Mapping Britain’s Service Community Financial Safety Net

What occupational returns charitable funding and community finance reveal about the case for interest free assistance

Britain does not lack organisations that support people who serve. It has trade unions and professional associations, regimental and occupational charities, state compensation schemes, credit unions, salary advances and commercial credit. The harder question is whether these institutions, taken together, provide the right form of help at the moment an essential expense cannot wait.

This study maps that system using 143 annual-return records from twelve occupational organisations, 381 Armed Forces Covenant Fund awards worth £60.25 million, a relevance screen of 1,474 Scottish charities, a 127-organisation partner map and selected official statistics and product terms. The resulting picture is substantial but incomplete. Existing provision performs many valuable functions, yet much of it is designed for legal representation, long-form casework, compensation, grants, savings or interest-bearing credit. None of those categories is identical to rapid, interest-free, repayable assistance.

That difference creates a plausible institutional space for InnerCircle, a proposed community-governed model built around regular contributions and restricted community funds. The evidence does not prove the size of demand for such a product, still less that a new organisation can deliver it safely. It does, however, define a testable proposition and shows what a credible pilot would need to measure.

Findings at a glance

Six conclusions survive the data audit.

1. Reported occupational benefits vary widely in intensity and purpose. Across nine organisations with a usable latest native-text AR21 return, reported benefits ranged from about £6 per member at UNISON to £155 at the Prison Governors Association. These figures cannot be read as household hardship payments. The underlying lines often cover legal defence, education and organisational services.

2. The Fire Brigades Union series records a marked increase in benefit expenditure without comparable membership growth. Membership remained close to 33,000 from 2019 to 2025, while reported benefits rose from £0.43 million to £1.88 million. The return does not identify how much reached members as cash, so this is evidence of rising support cost, not proof of rising financial hardship.

3. Armed forces charity funding is the strongest quantified part of the present evidence base. The parsed Covenant Fund data contain 381 awards worth £60.25 million. Within the complete 2023–2025 segment, annual value rose from £13.2 million to £26.6 million and the number of recorded awards from 49 to 222, while the median award fell from £100,000 to £60,000. These are calendar-year records in the parsed workbooks, not the Trust’s annual-accounts expenditure.

4. The grant data describe services delivered through organisations, not cash delivered directly to households. Health, housing, family and work-related themes dominate a model-assisted reading of the award descriptions. That classification is exploratory because the supplied manual-review fields were blank.

5. The surrounding system contains real alternatives, but each carries its own boundary. Credit unions offer community-based savings and affordable lending but charge interest and apply affordability criteria. Salary advances can be fast but depend on employer participation and bring the next payday forward rather than creating new income. Benevolent grants avoid repayment but require eligibility and casework. State schemes serve statutory purposes and can involve evidential and administrative processes.

6. The research establishes a plausible gap, not a market size. The 127 organisations in the partner map are prospects, not customers. Membership is populated for only ten. The 1,474-charity Scottish screen is a keyword map, not a measure of capacity. Any claim that millions of people are addressable would exceed the evidence.

The practical implication is narrow. InnerCircle should be judged first as a pilotable hypothesis: can a community-governed restricted fund deliver essential assistance more quickly and at lower household cost, while maintaining fair access, repayment discipline, sufficient liquidity and regulatory compliance?

The question behind the research

Financial shocks are defined by timing as much as by size. A relocation deposit, funeral cost, broken boiler, travel required after a family emergency or temporary loss of income may be manageable over a year and impossible this week. A household can have stable employment and still lack the liquid cash needed between the event and the next salary, grant decision, benefit payment or insurance settlement.

Service communities matter in this context for two reasons. First, their members can face distinctive events: postings, discharge, service injury, occupational trauma, irregular shifts, unsocial travel, professional disputes and transitions between institutional systems. Second, these communities already possess forms of social infrastructure. Membership organisations, messes, stations, branches, trusts, professional colleges and occupational charities create identity and routes of referral. The same common bond that supports representation or benevolence may also support a financial mechanism.

