Greensill Capital made receivables verification and funding concentration a supply-chain-finance accountability test

Summary

• Supply-chain finance was not itself the failure. Conventional programmes can shorten the time between delivery and supplier payment: a finance provider pays an approved invoice and later receives payment from the buyer. The accountability problem changes when financing extends beyond independently confirmed invoices, when a financier selects or describes assets that outside investors cannot verify directly, or when the same economic group dominates both origination and repayment risk. The UK Treasury Committee's Lessons from Greensill Capital explicitly resisted treating the technique as inherently objectionable while identifying serious weaknesses around non-bank data, regulatory perimeter and appointed representatives.

• The March 2021 collapse was a liquidity trigger built on a broader dependency system. Credit Suisse closed four Greensill-linked funds, German supervisors restricted Greensill Bank, and the availability of credit insurance and investor funding contracted. Greensill Capital (UK) Limited and a related UK management company entered administration on 8 March. Those events did not prove one simple cause. They exposed a structure in which continued financing depended on confidence in receivables, insurance protection, securitisation channels and a small number of important relationships.

• Receivable existence and concentration require different controls. A valid invoice can still create excessive exposure if many invoices depend on one obligor or connected group. A diversified list of nominal customers can still conceal common economic risk. Possible future receivables pose an additional issue because the underlying sale or contractual obligation may not yet exist. The responsible control is therefore not merely a document count. It is a traceable chain from purchase order and delivery through buyer approval, contractual debt, assignment, insurance, payment and exception resolution, combined with aggregation by ultimate obligor and connected borrower.

• The Credit Suisse funds supplied a specific supervisory finding, not a universal verdict. FINMA found that Credit Suisse had little knowledge and control over concrete claims in the four funds, relied on Greensill for selection and insurance, and did not initially recognise the consequences of possible future claims. It found serious breaches of Swiss supervisory law by Credit Suisse in risk management and organisation. Those findings are powerful evidence about the bank's duties and the funds, but they are not findings against every Greensill entity, every invoice, every investor or every other gatekeeper.

• The German, UK and Australian records have different scopes. BaFin's published concern addressed Greensill Bank's inability to evidence specified balance-sheet receivables purchased from the GFG Alliance. The UK FRC announced an investigation into Saffery Champness's audit of the UK company's 2019 financial statements, and that case remains listed as current. Australian disciplinary findings concerned a named lead auditor and the 2018 and 2019 consolidated group audits. None of those records can be silently expanded into an identical finding about all audits or all receivables.

• Open investigations and closed inquiries must be labelled accurately. The Serious Fraud Office says it is investigating suspected fraud, fraudulent trading and money laundering involving GFG Alliance companies, including their financing arrangements with Greensill. That is an investigation, not a conviction. Conversely, the FCA closed its Mirabella Advisors investigation in December 2025 after identifying no breaches requiring further action. A responsible history preserves both procedural statuses instead of presenting every inquiry as proof of misconduct.

• Exposure is not loss, and claims are not recoveries. The British Business Bank's public-backed lending record described nearly £335 million of maximum guarantee exposure if conditions were met and borrowers did not repay. Fund money invested at closure, creditor claims in an administration, bank deposits, estimated recoveries and cash actually distributed are different measures. Each has a date, legal perimeter and uncertainty. Combining them produces an impressive but meaningless total.

• Repair must be proved at the transaction and governance levels. A durable system would give fund managers, bank risk teams, boards, insurers, auditors and public accreditors independent access to underlying evidence; enforce connected-party concentration limits; surface prospective receivables; test insurance expiry and cancellation; reconcile cash to assets; preserve dissent; and stop new funding automatically when evidence is incomplete. Policy reform or a new dashboard is only design evidence. Repair is demonstrated when operating records show that exceptions are found, escalated, resolved and independently retested.

The financing technique and the accountability boundary

Supply-chain finance begins with a practical business problem. A supplier has delivered goods or services but will not receive cash until the buyer's payment date. A finance provider can pay the supplier earlier at a discount and collect the full amount from the buyer later. When the buyer has confirmed the invoice, the financier can assess a relatively short-dated claim against that buyer rather than relying only on the supplier's credit. The arrangement may strengthen smaller suppliers' liquidity and allow a buyer to manage payment processes. Nothing in that sequence requires deception.

