Chapter 1: Public Value

Background to the statistics

DLUHC has published Statistics on Statutory Homelessness in England since 1998. The statistics are used extensively by government, local authorities (LAs), and charities and are also of interest to journalists, academic researchers, and members of the public. Uses of the statistics by DLUHC include monitoring levels of statutory homelessness across England to understand the scale and causes of homelessness and to consider possible policy responses; allocating resources and monitoring LA performance in relation to implementing the Homelessness Reduction Act (HRA) and to support bids for funding from HM Treasury. LAs use the data to plan services, allocate resources, monitor performance and benchmark against other authorities. Charites and academics also use the statistics to monitor and evaluate housing policy; campaigning; and for further analysis and reporting.

The statistics were substantially redeveloped following the introduction of the HRA in 2017, as explained in this user note, which placed additional duties on English LAs to help prevent or relieve homelessness for anyone eligible for public funds. This is in addition to LAs’ main homelessness duties to provide temporary accommodation to households that are considered by LAs to be eligible and in priority need.

The legislative changes have been reflected in an enhanced Homelessness Case Level Information Collection (H-CLIC) data specification for local authorities to follow since April 2018. Whereas the previous data collection (P1E) recorded cross-sectional totals on the number of households accepted as statutorily homeless each quarter, and the number of households in temporary accommodation, H-CLIC collects detailed case-level information on households each quarter, as they progress through each of the prevention, relief, and main duty decision stages of the statutory homelessness system.

This has resulted in a more detailed case-level data set on which DLUHC’s current statistics on Statutory Homelessness in England are published each quarter, from the most recent reporting period (April-June 2021) back to April-June 2018. However, data on households in temporary accommodation is available on a consistent basis back to 1998. The statistics are accompanied by detailed LA tables and a Technical Note each quarter, with headline quarterly and annual time series available through separate live tables. A separate financial year annual report which contains detailed flows analyses is published each Autumn. DLUHC statisticians have also recently developed new data dashboards to help illustrate key homelessness measures, on topics such as the support needs of those owed HRA prevention or relief duties, and on LA key performance measures and data quality.

Enhancing accessibility and clarity to achieve broader public value

Homelessness as a concept is often misunderstood and can be misused in public debate, with people often using the terms homeless and rough sleeper interchangeably. It is also not always clear how terms like statutory homelessness relate to others such as hidden homelessness, sofa surfing, or rough sleeping. The potential for this confusion makes it particularly important that that the statistics are clear about their scope, the questions they answer, and what they include or exclude.

Users that we spoke to during the assessment recognise the value of DLUHC’s statutory homelessness statistics, seeing them as a reliable, consistent source for monitoring the statutory support available for homeless people in England. Users of the statistical bulletin told us that they appreciated the additional commentary that was added during the pandemic, to explain how the statistics collected during that period should be interpreted. The latest annual bulletin also includes links to ONS domestic violence analyses to provide broader context around increases in statutory homelessness related to domestic violence reported in the statistics.

Users told us that they particularly liked the new flows analysis at the end of the latest annual report, as this shows the potential of the H-CLIC data for answering key questions. The flows content is more accessible than that in earlier sections of the bulletin, though is situated at the end of the report as it is based on households granted prevention and relief duties in the previous year.

While the commentary around the flows analyses in the annual bulletin and additional commentary added during the COVID-19 pandemic is insightful, the overall messages earlier in the bulletin could be clearer and more accessible. The presentation of the statistics in the main part of the bulletin is primarily focused on reporting the caseload at different stages of the HRA system, rather than explaining the statistics’ main messages, or placing them in their broader homelessness context.

Both the quarterly and annual bulletins include a key points section. However, they are very long and while useful definitions for some key terms were provided in text boxes to aid interpretation, they appear to have been removed from the latest quarterly bulletin (April-June 2021). We note that some of the clearest, most accessible commentary is in the Technical Note. This means that it is not easy for non-experts to navigate or understand the statistics’ main messages, the insights they offer, or the questions they help answer. The statisticians acknowledged that users would have to understand the HRA in order to navigate the statistics.

More generally, there is a lot of supporting documentation in different places, and on different webpages, so users may end up following links around and not finding the information they are looking for. For example, the statutory homelessness time series data are available only via a separate live tables page, and the Technical Note, statistical bulletin and live tables all point to each other for ‘more information on data quality and limitations’, which is confusing.

Some expert users told us that the fact that the statutory homelessness statistics don’t capture all forms of homelessness could be clearer. These users see the statutory homelessness statistics as an important input into a broader measure of homelessness; as the statistics do not cover all homelessness, clearer guidance should be provided to help users place them within the broader homelessness picture. Users told us it would also be helpful for DLUHC to provide upfront advice in the bulletin about key strengths and limitations, how the statistics should and should not be used, or the questions that they can or cannot answer (such as in the similar approach developed for DLUHC’s rough sleeping snapshot annual statistics).