The research question is therefore not whether Britain provides support. It plainly does. The question is whether a rapid, interest-free and repayable instrument could complement the existing system when a grant is inappropriate, statutory support is not yet available and interest-bearing credit is an expensive or unsuitable response.

The word complement carries weight here. A repayable community fund cannot replace compensation for service injury, specialist debt advice, a benevolent grant for someone unable to repay, legal representation or safeguarding. Its potential value lies between those functions.

What the evidence covers

The study uses four main evidence families.

The first is the Certification Officer’s AR21 annual-return regime. Trade unions submit information on membership, finances, officers and governance.[^1] The research corpus contains 143 extracted returns across twelve organisations from 2012 to 2025. Native-text records from 2019 onwards form the main analytical window because older optical-character-recognition rows contain digit errors and failed plausibility checks. Nine organisations have both membership and benefit values in their latest usable native-text return.

The second family is charity and grant data. A screen of the Scottish Charity Register identifies 1,474 records whose names or purposes contain terms associated with service communities, benevolence, hardship or related needs. OSCR makes the full register available as open data and updates it daily.[^2] The screen used here is a frozen September 2026 extract. It is deliberately Scotland-only because the equivalent England and Wales bulk register was not present in the research delivery.

The grant analysis uses 381 Armed Forces Covenant Fund records parsed from 360Giving workbooks. The Trust publishes annual reports and accounts, while its 2026 mid-cycle review describes £14 million committed to 146 projects between April 2024 and September 2025.[^3] Those official reporting periods differ from the calendar-year grouping used in this study. The figures are therefore related views of the same institution, not values that should be forced to reconcile line by line.

The third family is official context: War Pension Scheme and Armed Forces Compensation Scheme statistics, selected insolvency data and workforce signals. The Ministry of Defence describes the War Pension Scheme series as covering claims, awards and recipients where disablement or death resulted from service before 6 April 2005.[^4] Its AFCS collection covers claims and awards under the post-2005 scheme.[^5] These sources show administrative scale, but claim counts do not measure unmet household liquidity.

The fourth family maps substitutes and partners. It includes occupational credit unions, benevolent funds, earned-wage access and no-interest lending experiments, together with 127 organisations that might inform or host future outreach. Product terms were checked against current provider pages in September 2026 wherever they enter the analysis.

The evidence hierarchy

The study gives primary regulatory and official sources the greatest weight. Derived CSVs and reproducible calculations support the figures. Narrative notes guide the investigation but do not prove a claim. InnerCircle’s own product and legal documents establish what the project proposes; they are not external validation of need or feasibility.

This distinction matters because the original inception pack contained two material inconsistencies. It stated that a 40-award manual sample achieved approximately 95 per cent classification accuracy, yet the delivered QA file contains 67 sampled rows and no completed verdicts. It also described Covenant Fund median awards of £190,000 in 2023 and £80,000 in 2025, while the supplied build script reproduces £100,000 and £60,000. This publication uses the reproducible values and treats automated thematic classifications as exploratory.

A system of specialised institutions

Britain’s financial safety net for service communities is not a single market. It is a set of institutions created for different purposes.

Trade unions and professional associations organise representation, legal defence, education and professional support. Their annual returns expose financial scale but rarely isolate the household circumstances behind each payment. Benevolent charities can make grants and arrange casework, often with deep knowledge of a profession or regiment. Their strength is discretion; their boundary is eligibility and the time needed to establish a case. State schemes can recognise an entitlement arising from service, employment or disability, but they operate through statutory definitions and administrative evidence. Credit unions and commercial products can deliver money quickly, but repayment cost, affordability assessment and membership rules remain part of the transaction.

These are not defects. They are consequences of institutional purpose. A charity should not lend indiscriminately simply because a household needs speed. A regulated lender cannot ignore affordability. A compensation scheme must establish causation and entitlement. A union must fund representation as well as welfare. Fragmentation appears when a household’s problem crosses these boundaries.

The most useful way to compare the instruments is through five questions.