Accountability depends on what evidence changes hands. The funder needs to know that the supplier exists, the delivery occurred, the buyer approved the amount, the obligation is enforceable, the receivable has not already been assigned, and any dilution through returns, disputes or set-off is captured. It also needs to know when the buyer must pay and what happens if the supplier or financier fails. An electronic platform can accelerate these checks, but automation does not make the source data true. If the platform accepts an originator's assertion without independent confirmation, it can scale uncertainty faster than a manual process.

The distinction becomes sharper with prospective receivables. A conventional approved invoice records an obligation arising from a completed or accepted transaction. A prospective receivable may depend on business expected in the future. That can resemble working-capital lending against forecast revenue rather than invoice discounting against a present debt. It may still be financeable, but the credit analysis, disclosure, insurance and valuation must reflect the different risk. Calling both assets “receivables” does not give them the same evidential quality.

Concentration is a second axis. Thousands of documents do not create diversification when the ultimate ability to pay comes from one industrial group. Systems should aggregate exposure by legal obligor, parent, guarantor, connected borrower, industry, geography and common risk driver. They should also identify circular relationships: whether a borrower, supplier, insurer, funder or originator depends on the same group for revenue or liquidity. A portfolio can look granular by invoice while remaining concentrated economically.

The Greensill case is therefore not best understood as a morality play about invoice finance. It is a control test. Who could reject an asset? Who independently contacted the obligor? Who could see total connected exposure? Who monitored policy expiry? Who reconciled securitised notes to underlying claims? Who could stop origination when information failed? Those questions apply even where no misconduct is alleged and no individual legal finding exists.

A funding model dependent on confidence, insurance and outside investors

Greensill connected several layers. It originated or purchased financing assets; securitisation vehicles converted exposures into notes; investment funds and other financiers supplied capital; insurers were expected to cover substantial credit risk; and a German bank operated within the wider group. Each layer could make economic sense in isolation. Together they created dependencies that required clear ownership of verification and liquidity stress.

Insurance can reduce loss severity, but only within its terms. Coverage has named insureds, eligible assets, exclusions, limits, expiry dates, notification duties and cancellation provisions. A control system needs policy-level evidence tied to each financed asset and should model what happens if renewal fails. It should never treat a historical policy, broker indication or expected renewal as permanent capital. It should also test whether claims would be payable if the underlying receivable was prospective, disputed, inaccurately described or concentrated beyond a limit.

Outside-investor funding creates another rollover dependency. A financier that continually originates assets needs new or returning capital as earlier assets mature. If a fund suspends purchases or redemptions, the originator may lose a major route for funding even if some underlying assets will ultimately pay. Liquidity can therefore fail before final credit losses are known. The distinction matters: inability to continue financing today is not proof that every asset is worthless, while ultimate recoveries do not prove the pre-failure liquidity design was safe.

The administrator's evolving record is available through the UK company's Companies House filing history, which contains proposals and successive progress reports rather than one final loss statement. The sequence shows why dated language is essential. Creditor claims can be submitted, rejected, revised or contingent; collections and litigation can continue; costs accrue; and estimated returns change as assets are realised. A filing proves what the administrators reported at that date. It should not be converted into a permanent aggregate for the entire international group.

The trigger in early March 2021 compressed these dependencies. Fund closures reduced a central investor channel. Supervisory action at the German bank constrained another part of the group. Insurance uncertainty affected the credit protection supporting assets. The UK company told the court it lacked a viable path to continue, and administration followed. A single-cause label—“insurance caused the collapse,” “a bank regulator caused the collapse,” or “one borrower caused the collapse”—would mistake the final break for the whole structure.

Credit Suisse: asset-manager duties could not be delegated away

The four Credit Suisse supply-chain-finance funds are the clearest official record of how an external fund manager's control can fail. FINMA's 2023 enforcement release said clients had invested about $10 billion in the four funds when they closed. That is money invested at closure, not a final realised loss. FINMA focused on whether Credit Suisse met Swiss supervisory duties in its relationship with Greensill.

