the key policy measures which reduce homelessness, including rough sleeping, are:
- Rapid rehousing quotas
- Greater prevention effort
- Greater use of Housing First
Changing approaches to social and affordable housing
Strands of Bramley’s research have fed into a policy shift around the building of new social housing. He shared a platform at the Crisis 50th anniversary conference in 2017 with the Secretary of State, with Bramley presenting his homelessness projections research and the minister speaking of the need to tackle homelessness.
This research, along with the Housing supply requirements research massively input into Crisis’ 2018 ‘manifesto’ Everybody In: How to end homelessness in Great Britain. Included in the report were the 160,000 core homeless estimates and the annual need for 100,000 new social rented housing, as well as discussion of a ‘predict & prevent’ approach, and a costing of the programme by PWC based in part on Bramley’s projected numbers. The social housebuilding targets have been supported by many organisations including Crisis, Shelter and the Welsh Government
The 2017 UK Government produced a Green Paper A new deal for social housing which committed to the building of more social and affordable housing. The Government made related announcements which allowed social housing to be funded through an expanded ‘affordable homes programme’, as well as allowing councils to borrow to fund new council house building.
The National Housing Federation (NHF) produced a report based on additional work by Bramley evidencing that for 3.8 million people across 1.6 million households social housing was the most suitable approach for them to afford rent. This figure is half a million higher than the extensively publicised number of people on official waiting lists. This work also led to a request from the Ministry of Housing, Communities and Local Government (MoHCLG) in November 2020 to deliver a seminar to analysts and policy staff from the department.
Bramley’s work for the NHF and Crisis fed heavily into a report by the cross-party House of Common Select Committee on Housing, Communities and Local Government in July 2020 (on ‘Building more Social Housing’ (HC173) A key section of this report (paragraphs 47-50 and 53-54) highlights and quotes extensively from Bramley’s NHF/Crisis study and endorses it as its recommended target basis, quoting support from independent academics as well as the Affordable Housing Commission and other organisations including the Chartered Institute of Housing (CIH).
Its key recommendation (para 53) is to assert in bold that ‘There is compelling evidence that England needs at least 90,000 net additional social rent homes a year’ (Bramley’s figure) and that the government should have published targets for each affordable tenure
To date, while there has been support from ministers for more affordable and social housing, there has been no official commitment to a target.
Read the research
Bramley, G & Watkins, D 2016, ‘Housebuilding, demographic change and affordability as outcomes of local planning decisions: exploring interactions using a sub-regional model of housing markets in England’, Progress in Planning, vol. 104, pp. 1-35. DOI: 10.1016/j.progress.2014.10.002
Bramley, G 2016, ‘Housing need outcomes in England through changing times: demographic, market and policy drivers of change’, Housing Studies, vol. 31, no. 3, pp. 243-268. DOI: 10.1080/02673037.2015.1080817
Bramley, G. & Fitzpatrick, S. (2017) ‘Homelessness in UK: Who is most at risk?’ Housing Studies, vol 33 pp. 96-116. DOI:10.1080/02673037.2017.1344957
Bramley, G 2019, Housing supply requirements across Great Britain for low-income households and homeless people: Research for Crisis and the National Housing Federation; Main Technical Report. Heriot-Watt University, Edinburgh. DOI: 10.17861/bramley.2019.04
Bramley, G. 2021 Research on Core Homelessness and Homeless Projections: Technical Report on New Baseline Estimates and Scenario Projections. Edinburgh: Heriot-Watt University. DOI: 10.17861/fex5-jg80
Fitzpatrick, S., Watts, B., Pawson, H., Bramley, G., Wood, J. Stephens, M. & Blenkinsopp, J. (2021) The homelessness monitor: England 2021. Report. London: Crisis.
Research funding and partners
- ‘Sub-regional and cross-regional market models feasibility study’.(2009-10)
National Housing and Planning Advice Unit
- ‘Gloucestershire Housing Affordability Model’ (2011-12)
Gloucestershire County & Districts Council
- Policy Modelling for JRF Anti-Poverty Strategy’. (2017-18)
Joseph Rowntree Foundation
- ‘Homelessness projections’. (2017-18)
- ‘Housing supply requirements across Great Britain for low income households and homeless people’. (2018-19)
Crisis and National Housing Federation
- ‘Homelessness Monitor’ programme 2011- (ongoing)
Crisis, with Joseph Rowntree Foundation
- ‘Homelessness Projections’, supplementary project to Homelessness Monitor (2019-21)
I-SPHERE won the Queen’s Anniversary Prize for Higher and Further Education in 2019-20 for its path-breaking and impactful research on homelessness and destitution.