• Who qualifies? Occupational membership, charitable objects, statutory tests, employer participation and credit-union common bonds define different populations.
• What can the money pay for? A benefit may fund legal representation, a grant may be restricted to essential goods, and an advance can only bring forward earned salary.
• How quickly can a decision be made? Digital credit may decide quickly; a complex grant or compensation claim can require evidence and casework.
• What does the household pay? Grants and compensation are non-repayable. Credit carries interest. Earned-wage access can carry a transaction fee. A no-interest loan still imposes principal repayments on future cash flow.
• Who absorbs loss? The state, donor, insurer, lender, charity or participating community bears different forms of risk.

InnerCircle’s proposed combination of restricted community funds and interest-free repayment tries to alter the last two questions. Removing interest lowers direct household cost. Keeping repayment recycles capital. That design does not remove loss; it transfers responsibility for underwriting, arrears, liquidity and fairness to the fund’s governance.

What occupational returns show

The AR21 analysis compares the latest usable return for nine organisations. Scale and intensity are separate dimensions. UNISON reported £9.18 million of benefits for 1.47 million members in 2025, or £6.27 per member. The Prison Governors Association reported £167,711 for 1,079 members in 2024, or £155.43 per member. The Royal College of Nursing reported £35.71 million for 589,717 members in 2024, or £60.55 per member. The Fire Brigades Union reported £1.88 million for 33,589 members in 2025, or £56.02 per member.

The range is analytically interesting but easy to misuse. A high per-member figure does not necessarily indicate generous hardship payments. Smaller occupational bodies may purchase intensive legal or professional protection for a narrow membership. A large general union can spread central services over more than a million members. Accounting categories, organisational structures and exceptional cases also differ.

The RCN series demonstrates the problem. Its extracted benefit figure falls from £131.5 million in 2020 to £29.0 million in 2021, then ranges between £34.8 million and £43.1 million before reaching £35.7 million in 2024. That discontinuity may reflect classification or reporting changes as well as underlying activity. The latest row was already marked for review in the source corpus. It remains useful as a reported value, but it should not carry a causal story.

Membership trends are clearer. Between 2019 and the latest available point, the RCN series grows from 469,237 to 589,717 members, an index rise from 100 to approximately 126. UNISON’s comparable series moves from 1.38 million to 1.47 million. The FBU remains broadly stable.

The FBU case

The FBU provides the cleanest example of a changing cost profile. Its membership remained between 32,501 and 33,939 across the seven annual points from 2019 to 2025. Reported benefits rose from £427,739 to £1,881,593. On a per-member basis, the series moved from £13.10 to £56.02.

There are several possible explanations: more cases, more expensive cases, wider benefit definitions, legal costs, inflation or changes in accounting presentation. The return series alone cannot select among them. The defensible conclusion is that reported support expenditure intensified while membership did not. The finding justifies further investigation; it does not prove a rising incidence of household distress.

What benefits contain

A 160-line sample, covering £139.2 million of reported benefit items, was coded into thematic categories. Fractional allocation was used when a line carried more than one code. Legal services form the largest classified share at £42.9 million, followed by work-related items at £21.0 million. Another £44.9 million, or 32.3 per cent of the sample value, remains unclassified.

This is enough to reject a common shortcut: the AR21 field labelled “benefits to members” is not a proxy for hardship cash. It is not enough to calculate the cash-hardship share reliably. A publishable estimate would require a documented codebook, completed manual review, treatment of ambiguous labels and reconciliation back to the source return for each organisation.

The distinction changes how the support gap should be discussed. Large union-benefit totals demonstrate institutional spending. They do not show that a member can obtain £1,500 for an urgent household expense.

Charitable funding is substantial and mediated

The Armed Forces Covenant Fund is the strongest quantitative grant source in the corpus. The parsed data contain 381 awards worth £60.25 million across six calendar years, although 2020, 2022 and 2026 are represented only partially. The comparable complete segment is 2023–2025.

Across those years, recorded annual value rises from £13.2 million to £26.6 million. Award count rises more quickly, from 49 to 222, while the median falls from £100,000 to £60,000. A lower median alongside a higher total is consistent with a portfolio containing more, smaller awards. It does not by itself reveal whether support became more accessible to individual households.