FINMA described an arrangement in which Greensill selected and reviewed claims and arranged insurance in its own name. It found that Credit Suisse's asset-management company had little knowledge and control over specific claims. Possible future claims entered the funds, and Credit Suisse did not initially understand the significance of that change or know how many claims were actually contractually owed. Critical questions were handled by people responsible for the commercial relationship rather than by an independent challenge function.

FINMA concluded that Credit Suisse seriously breached supervisory obligations concerning risk management and adequate organisation.

This finding identifies an ownership principle: delegation of work is not delegation of accountability. A fund manager may use an originator's platform and expertise, but it remains responsible for understanding the assets it buys for clients. Independent verification does not necessarily mean manually confirming every invoice. It means designing risk-based checks whose populations, exceptions and outcomes are visible to the manager; obtaining direct obligor evidence; reconciling note holdings to assets; testing future-receivable classifications; and retaining authority to suspend purchases.

Concentration monitoring also belongs with the fund manager. Nominal asset counts should roll up to the economic debtor and related groups. Limits must apply before purchase, not after a portfolio report arrives. Overrides should state who approved them, why, for how long and with what compensating protection. Senior management should see a consolidated relationship view that includes funds, loans, fees, prospective mandates and conflicts. FINMA's ordered measures—including periodic executive-level review of important relationships and clearer responsibility records—addressed that wider governance need.

The later FINMA report on Credit Suisse places the supply-chain-finance funds within a broader institutional history. Its inclusion does not make every Credit Suisse weakness a Greensill finding. It is useful because it reinforces the difference between responding to an incident and maintaining a bank-wide view of significant relationships, incentives and risk ownership. A remediation programme is credible only when subsequent testing shows those controls operating under commercial pressure.

Greensill Bank: preserve the exact BaFin finding

German supervision adds a different entity and a different evidentiary question. BaFin's Greensill Bank FAQ described the moratorium and the steps relevant to depositors. The measure protected the bank's assets while authorities assessed its condition; it was not a judgment that every affiliated company or receivable was fraudulent. Deposit-protection arrangements also concern eligible depositors under defined rules, not investors in the Credit Suisse funds or creditors of the UK company.

The contemporaneous BaFin Journal account stated that a forensic special audit found the bank unable to provide evidence of the existence of specified receivables it had purchased from the GFG Alliance. The scope matters. This was a finding about particular balance-sheet receivables and evidence at Greensill Bank. It does not establish that no Greensill invoice anywhere existed, that all GFG obligations were identical, or that the same finding applied to every fund asset.

At the same time, the limited scope does not make the control issue narrow. A regulated bank inside a financing group needs an independent credit file for assets bought from affiliates or concentrated customers. Evidence should include the contract creating the obligation, proof of performance, debtor acknowledgement, assignment, ageing, payment history, dispute status, collateral and insurance. Confirmation should be sent to a contact independently obtained from the debtor, not only routed through the seller. Related-party purchases require enhanced review because group incentives can weaken challenge.

Depositor protection also demonstrates why impact categories must be separated. A protection scheme may compensate eligible deposits and later pursue the estate. That payment is not proof of the bank's gross asset loss, and the scheme's outlay may change through recoveries. Depositors above limits, institutional creditors, noteholders and other claimants may have different positions. A public figure for protected deposits cannot be added to fund assets, UK creditor claims and government guarantees as though each measured the same harm.

The proper cross-border question is whether information travelled quickly enough. The UK company, Australian parent, German bank, Swiss-managed funds, insurers and borrowers sat in different legal and supervisory perimeters. No one regulator necessarily saw the entire group. Durable reform requires entity maps, intra-group exposure reporting, named supervisory contacts, prompt exchange of verified concerns and explicit ownership of the consolidated risk view. Information sharing must respect legal limits, but fragmentation cannot become an excuse for nobody assembling the whole.

The UK perimeter and the appointed-representative lesson

Greensill Capital (UK) Limited was not a bank authorised by the UK prudential regulator. For some regulated activity the group used an appointed-representative structure. Under that model, an authorised principal accepts responsibility for defined regulated activities of a representative. The arrangement can give smaller or specialist businesses access to a regulatory umbrella, but it creates an obvious control question: did the principal understand and supervise the business actually conducted under its permissions?

