Greater London Authority: Homelessness


The Department for Communities and Local Government (DCLG) and the Greater London Authority (GLA) developed a social impact bond (SIB) to try to improve outcomes for a group of 831 persistent rough sleepers in London. The SIB sought to help some of the most vulnerable people in society, who often have a life expectancy in the region of 40-44 years.


Overview of the issue

Over half of all rough sleepers in the UK are based in London. The capital is unique in the UK, and possibly the world, however, in having a comprehensive database (CHAIN) which records when individuals are seen sleeping rough on the streets and when they enter and leave hostel accommodation, along with data on support needs, nationality, and age. Evidence from CHAIN shows that about 1/6 of rough sleepers account for almost half of all recorded rough sleeping and suggests that despite a range of existing services for London rough sleepers, provided by the London boroughs, the GLA, and the VCSE sector, outcomes for this group of vulnerable individuals have remained consistently poor over a number of years. Their support needs often relate to drug and alcohol abuse, and mental health illnesses. Development of the SIB was supported by the government’s Ministerial Working Group on Homelessness.


What is the intervention?

The key early objective is to get people off the streets as early as possible, with incentives to then help them move into settled accommodation, and employment where possible, as well as to reduce chaotic use of the NHS (measured by number of A&E visits).

Although not measured directly as outcomes, it was seen as necessary to help cohort members to address sometimes longstanding drug, alcohol and mental health issues.

There are a range of existing services that cohort members already had access to such as rough sleeper outreach teams, hostel provision, day centres, health services and potentially access to employment services such as the Work Programme. The intention was not to replace these services, but to work with cohort members and service providers to ensure that better use is made of what exists, as well as filling in service gaps where necessary. It was anticipated that in many cases, a “consistent trusted adult” who will be able to provide support over an extended period, would be important in achieving these goals, but providers have had the freedom to innovate and to do what works best for each individual case.

Whilst the GLA has set out what outcomes it wishes to be achieved, it has left the nature of the intervention to the discretion of the contract holders.


How the intervention is funded

The cohort of 831 rough sleepers has been split roughly in half and the GLA held two separate contracts with two service providers: Thames Reach and St Mungo’s.  The GLA held a contract directly with one provider (Thames Reach) while the contract for the second half of the cohort is with an SPV which includes the provider (St Mungo’s) as well as intermediary Triodos and investors.

In both cases, the service was funded by a combination of investment from social investors and from the providers themselves. As such, the providers shared in the risk but also stood to make a financial return if the project is successful. Outcomes were paid quarterly and payments could be reinvested, so that only a proportion of the service costs needed to be funded by investment if expected outcomes were achieved.

A number of investors, including CAF Venturesome, the Orp Foundation, Big Issue Invest and several smaller charitable foundations and private individuals provided capital to support the project.



 Outcome payments were made by the Greater London Authority and funded by DCLG. Payment by DCLG was capped at £5M including administration and development costs.

 The specific outcomes measured were as follows:

  •  Reduction in rough sleeping

The reduction in the number of the cohort seen sleeping out in each quarter is compared to baseline expectations. This is measured using the CHAIN database with a baseline of expected levels of rough sleeping each quarter based on historic virtual cohorts who would have met the SIB criteria in the past.

In order to provide a smoothed baseline profile, the proportion of individuals seen rough sleeping each quarter were averaged across each year e.g. adjusted historic performance for the first four quarters after cohort generation had been averaged to produce an average figure, for use quarterly. Providers’ achievement in rough sleeping was measured against this average yearly baseline.

  • Moves to settled accommodation in the UK

Small payments were made for moves into settled accommodation outside the hostel system, with larger payments made for sustainment of accommodation for 12 months and 18 months. The definition of settled accommodation was flexible, and could include a social housing tenancy, a private rented tenancy or moving back in with family and friends anywhere in the UK. A return to rough sleeping or leaving the accommodation in an unplanned way meant that the 12 month and 18 month payments were not made, though moves form one form of settled accommodation to another are not penalised.

Evidence for this outcome includes tenancy agreements, letters from landlords etc.

  • Reconnections to accommodation abroad

Small payments were made for evidenced moves to an address in another country (likely to be the country of origin), with larger payments being made after 6 months if there had been no return to a rough sleeping lifestyle in London.

  • Increased employment

Small payments were made for a sustained period of volunteering or achievement of an NVQ level 2 qualification.