The thematic profile helps explain what the grants support. A model-assisted classification gives health the largest fractionally allocated value, followed by housing, family and work. Health appears in 221 raw tags and family in 128. These themes reflect programme descriptions rather than verified outcomes.

The strongest interpretation is institutional. Covenant Fund awards usually finance organisations and projects: advice pathways, wellbeing programmes, housing support, research, family services and capacity. A household may benefit significantly, but the route is mediated. The grant is not ordinarily a revolving emergency facility placed directly in that household’s hands.

The portfolio is also concentrated. The top five recipient identifiers account for 23.4 per cent of recorded value and the top ten for 33.4 per cent. Concentration is not inherently undesirable. Large or repeated recipients may possess specialist capacity, national reach or the ability to manage multi-year programmes. It does mean that aggregate funding cannot be interpreted as uniformly distributed provision.

What Scotland adds

The Scottish Charity Register offers a mappable subsystem. The keyword screen identifies 1,474 charities, of which 1,374 were active in the frozen extract. The largest overlapping buckets are hardship with 737 matches, health and NHS with 558, armed forces and veterans with 200, and benevolence with 163. Police, fire, ambulance and prison terms produce smaller groups.

These numbers show organisational density, not spare capacity. A charity can appear in several buckets. A reference to hardship in its purposes does not prove that it currently makes grants, accepts new cases, serves a particular local authority or holds enough unrestricted funds to respond quickly. The dataset lacks a consistent income field for this analysis. Keyword score measures textual relevance, not institutional strength.

Scotland could nevertheless be useful for a pilot because the ecosystem can be mapped against a single register and a defined regulatory geography. That makes referrals and exclusions easier to study. It does not make Scotland representative of Britain.

State support and household pressure

Official statistics establish that formal schemes operate at scale. The research extract records 13,780 War Pension Scheme claims in 2024/25 and 90,743 people in cumulative AFCS injury and survivors’ claims since 2005/06. These values belong to different periods and populations. They cannot be added, compared as annual flows or translated into cash available during an emergency.

The same discipline applies to insolvency. The supplied state table includes a seasonally adjusted monthly figure for August 2026 and a 2026 calendar-year row that is necessarily year-to-date. Because the latter is labelled as though it were a completed annual total, this publication excludes it. A public article should prefer a stable annual series or label partial periods explicitly.

The broader national context is clearer. The Financial Conduct Authority’s 2024 Financial Lives survey found that 8 per cent of UK adults, about 4.5 million people, were in financial difficulty under its definition; the rate was higher among renters, lower-income households and younger adults.[^6] In a January 2024 recontact survey, 14 per cent felt heavily burdened by bills and credit commitments, 11 per cent had missed a bill in the previous six months and 28 per cent were not coping or found it difficult to cope financially.[^7]

Those figures are not service-community estimates. They describe the national environment in which service households live. Occupational identity may change exposure, eligibility and referral routes, but it does not remove ordinary housing, energy, childcare, debt and family pressures.

Serve and Protect Credit Union’s 2025 survey adds a relevant, though non-random, occupational perspective. It drew responses from 5,599 members across the armed forces, police, prisons, health and fire services and reported that almost a quarter had previously been denied credit.[^8] As a member survey, it cannot estimate prevalence across the workforces. It does show that exclusion from mainstream borrowing exists inside the same common-bond populations that occupational finance serves.

The alternatives are real

Any case for a new instrument must acknowledge the products already available.

Credit unions

Credit unions are member-owned financial co-operatives. Membership depends on a common bond such as occupation, employer, locality or association. MoneyHelper notes that members may need to save or wait before borrowing, although terms vary, and that credit-union loans generally cost less than high-cost short-term credit.[^9] In England, Scotland and Wales the statutory cap is 3 per cent per month, equivalent to 42.6 per cent APR, though many products are priced well below it.