The historical boundary is unusually important here. The FCA opened an investigation in 2021 into Mirabella Advisors LLP's oversight of Greensill Capital Securities Limited. In December 2025 the FCA published its closure statement, saying it had not identified breaches by Mirabella requiring further action. The investigation closed, subject to the regulator's ability to revisit the decision if new information emerged. An article that still describes the Mirabella inquiry as unresolved is outdated; one that implies the closure cleared every Greensill company or every separate FCA matter is equally wrong.

GAM is a separate, resolved actor-specific record. In March 2022 the FCA fined GAM International Management Limited and former investment director Timothy Haywood for failures involving due care and conflict management. The regulator said three transactions were involved, two linked to Greensill, and that potential incentives offered to GAM or its parent were not taken up but were not handled properly. It also found that Haywood recorded gifts and entertainment late. Crucially, the FCA said it did not find evidence that those benefits caused his investment decisions. The findings and fines belong to GAM International Management and Haywood for the stated periods; they do not establish misconduct by Mirabella, Greensill, every GAM fund, every investor or every Greensill-linked transaction.

The appointed-representative debate extends beyond one principal. The FCA's consultation on improving the regime examined information, notification, due diligence, oversight and termination controls. A consultation records proposed policy reasoning, not proof that later rules have operated successfully. Its value for the Greensill control map lies in the questions it formalises: what business a representative plans to conduct, whether the principal has skill and resources to supervise it, how complaints and revenue are monitored, and when permissions should be restricted or the relationship ended.

Public-backed lending: speed did not eliminate accreditation duties

The pandemic lending schemes brought public money into the accountability perimeter. The National Audit Office's investigation into the British Business Bank's accreditation reported that Greensill received accreditation under CBILS and CLBILS, made £418.5 million of loans, and had up to nearly £335 million of government guarantee exposure based on the 80 per cent guarantee. The last figure was maximum exposure if guarantees applied and borrowers failed to repay, not a crystallised taxpayer loss.

The NAO explained that the Bank used a streamlined accreditation process designed for speed and relied more heavily on information from applicants and post-accreditation audit. Greensill was approved to lend up to £400 million under CLBILS, with limits applying to borrowers. The Bank later became concerned about seven loans totalling £350 million to borrowers within the GFG Alliance, opened an investigation and suspended the guarantees. Greensill, through its administrators, disputed that it had breached scheme rules and challenged aspects of the process. Those positions must remain in the record.

The Public Accounts Committee's accreditation report examined whether the Bank and departments balanced urgent support with taxpayer protection. Its conclusions are parliamentary findings and recommendations, not a court judgment on Greensill's compliance. The government's formal response recorded its position on the recommendations and the suspended guarantees. A response is evidence of commitment and interpretation; completion requires later implementation evidence.

Accreditation should test capability and incentives as well as policy eligibility. A public body needs verified ownership, financial resilience, operational capacity, underwriting standards, connected-borrower aggregation, fraud controls, data access and a credible wind-down plan. It should define what the guarantee covers and reserve audit rights down to underlying borrower files. Rapid approval can use staged limits, enhanced sampling and automatic pauses rather than simply accepting less evidence.

Post-accreditation monitoring must also be fast enough to matter. Lending data should identify ultimate connected groups before a limit is consumed. Alerts should not wait for manual aggregation after disbursement. When a concern arises, the system should freeze additional guarantees while preserving evidence and giving the lender a fair process to respond. The record should distinguish an allegation of non-compliance, a suspended guarantee, a final eligibility decision, a claim on government and cash actually paid.

Government access and NHS continuity were separate governance tests

Greensill's contacts with government produced intense scrutiny, but lobbying, procurement and financial failure should not be collapsed. The Cabinet Office's Boardman review examined the development and use of supply-chain-finance arrangements in government and made recommendations. The Treasury separately released communications with Greensill, allowing the channels and timing of approaches to be examined. These records support transparency analysis; they do not by themselves establish a corrupt bargain or prove that every approach changed a decision.

