Measuring the Return on Our Work with Crisis

In line with our commitment to integrity, this model follows the principles set out in HM Treasury's Green Book (appraisal) and Magenta Book (evaluation), using conservative UK public expenditure data drawn from published sector research.

Helium Capital's partnership with Crisis supports the Housing Access Fund. Since 2019, this Fund has supported people out of homelessness and into safe, secure, and appropriate housing through services and support at Crisis' London Skylight.

1. The Core Economic Principle

The framework quantifies the direct fiscal pressure removed from UK public services when someone has their homelessness ended. Rather than applying broader social value multipliers, it measures hard fiscal cost-remediation: money saved by the public purse across healthcare, criminal justice, and local government statutory services.

2. The Evidence Base

Baseline assumptions are drawn from published UK research on the cost of homelessness, principally Crisis' own cost-of-homelessness work and analysis by the University of York's Centre for Housing Policy.

Crisis' research finds that people suffering from homelessness for three months or more cost, on average, around £4,300 a year to the NHS, £2,100 to mental health services, and £12,000 through contact with the criminal justice system. This results in a combined cost of roughly £18,400 per person. Separate estimates of the total annual cost of rough sleeping, including temporary accommodation, range from £20,000 to £28,000.

Against this evidence, we use a baseline of £15,000 gross saving per person per year — below the combined figure Crisis identifies, and set deliberately low to maintain a conservative measurement.

3. Adjusting for Caution

ot everyone we support would have remained homeless without us, and we don't work alone. The Green Book requires these effects — deadweight and attribution — to be estimated against the best available comparator evidence, not assumed.

Our closest UK benchmark is the government's national evaluation of the Housing First Pilots, delivered by ICF Consulting and published via GOV.UK in October 2024. It found 84 per cent tenancy sustainment after three years and estimated long-term public savings of £15,880 per person per year, arrived at independently of our own baseline. Earlier work by Bretherton & Pleace (2015) and Bramley et al. (2016) points the same way. Neither is a randomised controlled trial; both rely on before-and-after comparisons, a limitation shared by most UK evidence in this area.

For genuinely peer-reviewed evidence, we look to Canada's At Home/Chez Soi trial, published in the academic literature and built on a randomised design. It supports the same broad mechanism internationally: that stable housing reduces downstream public service costs. It's a non-UK study, so we use it to test whether our approach is plausible rather than to size our own figures.

We apply two downward adjustments to the £15,000 baseline:

• Deadweight (15%): the proportion of the cohort the wider evaluation literature suggests might exit homelessness through alternative routes without our specific involvement.

• Attribution (20%): the contribution of other statutory actors, such as community mental health teams, social care, and local authority housing options services, typically involved alongside our support.

£15,000 × 0.85 × 0.80 ≈ £10,200 net benefit per person, per year.

Both rates are reasoned assumptions, informed by the closest available evidence rather than a controlled study of our own cohort.

4. What We Don't Claim

We count savings for one year only, not compounded over a person's lifetime, even where housing is sustained for longer. This likely understates lifetime impact.

For our earliest years, outcome data wasn't systematically recorded. Rather than use an external industry average, we took our lowest verified annual transition rate from later, better-recorded years and applied it as a cautious proxy for the unrecorded cohorts, rounding down.

The £10,200 figure is a current-price estimate, built from today's evidence, and applied consistently to every person supported since 2019 rather than adjusted year by year. This assumes the real cost of supporting someone out of homelessness — after stripping out general price inflation, which the GDP deflator would otherwise handle — has stayed broadly stable over that period. We haven't tested this assumption directly. It's also not obviously safe: Crisis' Homelessness Monitor series and a 2025 cohort study of homeless hospital access in London both point to rising health need among people experiencing homelessness over roughly this period, including self-reported mental health diagnoses climbing from 45 per cent in 2014 to 82 per cent in 2022. If real per-person costs have in fact risen, applying today's £10,200 uniformly to earlier years likely overstates the real benefit of those earlier transitions, and understates it for the most recent ones. We flag this as an open assumption rather than a settled one, and would revisit it if better evidence on the trend becomes available.

5. The Result

Since 2019, 462 people have had their homelessness sustainably ended through the Housing Access Fund, generating an estimated £4.7 million in public service savings at £10,200 net benefit per person.

6. Limitations

This is a reasoned estimate, not an audited financial statement. The £15,000 baseline and the deadweight and attribution rates are judgement calls, benchmarked against the best available evidence rather than derived from a controlled study of our own cohort. A formal comparison-group evaluation, of the kind described in the Magenta Book, is the natural next step as our tracking data matures.