Serve and Protect Credit Union serves police, prison, probation, military, fire and health communities and reports more than 55,000 members. Its current material describes payroll-based savings and loans, with rates starting from 8 per cent APR in one 2026 employer launch.[^10] No1 CopperPot serves the police family. It assesses affordability, does not rely on credit score alone and aims to decide applications within three working days.[^11]

These providers are close to InnerCircle’s intended audience and already possess regulated lending capability. They are therefore potential comparators and potential partners, not merely competitors. Their boundary is that credit remains interest-bearing and subject to lending criteria.

Earned wage access

Salary Finance Advance gives participating employees access to a portion of pay already earned. Its public page describes a typical threshold of up to 50 per cent of earned income, up to three withdrawals per pay cycle, with a stated standard fee of £1.69 per withdrawal subject to employer terms.[^12] The product can solve timing inside a pay period. It cannot cover someone without a participating employer, create capacity beyond accrued pay or prevent the following payday from being smaller.

No-interest lending

Fair4All Finance’s No Interest Loan Scheme pilot is the closest public-policy analogue. Its 2024 interim report covered nearly 4,000 loans worth £2.1 million, with an average value of £591, and reported that 80 per cent appeared to reach people in financially vulnerable circumstances.[^13] The official page said a final evaluation would follow completion in 2025. No final evaluation was located in the sources reviewed for this publication, so the article does not claim that the model has been adopted, discontinued or proven sustainable.

The pilot matters because it demonstrates that no-interest lending can be delivered at meaningful volume through community-finance institutions. It also underlines the funding question. Forgone interest does not remove origination, servicing, arrears or capital costs. Someone must pay them.

Grants and advice

Benevolent funds can be the right answer when repayment would worsen hardship. Debt advice can be more valuable than another liability. Statutory support can recognise a right that should not depend on mutual contribution. A sound referral system must preserve those options.

InnerCircle’s proposed product should therefore contain a clear “not a loan case” route. Applicants facing unmanageable debt, coercion, fraud, safeguarding concerns or an expense that cannot reasonably be repaid need specialist advice or grant support. Approval volume is not a success measure if the instrument is unsuitable.

The case for a missing middle

The research supports three different statements, and they should not be collapsed into one.

An identified gap is visible where the evidence directly shows a mismatch. AR21 benefit totals cannot be treated as emergency cash. Covenant awards primarily finance organisations and projects. Salary advance depends on accrued pay and employer access. Credit-union loans charge interest. Each instrument differs from interest-free repayable assistance.

A plausible gap follows from timing and eligibility. A working household may need money before a charity completes casework, may not qualify for a grant, may wish to avoid interest and may still be able to repay principal over a year. The national financial-resilience data and occupational credit evidence make that scenario credible.

An unmeasured gap is the number of eligible people who would apply, receive a suitable offer and repay it. The present corpus does not estimate that population. It also does not establish the distribution of requested amounts, urgency, default risk, repeat demand or willingness to contribute £15 per month.

This three-part distinction protects the project from a familiar error in early-stage research. Mapping an institutional difference does not calculate demand. Listing organisations does not calculate distribution. Adding known memberships does not calculate an addressable market.

The partner map illustrates the point. It contains 127 organisations and marks 32 as priority one. Only ten have a populated membership field. Seventeen of the priority-one organisations are in the veterans sector, five in police, four in fire and three in health. The weighting reflects evidence coverage and outreach logic; it does not rank need.

The strongest current case for initial fieldwork lies in the armed forces and veterans ecosystem. It has the deepest grant data, defined statutory schemes, mature charities and occupational credit-union coverage. That strength is methodological rather than moral: it means a pilot can distinguish the proposed product from existing provision more clearly.

InnerCircle as a proposed institution

InnerCircle is pre-launch. It has no operating loan book, proven repayment record or confirmed partnership network. Its legal and governance materials remain under development. Any public description should preserve that status.

The current product concept has four elements.

First, a participant contributes £15 per month. The proposal allocates £10.50 to the relevant community fund and £4.50 to operations. The split makes operating cost explicit, but it also creates a demanding value test: participants must accept that only 70 per cent of their contribution enters the restricted fund.

Second, eligibility begins after three cleared contributions. The proposed products are £1,500 over twelve months for emergencies and £3,000 over eighteen months for bereavement and relocation. Both are described as interest-free and fee-free.