The Treasury Committee found that ministers and officials were right to consider emergency financing proposals and right to reject Greensill's requested access to the Covid Corporate Financing Facility. It criticised informal lobbying and recommended stronger processes, while stating that Treasury ministers and officials had acted with integrity in handling the approaches. The Chancellor's published response to parliamentary correspondence described the requests and existing prompt-payment policy. The balanced lesson is not that access always secures a result. It is that privileged access needs a complete, searchable record so outsiders can see what was requested, considered and decided.

NHS arrangements create a continuity question. A parliamentary report on pharmacy early payment and salary advance schemes found weak evidence of promised benefits, low pharmacy participation and limited public evidence curiosity about a free salary-advance service that could create reputational value. These were committee findings about public administration. They should not be converted into a finding that every pharmacy payment or salary advance was invalid.

The Public Administration and Constitutional Affairs Committee's interim governance report examined Lex Greensill's government role, appointments and propriety. Its scope differed from the financial-regulation and public-loan inquiries. Together, the reports show why an institution needs a relationship register: advisory roles, commercial interests, meetings, proposals, contracts, pilots and decisions should be linked without assuming that the existence of a relationship proves impropriety.

Continuity planning matters when a private intermediary supplies a public-facing process. Before adoption, the public body should know how pharmacies, employees or trusts will continue if funding stops or the provider enters insolvency. Data portability, payment reconciliation, substitute suppliers, communications and termination rights should be tested. A service offered free of charge is not free of dependency risk; the business rationale and cross-selling incentives should be understood.

Audit accountability must follow the entity and audit period

Audit evidence in the Greensill history is easy to overstate because different firms and jurisdictions examined different financial statements. The UK Financial Reporting Council announced an investigation into Saffery Champness in relation to the audit of Greensill Capital (UK) Limited's financial statements for the year ended 31 December 2019. The FRC's current enforcement-case register still lists the case as current. Opening an investigation is not a finding of breach, and a current investigation should not be described as exoneration or misconduct.

Australia produced a resolved but differently scoped record. ASIC's December 2024 release reported that the Companies Auditors Disciplinary Board suspended registered auditor Joseph Santangelo until June 2026 after findings concerning his duties as lead auditor and engagement partner for the consolidated Greensill Group audits for 2018 and 2019. The Board's media release described accepted failures across specified audit responsibilities. Those findings apply to that auditor, role, group audit and period; they do not decide the pending UK investigation.

The group-audit problem is structurally important. A parent auditor may rely on component auditors across countries, but it must identify significant components and risks, communicate instructions, evaluate component work, challenge unusual balances and document its conclusions. Receivables, related parties, insurance, concentration, going concern and funding dependency deserve explicit treatment. A clean opinion is not proof that every asset existed; it is a professional conclusion based on the audit scope, materiality and evidence obtained.

Audit repair needs more than rotating a firm. The audit committee should map which auditor covers each entity and which balances cross those boundaries. Direct confirmations should be independently controlled. Prospective receivables should be separated from billed amounts. Insurance should be confirmed with carriers and matched to eligible assets. Cash receipts after year-end should be reconciled to specific debtors. Concentration should be visible at ultimate-group level. Going-concern testing should model loss of insurance or fund demand rather than assuming renewal.

The legal status field is as important as the technical finding. An internal review, regulator investigation, disciplinary admission, tribunal decision and court judgment carry different authority. A public accountability ledger should record the actor, mandate, period, allegation or issue, present status, findings, appeal and remedy. Without that structure, a pending UK case can be mistakenly reported as resolved because an Australian case ended, or the Australian finding can be diluted by pointing to a separate unresolved inquiry.

Criminal investigation is not a substitute for transaction evidence

The Serious Fraud Office's GFG Alliance case page states that it is investigating suspected fraud, fraudulent trading and money laundering in relation to the financing and conduct of GFG Alliance companies, including financing arrangements with Greensill Capital UK. The page records an open investigative posture. It does not establish a charge, plea, verdict or conviction against Greensill, Lex Greensill, the GFG Alliance as a whole or every person associated with either group.

