Homelessness is transient and can be difficult to quantify. Nevertheless, there are substantial changes that could improve the quality and accuracy of current data collection processes. These improvements would: i) provide a more robust figure on the level of need and scale of the problem, including demographics and type of homelessness ii) cover the use and effectiveness of interventions iii) feed into funding and commissioning decisions to improve service design and delivery to address homelessness. This chapter examines current data collection methods measuring the scale and profile of:
• rough sleeping
• those receiving homelessness assistance within the statutory framework
• those falling outside it across England, Scotland and Wales.
It then suggests improvements for data collection for each country. These should ensure a more comprehensive and high quality data set to inform decision making for policy and practice and ultimately improve services and outcomes for homeless people.
The final section of the chapter argues these efforts should build on the work by the Centre for Homelessness Impact to develop a shared outcomes framework for homelessness interventions.
Rough sleeping
The official national rough sleeping statistics in England are widely interpreted as substantially understating the true scale of rough sleeping. These figures are calculated using a methodology introduced in 2010. The methodology involves counts and estimates from local authorities of the number of people thought to be sleeping rough in a local authority area on a ‘typical night’. This night is a single date chosen by the local authority between 1 October and 30 November. It is a snapshot and will not include everyone in the area with a history of rough sleeping. In 2017, 87 per cent of councils provided estimated and 13 per cent counted.
In 2015, the UK Statistics Authority (UKSA), which oversees the validity of official government data, investigated the homelessness statistics. UKSA concluded that government data on rough sleeping in England does not meet standards required to be considered ‘national statistics’; it falls short in terms of ‘trustworthiness, quality, and value.
There are also wider categories of people sleeping in precarious and dangerous situations not officially counted in the annual counts and estimates data. These include people sleeping in cars, tents and public transport.
A Heriot-Watt University report for Crisis in 2017 identified a mid-point estimate of 8,000 rough sleepers in England and a further 8,000 people under the cars, tents and public transport category. Heriot-Watt used both secondary data sources and triangulation methods to achieve these estimates.
Notwithstanding the problems with official figures in England, the statistics do show trends over time, and are best regarded as a trend analysis. They also show the rapid upward trajectory of levels since the new methodology was introduced in 2010.
Data collection of demographic information and information about gender, nationality and age have also been introduced. For example, we know that rough sleeping in London has accounted for approximately a quarter of the national problem consistently over the previous seven years. We also know that approximately 14 per cent of rough sleepers are women; and that very few (an estimated 0.1%) are under the age of 18.
The most robust and comprehensive rough sleeper monitoring data in the UK are the statistics collected routinely by the CHAIN system funded by the Greater London Authority in London. This database is able to collect ‘flows’ of rough sleepers rather than snapshot annual counts.
It allows outreach teams and services to know if someone is new to the street, a returner or a long-term rough sleeper. Data is also collected about: support needs; reason for homelessness; if they have previously been placed in homelessness services (eg emergency accommodation and longer term supported housing), and if they have experienced rough sleeping before.
Although the CHAIN database is the most comprehensive dataset on rough sleeping, it does not routinely align with statutory datasets and is only fully accessible to commissioned services in London. Consequently, data does not routinely show whether someone has approached his or her local authority for assistance before experiencing rough sleeping, nor is this reported on.
Non-commissioned services have limited access to the database. This leads to criticisms that those who are more hidden (eg women), or are ineligible for services (eg those with no recourse to public funds), are not recorded.
Some short and long-term homeless accommodation projects record outcomes on the CHAIN database, but this is inconsistent. This means we have an incomplete picture of what happens to rough sleepers in London once moved off the streets.
Scotland, unlike England and Wales, does not conduct an annual rough sleeping count. Instead rough sleeping levels are recorded in the statutory homelessness statistics when people present for homelessness assistance. When they present to Housing Options teams, individuals are asked if they have slept rough the night before or in the last three months. These figures are a measure of the ‘flow’ of people over a year, rather than the ‘stock’ or point-in-time figures in England that relate to a given night.
The weakness in the published data is that only those applying for local authority assistance will be counted and within a short timeframe of experiencing rough sleeping. This will capture a wider range of people compared to England and Wales because all eligible unintentionally homeless people are entitled to rehousing in Scotland. However, it is likely to underestimate the level and frequency of rough sleeping as many people will not present to their council after they have slept rough.
Research shows that many people sleeping rough need to be found proactively by outreach teams rather than through services waiting for individuals to come to them. It shows that people often experience other forms of homelessness before they become street homeless.
The Scottish Government’s Homelessness and Rough Sleeping Action Group has put forward recommendations on data collection. This includes introducing a CHAIN-style system to achieve real-time, by-name data sharing between the agencies working with people who are rough sleeping or at risk of rough sleeping. The system would enable frequent and regular reporting of numbers, locations and other data to support monitoring the reduction in rough sleeping across Scotland.