Larger payments were made for part time employment of 13 weeks and 26 weeks and even larger payments were made for full time employment of 13 weeks and 26 weeks.

Evidence of employment was required such as payslips, letter from employer, and P45 information. Periods of employment could be built up over time and do not necessarily need to be continuous.

  • Reduction in use of Accident and Emergency services

Payments were made for annual reductions in the use of A&E services against a baseline for each cohort drawn from the NHS Hospital Episode Statistics. Each cohort had a proportion of individuals who are very frequent users of A&E, and stabilising their NHS usage will be important to do well against this outcome.

Payments for each metric were sense checked with potential providers during the competitive dialogue process, and maximum tariffs for each outcome measure were then set at the Invitation to Tender stage. Providers were asked to bid to provide these outcomes at a discount from the maximum tariff. Levels of ambition were also assessed to ensure that price per outcome was balanced against working to achieve positive outcomes for the maximum number of cohort members.


Impact measurement and counterfactual

 The measurement and payment model has been designed to encourage providers to work with every cohort member to achieve the best possible outcomes for them over the three year contract period. Specifically, the aim was to ensure a financial incentive to work with people with complex needs as well as an incentive to help those with lower needs progress into employment wherever possible.

Baselines have been set for expected reduction in rough sleeping and A&E visits based on virtual historical cohorts who would have met the same criteria in the past, and whose outcomes have been tracked. Accommodation, Reconnection and Employment outcomes do not have a baseline, as these would have been a) very low and b) difficult to measure.

Because of the different services involved in the scheme, and who will be seeing reductions in costs, the measurement of the outcomes is also taking place across different sectors. Rough Sleeping outcomes were measured by CHAIN; A&E outcomes by NHS Hospital Episode Statistics; other outcomes were evidenced by providers for each cohort member and audited by GLA. An evaluation of the Homelessness SIB has been commissioned by DCLG, supported by the Cabinet Office.


Savings to the commissioner

 Funding is from DCLG, but there is potential for other government departments to realise savings if the SIB intervention is successful, in particular from reductions in crime, reduction in chaotic use of health services, movement to settled accommodation from the hostel system, and decreased benefit claims caused by higher employment. There is also a group of foreign national rough sleepers, who have not established a life in the UK away from the streets, and whose needs are best served by returning to accommodation in their home country.

Social Finance estimated average costs to the public purse of the cohort at £37,000 per person Net Present Value over 5 years. For 831 cohort members this equates to over £30M over a 5 year period. In contrast, the maximum cost of the SIB is £5M if optimum outcomes are achieved.

There are, of course, other costs associated with helping persistent rough sleepers beyond the SIB itself. These include accommodation costs where cohort members have previously been sleeping on the streets or squatting, outreach service costs, costs of drug and alcohol treatment, and costs of properly treating persistent health problems. There are important questions to be answered about how, if at all, these costs should be counted in assessing net savings to the public purse.



The SIB focused on 831 persistent rough sleepers split into two cohorts with a separate contract for each. The two groups were geographically split across London, but were as equal as possible in terms of previous frequency of rough sleeping, nationality, and support needs. These rough sleepers were recorded on the CHAIN database if they have been seen sleeping rough in the previous three months or living in a rough sleeping hostel and seen bedded down on the streets at least 6 times over the previous two years. Although the cohort was less than 1/6 of the total number of people seen sleeping rough in London over the previous year, they accounted for 47% of all bedded down street contacts over the year because of their relatively frequent rough sleeping. Rough sleepers who are being worked with through the existing “205s” programme were excluded.



Key innovations

  •  Development of a basket of outcomes indicators designed to encourage work with the whole cohort over the three year period.
  • Payment for outcomes each quarter, reducing the amount of investment needed, and the length of time investors cash is tied up, thus reducing the level of return necessary to attract investment.
  • “Dragons Den” investment event designed to introduce likely bidders to potential social investors.
  • Competitive Dialogue Procurement which is normally used for complex infrastructure projects. This allowed the commissioners to facilitate the introductions between social investors and shortlisted providers (i.e. the Dragon’s Den event), discuss the proposed tariffs for the outcome metrics as well as explore and comment on service solutions proposed by the short listed providers.
  • Evaluation of commercial aspects of the tenders by scoring the level of discount given against the tariffs for the outcomes as well as scoring the level of ambition (how many outcomes will be achieved).
  • Providers provided some of the upfront funding themselves and so shared the risk with investors, as well as potential upside.