That boundary should shape every sentence. “The SFO is investigating suspected conduct” is accurate. “The SFO found fraud” is not supported by the case page. Nor can the investigation prove that a particular invoice was false, that all prospective receivables lacked a basis, or that every GFG exposure will produce a loss. If a later charge or disposition occurs, it must be appended with its date and exact defendants; it should not retroactively change what the 2021 investigation announcement established.

Criminal investigation also cannot replace operational verification. A fund manager, bank or public lender must decide whether an asset is eligible before prosecutors could possibly complete a case. The control evidence is commercial and contemporaneous: obligor confirmation, contract, delivery, acceptance, assignment, insurance and payment. Suspicion can trigger enhanced review, but absence of a criminal case is not validation, and an open case is not a basis to assign guilt.

The same logic applies to recovery. Administrators may pursue receivables or claims under civil and insolvency rules while criminal investigators examine suspected conduct. A payment, settlement or asset sale can occur without a criminal finding. Conversely, a criminal conviction would not automatically determine every creditor's recovery. Separate ledgers should track evidence, legal process and cash.

For boards, the lesson is to preserve records before a crisis. Each financing asset should have immutable provenance; changes should be logged; emails and approvals should be retained; related parties should be identified; and exceptions should show who decided what. That record assists routine risk management, audit, administration and any later investigation. It also protects people and institutions from overbroad accusation by allowing actor-specific facts to be tested.

Loss accounting requires a measurement dictionary

Greensill generated many large figures, but they do not share a denominator. Approximately $10 billion invested in four funds at closure describes fund scale at a date. Nearly £335 million described potential UK government guarantee exposure under conditions. A bank's protected deposits measure eligible liabilities, not a final asset shortfall. Administrator creditor claims may be contingent, duplicated across entities or subject to adjudication. Estimated realisations can change. Fees, enforcement penalties and litigation settlements are different again.

A sound accountability report begins with a measurement dictionary. Every number should carry an entity, currency, valuation date, gross-or-net basis, legal status and source. “Exposure” means an amount at risk before recovery. “Loss” means a realised or recognised reduction under a stated method. “Claim” is an asserted entitlement, not necessarily admitted. “Provision” is an accounting estimate. “Recovery” may mean cash collected by an estate, cash distributed to a creditor, or an insurer's subrogated collection; those are not interchangeable.

Time matters as much as classification. A fund may return cash over years, changing the investor shortfall. A government guarantee may remain suspended, be denied, reinstated or paid. Administrators update expected dividends as litigation and collections progress. Depositor-protection schemes may recover from an estate after compensating depositors. Reporting should retain each dated snapshot and show movement rather than overwriting history with the newest headline.

A repaired control system starts with receivable-level proof

The minimum data model for receivables finance is straightforward to describe and difficult to operate consistently. Each asset needs a unique identifier; supplier and obligor legal entities; purchase order; invoice; delivery or performance evidence; buyer approval; amount and currency; due date; assignment history; dispute and dilution status; insurance mapping; funding vehicle; and cash settlement. The record should distinguish a present contractual receivable from an expectation of future business.

Verification should be risk based but independent. Low-risk, repeat invoices can be sampled and reconciled automatically. New obligors, unusual growth, manual entries, round amounts, long tenors, related parties and prospective claims require enhanced review. Direct confirmation routes must be obtained independently of the originator. The system should detect duplicate invoice numbers, repeated bank accounts, conflicting assignments and payments that do not match the expected obligor.

Concentration controls sit above the transaction layer. Exposure should aggregate across spelling variants, subsidiaries, guarantors and connected groups. Limits should cover funded amount, insured amount, uninsured tail, prospective receivables and stressed recovery. A board should see how exposure changes if insurance ends, one investor channel closes, a key borrower delays payment or a rating falls. The risk view should include revenue concentration too: large fees from a relationship can weaken challenge even when credit exposure seems limited.

Insurance controls require direct carrier evidence and expiry alerts. The institution should know whether the policyholder, beneficiary and financed asset align; whether the policy can be cancelled; and whether exclusions could apply. Renewal assumptions belong in stress testing, not in the base evidence column. If coverage becomes uncertain, new purchases should pause automatically until an authorised independent function approves a documented exception.