Third, applications use two or three supporter confirmations depending on the amount. Supporters confirm community standing and circumstances without becoming guarantors. This may use social knowledge without transferring legal liability, but it raises questions about privacy, exclusion, power and access for people with weaker networks.

Fourth, each community’s money is intended to remain in a separately governed fund. That structure can make allocation legible to participants. It also fragments liquidity. A small community could face several claims at once, while another fund holds unused cash. Cross-fund support, reserves and rationing rules would need explicit governance.

Where the model may add value

The proposal differs from a conventional loan in household price, from a grant in capital recycling and from salary advance in amount and repayment period. It differs from a general no-interest scheme through the common bond and community governance.

Those differences could improve trust, referral and repayment behaviour. They could also produce adverse selection, peer pressure or a false sense that social confirmation substitutes for affordability. The project should treat these as empirical questions.

The strongest proposition is operational: a defined community may be able to pool regular contributions, triage an urgent request quickly, lend without interest and recycle principal as repayments return. The weakest proposition would be that community identity alone solves underwriting. It does not.

What a credible pilot must establish

A pilot should start with a narrow population, a capped fund, a fixed product and pre-agreed decision rules. Its purpose is to test the missing-middle hypothesis, not to demonstrate national scale.

Demand

The application form should capture the expense, amount, urgency, alternatives already tried and the consequence of delay. Reporting should distinguish enquiries, completed applications and approved cases. A high enquiry count with few suitable applications may indicate interest without product fit.

The pilot should record why applicants do not use an existing grant, benefit, insurer, employer programme or credit union. This creates a direct test of complementarity. It also exposes cases where referral is more appropriate than lending.

Access

Measure time from first contact to a complete application, decision and payment. Record where evidence requirements cause delay and whether supporter confirmations add information. Review approval rates by age, gender, community role, employment status and network characteristics where lawful and proportionate. A community mechanism should not quietly exclude newcomers, isolated people or those in conflict with local gatekeepers.

Repayment and household outcome

Track scheduled principal, on-time payment, arrears, forbearance, write-offs and early repayment. Do not treat repayment alone as evidence of benefit. A borrower can repay by cutting essentials or taking more expensive debt. Follow-up should ask whether the expense was resolved, whether another debt replaced the loan and whether the participant would choose the product again.

Liquidity

Report fund inflow, disbursement, repayments, available cash, committed-but-undisbursed amounts and the number of suitable applications delayed for lack of funds. Stress tests should model clustered claims and slower repayments. The £10.50 monthly community allocation means that loan capacity grows gradually unless external capital or donations provide the initial pool.

Cost and governance

The £4.50 operating allocation should be tested against the true cost of onboarding, verification, payment, case management, safeguarding, arrears and reporting. Volunteer labour and donated services should be valued, not treated as free. Complaints, conflicts of interest and exceptions should be reported to the governing body.

Decision thresholds

Before launch, the pilot board should define the evidence needed to expand, redesign or stop. A sensible dashboard would include application rate, median decision time, approval rate, referral rate, thirty- and ninety-day arrears, principal loss, fund utilisation, operating cost per case, repeat applications and participant-reported outcome.

Thresholds should not be invented after the results are known. A pilot that attracts many applications but produces unacceptable arrears has learned something important. So has a pilot with low demand. Neither result should be recast as success through selective metrics.

Risks that the model must own

Regulation and charitable purpose. Interest-free does not automatically mean unregulated, and charitable status does not remove consumer-protection obligations. The proposed legal route, lending exemption analysis, AML position and restricted-fund governance require professional confirmation before public launch. Draft documents are not regulatory clearance.

Affordability. Removing interest improves price but does not make principal affordable. The model needs a proportionate assessment that does not recreate a conventional credit score while still identifying likely harm.

Liquidity. A revolving fund can fail while its borrowers are repaying as agreed if demand arrives faster than principal returns. Reserves and a transparent queue matter as much as long-run loss.

Social pressure. Supporter confirmation may strengthen trust, but it can also disclose sensitive circumstances or create informal obligations. Supporters must understand that they are not guarantors, and applicants need a route that does not depend on popularity.