Welsh Government figures reflect two separate measures. These came into use in 2014 after a five-year gap in capturing any rough sleeping data at all. The first measure is a local authority estimate over a two-week period, and the second is a count on one night. The use of these complementary approaches was in recognition that conducting a nightly street count has several limitations. The Welsh Government has argued that, by comparison with a traditional street count, this hybrid enumeration approach ’provides a better understanding of the incidence of rough sleeping.
However, there are still limitations to this combined approach. It is still reliant on a snapshot estimate and the two-week log of rough sleeping activity only provides a time-limited enumeration of the issue. The Welsh Government has also recognised that the count is limited in rural and coastal areas due to the wide geographic area covered as part of a count. The Welsh Government has committed to developing a database on rough sleepers in Wales. This is the Street Homeless Information Network, under development by Homeless Link and the homelessness charity, the Wallich.
Statutory homelessness
All three homelessness legislative systems in England, Scotland and Wales collect data on households who have approached their local authority for assistance. All three data collection methods have strengths and weaknesses, but the common weakness is that data is only reported for people who have approached their local authority for help.
This approach does not capture forms of homelessness such as sofa surfing and those placed in hostels that are not recorded via statutory homelessness statistics (private hostel placements, for example). There is also no current way of linking data on rough sleepers to longitudinal statutory homelessness data. This means it is not possible to understand or quantify how many people have presented to their local authority before or after they have slept rough.
England In
England, local authorities record the outcomes of decisions for all households who apply for help with their housing when they are currently or imminently homeless. The dataset holds information on ‘formal actions’ regarding local authority assessments under the homelessness legislation.
This gives overall numbers of people being assessed or ‘decisions’. However, it only breaks down to demographics, nationality, household type, support need and reason for approaching for those households who have the full homelessness duty accepted. This is around 50 per cent of total approaches.
Three consecutive Homelessness Monitors have also asked local authorities about the overall ‘footfall’ to their services as an indicator of demand of services. It is consistently reported by two thirds of local authorities that footfall is increasing while the statutory figures have remained fairly stable during the same period. This indicates that the statutory homelessness statistics in England are not a true representation of those approaching for homelessness assistance.
Some people are either being turned away for help or having their homelessness resolved through actions that go unrecorded.
In 2009, local authorities in England began to record data on people who approached for assistance outside of the main homelessness duty. They also record how local authorities have helped people resolve their homelessness before a formal homelessness application has taken place.
Referred to as ‘prevention and relief activity’ the statistics show to some extent successful prevention action and how this has changed over time. For example, help to prevent homelessness through resolving Housing Benefit problems has increased fourfold since 2010/11.
It is useful to report on the type of prevention and relief activity that local authorities are using. But there is no way of assessing the effectiveness of the interventions, the quality of the service provided and the sustainability of the outcomes for households approaching them for assistance.
The Homelessness Reduction Act (2017) has prompted the introduction of a new system for local authorities to record prevention and relief data, called H-CLIC. This is due to report in July 2018 and will provide information about all households owed a prevention duty including reasons why the prevention duty has ended.
Find out more about preventing homelessness.
Scotland
In Scotland, homelessness statistics are collected so that each person has a unique identifying number. This allows local authorities to track households/ individuals through the homelessness system and can help identify if they have been homeless before. Local authorities can then understand how many households made a unique application for homelessness assistance. The collection method stops double counting and indicates the proportion of households making a repeat application after receiving help.
The HL1 data, which records the number of homelessness applications, is compulsory. Local authorities have to collect it from anyone they have reason to believe is homeless (or will be in 56 days). The PREVENT1 statistics were introduced in 2014. Some people recorded under Housing Options may fill out a homelessness application, but there is no statutory requirement to fill in a PREVENT1 application.
The ability to link both datasets is useful to measure an overall homelessness caseload figure, however there is varied practice in how these are recorded across Scottish local authorities. As in England, there are elements of PREVENT1 that limit the ability to drill down into the specific activities undertaken by local authorities in addressing homelessness prevention.
There are also issues with the HL2 and HL3 data used for monitoring of households placed in temporary accommodation through local authorities homelessness duties. HL3 has been developed to understand more about the length of time households are in temporary accommodation and the proportions of households needing temporary accommodation where an offer was made.
HL3 recording has only been mandatory since 2016. Due to data quality issues HL3 data has not yet been published. Early analysis shows that in some cases there is a 40 per cent discrepancy between local authorities recording of HL2 and HL3 data.
A positive development in the Scottish statutory homelessness statistics has been data linking between ‘HL1’ and health service data, originally trialled in Fife. This idea of data linkage has the potential to revolutionise our understanding of what works to achieve positive outcomes for homeless people across public services.
Data linkage and tracking individuals through homelessness datasets, and in all public services datasets, would show the extent to which services are meeting the needs of all homeless people. It would also show the cost effectiveness of interventions
In the US and Denmark, data linkage has explored patterns of service use and the cost associated with them for some time. Large-scale data merging across Great Britain could help to facilitate the cost effectiveness of services such as Housing First and Critical Time Interventions. It could also explore how to improve prevention services and integrate these across statutory services. This approach is highly recommended.