Explaining coherence between different homelessness data sources is also important for setting the statistics in their proper context within a broader homelessness picture. DLUHC’s Homelessness statistics webpage has potential to aid user navigation of the overall statistics package in an informed way and enable an understanding of where the statistics sit in the broader homelessness landscape.

However, the webpage contains a mixture of current and discontinued material, but using similar titles and with limited guidance for users on how to distinguish between the different outputs to find what they need. For example, differences between statistics on the previous (pre-HRA) and current statutory homelessness, and prevention and relief statistics; or the rough sleeping snapshot and separate management information on rough sleepers supported during the pandemic, are not clearly explained. And the homelessness notes and definitions pages linked from the homepage page relates to the old P1E data return, last updated in 2018. This makes it difficult to navigate to the most relevant data and guidance for particular uses.

Some expert users make use of the detailed data tables, rather than the bulletin. We heard positive comments that the table footnotes clearly reported the main caveats. One user commented that while the DLUHC glossary of terminology in the Technical Note is useful, it would help if these definitions were included in the data tables themselves.

Engagement with homelessness statistics users

The new H-CLIC system’s development was an innovation which drew on collaboration with a range of users and stakeholders. The redevelopment followed OSR’s 2015 assessment of DLUHC’s statutory homelessness statistics based on the old P1E system, which identified a pool of users keen to contribute views on the improvements. DLUHC statisticians engaged broadly with these users while H-CLIC developed, and intensely with LAs as both users and data suppliers, to support them through the transition. LAs that we spoke to were very positive about the engagement that they have received from DLUHC statisticians.

While broader user engagement since the introduction of H-CLIC has been more limited, partly due to the pandemic, DLUHC carried out an online event with over 100 users and data suppliers in May 2021. This was advertised through a link on the gov.uk website alongside the statistical release, and promoted to existing contacts. The statisticians sent out a user survey in advance of this to ask for feedback on the statistics and the data collection (discussed further in the next section). However, some users that we spoke to were not aware of the event.

The statisticians told us they include an inbox email address on all homelessness statistics products so that users can reach out to them. The team told us that it evolves the statistics accordingly in response to user interest. The team also maintains a user forum webpage on the Government Statistical Service (GSS) website. The Technical Note states that “a website has been created to keep users up to date with all the latest developments in DLUHC’s homelessness statistics. It contains details of user events and information about planned developments. Further information on the timescales or the consultation of user feedback will also be published on this forum.” However, the user forum page primarily contains information for LA data suppliers, and there is no general information for users on this page, such as the user consultation event that was held in May 2021.

DLUHC statisticians told us that following the May 2021 online user event they are now thinking about more formal user engagements. They have a vision for GSS web forum pages and would like to have section on them about developments for users to directly feedback on. The statisticians told us that the GSS was currently revamping the web pages and reviewing old documents to make the pages and documents more accessible. However, they can start preparing documents to go on to that page once the GSS page is ready.

DLUHC has a departmental user engagement strategy with an annex on engaging with housing and homelessness statistics users, but this has not been updated since 2015. The statisticians told us that they were not aware of plans for updating the DLUHC departmental strategy, though their future engagement activities would link in with any overall departmental approach. The GSS published a new user engagement strategy in early 2021, to support departments in ensuring statistics meet society’s needs, which we recommend DLUHC draws on.

The May 2021 online user event was a positive development, that was well attended. However, this kind of engagement appears to have been limited since the introduction of H-CLIC in 2018, and it’s not clear what DLUHC’s overall user engagement approach for the statistics is. However, broad and transparent engagement will be essential for informing the further development of the statistics to realise their public value; to inform users and manage their expectations; and to balance competing demands.

Developments to enhance analyses and insight

Throughout the assessment it has become clear that while a lot of progress has been made by DLUHC statisticians throughout the redevelopment of the statistics, the full potential of the H-CLIC data has yet to be realised. Case-level data have potential to answer new questions of importance to society in terms of households’ experiences, or flows through the homelessness system. DLUHC told us about a range of planned innovations, some of which are short-term improvements to outputs, while others are longer term developments to maximise value, such as through data linkage. Users also told us about the developments that they would like to see realised in future.