Safeguarding and fraud. Urgency can reduce scrutiny. Staff and volunteers need escalation routes for coercion, domestic abuse, scams, self-harm, illegal lending and misuse of identity.

Mission drift. A system designed for essential shocks can become a general credit provider through repeated exceptions. Product boundaries and published reasons for decline protect both the fund and participants.

Evidence bias. Early adopters are unlikely to represent the whole community. Partner organisations may refer the most engaged members. Repayment during a small, high-touch pilot may overstate later performance. The evaluation should document these differences rather than generalise from them.

What this study does not establish

The research does not calculate a total addressable market. It does not estimate the number of households that currently borrow at high cost, the number refused by charities or credit unions, or the proportion willing to contribute each month. It does not provide UK-wide charity coverage because the structured charity screen is Scotland-only.

It does not prove that union-benefit expenditure is rising because of financial hardship. It does not turn Covenant Fund project descriptions into verified beneficiary outcomes. It does not compare customer-level approval speed or loss rates across providers. It does not establish that the InnerCircle legal structure is authorised, registered or operational.

The benefit and grant theme charts use model-assisted classification without completed manual QA. They help readers see patterns, but should not support a precise causal or programme-performance claim. A release intended for academic peer review should complete blinded manual coding and report agreement statistics.

These limits do not invalidate the work. They determine what kind of conclusion is justified.

Conclusion

Britain’s service communities are surrounded by support institutions, but the institutions are specialised. Unions defend and represent. Charities assess need and fund services. State schemes determine statutory entitlement. Credit unions lend on fairer terms than many commercial alternatives. Earned-wage access solves some timing problems. Each performs a recognisable task.

The evidence suggests that a household can still fall between them when an essential cost is urgent, a grant is unsuitable or slow, and interest-bearing credit is avoidable but affordable principal repayment is possible. That is the proposed missing middle.

InnerCircle offers one design for testing it: regular community contributions, separately governed funds and interest-free repayable assistance. The design is coherent enough to pilot and unproven enough to require restraint. The next serious step is not a national launch or a large market claim. It is a small, transparent trial that records who needs the product, what other support was unavailable, whether the decision was quick and fair, how repayments affect households and whether the fund remains liquid.

If the pilot produces that evidence, InnerCircle can make a stronger claim. If it does not, the project should change. That is the value of research at this stage: it turns a compelling idea into a proposition that can be proved, limited or rejected.

Method and reproducibility

The analytical build was frozen on 18 September 2026. Headline tables were regenerated from the supplied CSVs before drafting. AR21 headline comparisons use native-text rows only. OCR-only years are excluded. Multi-label benefit and grant classifications use fractional allocation so that one item’s full value is not counted in every category. Covenant Fund records are grouped by the year in the award date. Partial-coverage years are marked and are not used for growth claims.

The web package accompanying this article contains the derived chart tables and all 14 publication figures. It does not contain private legal drafts or personal data. Product terms were checked against provider pages accessed in September 2026.

Source notes

Editorial data audit

Item in the supplied inception pack	Audit result	Treatment in this publication
Covenant classification QA reported as 40 records and approximately 95 per cent accurate	Delivered QA table has 67 sampled rows and no manual verdicts	Accuracy claim removed; classifications labelled exploratory
Covenant median described as £190,000 in 2023 and £80,000 in 2025	Reproducible build returns £100,000 and £60,000	Corrected values used
AR21 analysis described as twelve organisations	Twelve organisations exist in the corpus; nine have usable latest native-text membership and benefit values	Figure 3 states nine organisations
2026 individual insolvency value described as a calendar-year total	September publication date means the row is year-to-date	Excluded from headline analysis
£13.58 billion National Lottery total	Workbooks contain overlapping vintages and are not deduplicated	Excluded from headline findings
1,474 Scottish charities	Valid as a keyword-screen result, not a UK market size	Geography, overlap and relevance limits stated
127 partner targets and approximately 3.4 million known members	Membership populated for only ten records	Used only as an outreach map; no TAM or SAM claim