Wales
There have been changes to statutory homelessness statistics in Wales since the introduction of The Housing Wales Act (2014). Statistics are collected on the number of outcomes but not in relation to individual households. This makes it difficult to use them for statistical purposes and attributing the overall need of homelessness. The Welsh system means that each household could have up to three outcomes: prevention; help to secure accommodation (relief), and duty to secure accommodation (discharge).
Unsuccessful prevention should subsequently be assessed as homelessness (duty to help to secure accommodation). Unsuccessful relief may then be assessed as priority need (duty to secure accommodation). This partly explains why the total number of applications is higher than before The Housing Wales Act (2014) when decisions were made at a single stage.
Most strikingly, in Wales there is no longer a single figure for homelessness. This is because the same household may be counted under one or more of the preventative, relief and duty to secure categories within a single year. And the categories cannot be totalled together to ascertain an overall figure. As with English data, households are not followed through the system. There is no way of understanding the proportion of households who experience repeat homelessness and, for example, become homeless again after a prevention outcome.
Shelter Cymru have reported that under the new system some people are receiving interventions from partner agencies that do not show up in the official statistics. This is because the type of support that they receive is not being recorded. Support might include unplanned interventions by hostel staff, for example.
This means that the extent of homelessness in Wales, the amount of related work, and the funds required, may all be underestimated.
The Wales Audit Office also recently raised issues around the extent to which StatsWales data on homelessness measured the quality of service provided and local authority success rates in addressing homelessness.
Other forms of homelessness
A main constraint of official or statutory homelessness statistics across Great Britain is insufficient data. This relates to households or individuals not approaching local authorities for assistance, and those identified by outreach teams through annual rough sleeping counts. These cases are often referred to as ‘hidden homelessness’. They are generally, but not exclusively, single households who may be living in hostels or other forms of supported accommodation, squatting, living in tents, cars or other forms of transport.
Hidden homelessness can also describe the cases of people forced to live in circumstances that are dangerous or transient. They may not know from one night to the next where they will be living – for example they might be sofa surfing.
In England, Scotland and Wales there are data recorded on the number of bed spaces in hostels and long-term supported homeless accommodation. But these are all a measure of supply of this type of accommodation rather than a measure of demand or need.
Homeless Link manages a database of all homeless accommodation projects across England. This is reported on an annual basis through the publication of a larger piece of research looking at trends and outcomes of this type of accommodation.
While homelessness has been rising in England since 2010 the number of bed spaces has decreased by 17 per cent. This figure also omits numbers of night shelters from its bed space reporting and these are not routinely reported on in other datasets.
Scotland records the hostel data in the HL1. This means it can follow the household through the system. In Wales the data is recorded through temporary accommodation records in the new homelessness statistics. But this is a ‘low’ estimate; it is only those people accepted as homeless and placed in temporary accommodation who are included in the figure.
The work by Heriot-Watt University in 2017 estimates the level of these other forms of homelessness. This relies on triangulation of several secondary data sets, which extrapolate from survey data. Estimates of people in private hostels or unsupported temporary accommodation, cars, tents and public transport, caravans, squatting and people living in non-residential buildings are included.
It is not easy to access or enumerate forms of homelessness that fall outside official statistics. But there are several voluntary sector services routinely collecting data on individuals they have accommodated or helped into other forms of housing.
There are also local systems and data sets administered across Great Britain which identify numbers of people accessing the homelessness system, their support needs and the assistance they receive. Examples include: the MainStay database in the Liverpool City Region; Glasgow Homelessness Network’s Annual Homelessness Monitoring System, and the Wallich’s South Wales street-based lifestyle monitor.
Each of these demonstrates that more can be done to bring data sources together. What is missing is the national coordination in all three nations to ensure consistency across localities and a complete approach to data collection.
Improved data collection on homelessness is only part of the solution. To achieve better outcomes we also need to use data in an insightful way to commission and design services for homeless people. One of the ways of doing this is by creating a common outcomes framework.
The purpose of an outcomes framework is to ensure that the aims of policy makers and service providers are consistent.
A good example of such an approach is the Getting It Right For Every Child (GIRFEC) framework for children’s services in Scotland. GIRFEC sets out the positive outcomes sought for every child in Scotland and was established by the Scottish Government. It allows for consistent design of services towards achieving agreed outcomes (eg in physical health, safety, and educational attainment) and of reporting progress towards these outcomes.
If we are all working to a common and agreed description of ‘homelessness ended’ and of the indicators towards that goal, we will have more chance of success. This approach is strongly recommended.
The Centre for Homelessness Impact is currently developing a proposed outcomes framework. Its purpose is to help policy-makers, independent funders and practitioners to design and commission services that produce better outcomes for homeless people.
This framework will provide a consistent explanation of what it takes to achieve better outcomes for people experiencing homelessness, across areas like housing sustainability, employability, and wellbeing.
The development process for this framework is ongoing and it will be designed in consultation with the homelessness sector. It will be published within the next 12 months.
To improve data and outcomes measures across Great Britain, the following reforms are recommended.
England/Westminster
Statutory homelessness data collection should be redesigned to follow individuals through their journeys within the homelessness system
Scotland
Wales