Ahead of DLUHC’s user event in May 2021, the statisticians sent out a user survey asking for users’ opinions on how the statistics might be further improved. Key enhancements identified by the survey included:

• Providing users with options to benchmark homelessness figures for some LAs against others, for example, being able to select more than one LA at a time, or simultaneously selecting different time periods to allow for more-detailed interrogation of the data
• the ability for users to download the raw data and feed it into their own reporting tools
• reducing the elapsed time between data collection and publication
• making overall presentation more visual.

These areas also came up in our discussions with users. Some LA users told us that having the data published in standalone quarters in formatted tables made it tricky to produce time series to help benchmark between LAs. These users cited that they would like a large .csv raw data file (such as the data made available in a zip file alongside the Ministry of Justice’s mortgage and landlord possession statistics) to enable such comparisons to be made between LAs over time using their own software. Some LA users also commented on a perceived long length of time between data submission and the statistics being published as being the reason why they rely more on their own internal reporting data for monitoring purposes and driving key decisions, than the published official statistics. Enabling timely comparative, time-series analyses between different LAs is therefore a key goal that would enhance the value of H-CLIC statistics for LA users in particular.

DLUHC’s response to the user needs survey included a list of existing planned development projects, many of which would help address areas that users identified. However, the plans do not account for all the issues raised and it is not clear what the team plans to do about addressing the users’ other concerns, such as those related to data submission. DLUHC told us that it is piloting a new Application Programming Interface (API) method of extracting data from suppliers this autumn, with the aims of making the process more automated; reducing the time taken between data collection and publication; and reducing the data collection burden experienced by LAs.

In response to user interest in more visual and interactive content, DLUHC has begun introducing interactive data dashboards. These include on topics such as the support needs of those owed HRA prevention or relief duties, and on LA key performance measures and data quality, providing an indication of LAs that are more or less compliant with meeting HRA and the H-CLIC data requirements each quarter.

In terms of enabling benchmarking, the LA key performance measures, and data quality dashboard, presents the national average for England alongside the number of ‘Households assessed as owed a prevention or relief duty (per 000)’ and ‘Households with children in temporary accommodation (per 000)’ in individual LAs. However, there is not yet the capability in the dashboard for LAs to compare themselves with other individual, or groups of, LAs or the averages for their region, which LAs have expressed a clear interest in being able to do.

In the case of one LA, the percentage of prevention and relief duties owed that ended in accommodation being secured, is reported as 119.35%. Where percentages are greater than 100 this would imply that more cases were closed during the quarter than were open. In cases such as this it should be explained clearly how this situation can arise, to avoid any potential confusion. We also found that the links to each dashboard could be more prominently located on their relevant webpages. At present there does not appear to be any mechanism to allow users to provide feedback on the dashboard, or any communication of whether there are plans to further enhance the information provided within the dashboard, or its functionality.

The statistics team told us about several other planned developments including:

• Slimming down the quarterly bulletin from April-June 2021, to focus more on headlines and changes in trends from the last quarter, while extending the narrative, commentary and insight for the annual bulletin, each autumn
• An ambition to have a dashboard drawing on all the current data tables where users can run their own analyses.However, it would be necessary to carefully manage the security of the data, as allowing people to take their own cuts of data would require case-level data to be made available, which could be disclosive. The statistics team had wanted to put out a flows dashboard along these lines to accompany this year’s annual release, but had to restrict it to a pdf pack of the screenshots to protect confidentiality. DLUHC statisticians told us though that they are aiming to have this available for the next annual release
• Facilitating the linking of H-CLIC data over time with information from other government departments as part of the Homelessness Data England project. This will improve the evidence base for homelessness and allow researchers and policy makers to analyse data that provides a multidimensional profile of people experiencing homelessness, in a secure setting
• Timeliness – by programming in R and automating to improve processing times, the team has slimmed production times down to four weeks. The team is hoping the API project will improve this further and free up more time, which it can then use to help LAs still struggling with submitting H-CLIC data.

Some users we spoke to outside of government were not aware of DLUHC’s development plans. They told us that they would welcome further engagement from DLUHC so they can feed in their ideas on the plans. Users have said that it would be good if DLUHC’s development plans were more visible, rather than them hearing about them via word of mouth or after developments are launched.

Some other topics were raised by users during our engagement that users saw as within potential of the H-CLIC data collection. As previously mentioned, there were lots of requests from users to make the data more accessible and interrogable to aid further analysis. Some users are still raising Freedom of Information requests with LAs to get the homelessness data they require, which is an indication that they aren’t getting everything they think is there from the statistics. Other user requests include:

• Users want more cross-tabulations and a further disaggregation of the published data, and preferably access to the microdata (anonymised) if possible, to allow for more-detailed analysis of characteristics
• LAs would like to be able to make comparisons across larger urban areas, or second tier government (for example Greater Manchester)
• Missing LA data means that some users can’t carry out the analyses they would like to or present the data meaningfully in data dashboards. They acknowledge though that data completeness is improving over time, but see some LAs as consistently not reporting data every quarter. They acknowledge that missing LA data are clearly identified in the tables, but LAs are not labelled in the maps within the Technical Note
• H-CLIC originally promised to measure repeat homelessness cases but these estimates have still not been published
• In relation to temporary accommodation, users report that they are now no longer able to tell how long households have been in temporary accommodation, and they are unclear why these statistics are no longer available. They would also like to know the number of individuals in temporary accommodation, not just households
• Some users identify that not as much detail is available on the characteristics of households in the prevention and relief stages, compared to the main duty stage including: the number of children in the household; the number of age 16-24 presentations; and the characteristics of those ineligible for support or with no recourse to public funds, discounted at the eligibility stage
• Users would like to have more-detailed data to understand outcomes through the system by support needs and characteristics like ethnicity or disability
• Users also want more robust data in trans-homelessness, including more detail on gender identity categories (trans-male, trans-female, and non-binary)
• User would also like more information about the characteristics of households that fall out of the system at or between each duty stage ­– related to repeat homelessness
• Users would also like to see a further disaggregation of rough sleepers in H-CLIC by different support needs, and whether they slept rough in the last year, and not just at the homelessness application stage.

The statisticians told us that ineligible people fall out of scope of H-CLIC and so their characteristics are not recorded. In relation to measures of repeat homelessness, they said that they had been looking at what they could identify in H-CLIC to see the same people going through the system. They identified some possible issues with particular LAs not recording previous case data properly due to poor IT systems, but they were working these through. These was also a time lag involved when they are able to report complete flows, as some cases can take up to two years to go through the system.

The statisticians told us that meeting some user requests depended on other initiatives taking place first and that they were thinking about how best to communicate and update users on where they currently are with their planned developments. For things that users would like to see that the statisticians cannot do, or were out of scope, the team would consider adding more content explicitly on the limitations of the data. There is considerable potential for DLUHC to be more open about its future development plans and seek users’ views in a transparent way to inform developments that add value, and to seek feedback on new products.

Enabling data linkage while limiting additional burden

H-CLIC is a powerful case-level dataset that has the potential to answer a range of questions, particularly of different households’ experiences or flow through the homelessness system, and in understanding the nature and extent of repeat homelessness. While the data are available down to LA level, user feedback has highlighted a need for more-granular data. Users have also expressed a desire for the microdata and anonymised case level data to be made available for research purposes. There is a lot of potential to link H-CLIC data with other administrative sources to answer key questions, in greater granularity. But there is a need to balance increased insight with demands on LAs and applicants, particularly for LAs reporting that H-CLIC has substantially increased burden and those that have so far been unable to submit H-CLIC returns.

LAs reported that meeting the H-CLIC requirement increasingly takes over the focus of their engagement with homeless people. A two-hour assessment is required to collect all the data required, which is much more burdensome on homeless applicants and officers than the previous P1E collection. Many LAs have adapted but some are still struggling to get on top of the data requirement (discussed further in chapter 2).

The evaluation of the implementation of the Homelessness Reduction Act (March 2020) highlighted that LAs found transitioning to the new H-CLIC reporting requirements was a difficult process, especially when it was first introduced. LAs noted it was a significant challenge to meet the new data requirement and that with hindsight a different and later timeframe (to that of HRA) for the introduction of H-CLIC would have been better. The LAs in the case study all said that their existing IT systems were not fully equipped to meet the new data requirements and they had either modified, or in most cases purchased new systems – none described this as straightforward. DLUHC told us that the new API system being developed will be the key to reducing burden on LAs and to enable LAs to realise the benefits of being able to see more of the data themselves (for example, through a new quality dashboard).

Despite these challenges for some LAs, a range of ambitious linkage projects is planned through DLUHC working in collaboration with ADR UK and ONS on the Homelessness Data England project. Case-level data will sit on a research level platform where users can request access. The team told us that this was still a while away from becoming a reality as it needed data from other government departments too, but there would be a dataset that researchers can access securely through ADR UK.

DLUHC is currently looking to get data sharing agreements signed with participating LAs to allow them to share the personal data associated with cases submitted on H-CLIC. This will mean that DLUHC can create an ID spine to link H-CLIC data with administrative health and benefits data – this is currently being piloted with a small group of LAs. DLUHC told us it has separately held stakeholder events with LAs to explain its plans and continue to engage on a regular basis. The medium-term aim is to collect this personal data using the API so would require no input from LAs at all. It is not clear how DLUHC plans to enable data linkage for LAs not yet returning data via HCLIC.