Governance controls must make refusal possible

Control design fails when everyone can raise a concern but nobody can stop a transaction. The repaired model assigns explicit decision rights. Originators may propose assets; an independent credit function verifies eligibility; concentration systems aggregate risk; insurance specialists validate coverage; treasury confirms funding; legal teams map assignment and regulatory perimeter; and the board risk committee approves limits. Each function needs authority to pause funding without commercial retaliation.

Overrides should be rare, time limited and visible. An override record should state the failed control, evidence considered, decision owner, dissent, compensating protection, expiry and required remediation. Repeated overrides for the same customer or product should trigger a portfolio-level review. Senior executives should not be able to turn an exception into normal policy through a series of one-off approvals.

Auditors and fund managers need their own independent data access. Reliance on management representations is sometimes necessary, but significant balances should be corroborated. Boards should ask whether the person responding to criticism owns the commercial relationship under review. If so, another function should validate the answer. The control objective is not to eliminate expertise from relationship teams; it is to prevent expertise from becoming unchallenged advocacy.

Cross-border governance needs a responsible group-level owner even where no single regulator has consolidated jurisdiction. The entity map should identify regulated status, auditors, key counterparties, intra-group transfers and supervisors. Material exceptions should be shared through legally permitted channels. A concern at a bank, fund or auditor should prompt the group board to test analogous exposures elsewhere without assuming the finding automatically applies.

Proof of repair is operating evidence, not policy language

The post-Greensill reforms and supervisory measures establish expectations, but a rulebook does not show whether people follow it. Proof of repair should be empirical. A board should receive the population of receivables, the number independently confirmed, exceptions by type and age, overrides, connected-group concentrations, insurance gaps, reconciliation breaks and subsequent cash performance. Metrics should retain both numerator and denominator so improvement cannot be manufactured by redefining the population.

Testing should include adverse cases. Can the system detect an invoice submitted twice through different vehicles? Does it identify two legal entities controlled by the same group? What happens when an insurer declines renewal? Can a fund manager retrieve obligor evidence without asking the originator? Does a public guarantor see connected-borrower exposure before disbursement? Can administrators export the records needed to collect assets if the platform owner fails?

Independent assurance should follow remediation through closure. A control described as implemented should have an owner, start date, test method, exceptions and retest result. Supervisors should publish enough aggregate information to show whether ordered measures were completed while protecting confidential data. Audit committees should track unresolved findings across entities and years. A case should not disappear from a dashboard merely because responsibility moved to another team.

Outcome measures matter. Lower exception rates are useful only if sampling remains robust. Diversification is credible only when calculated at ultimate-obligor level. Insurance coverage is meaningful only when eligible and collectible. Liquidity resilience is demonstrated by successful stress tests and controlled wind-down exercises. Public value is shown by realised service and repayment outcomes, not only by quick accreditation or projected benefit.

Finally, procedural truth is part of repair. The organisation's public record should distinguish an investigation from a finding, a committee conclusion from a judgment, a suspended guarantee from a paid claim, and exposure from realised loss. The written parliamentary answer on UK regulatory roles is an example of why entity-specific descriptions matter: the UK company was not itself FCA-authorised, while related activity and overseas entities sat under different arrangements. Clear labels prevent both unjustified accusation and convenient evasion.

What boards, regulators and public accreditors should require

The Greensill record supports a practical accountability covenant. No receivable enters a financed pool without a traceable contractual basis or an explicit prospective-receivable classification. No portfolio report stops at invoice count; it shows ultimate-obligor and connected-group concentration. No insurance amount is treated as capital without policy eligibility, expiry and cancellation data. No fund manager delegates selection without retaining independent verification and stop authority.

No bank buys affiliated or concentrated assets without direct evidence and enhanced review. No appointed-representative relationship is described merely by the principal's name; the scope of permissions, supervision and actual business is recorded. No public guarantee is attached without borrower aggregation, data access, staged limits and a tested wind-down process. No government approach is kept only in informal channels when public money or policy change is requested.

No audit conclusion is detached from entity, period and mandate. Group auditors demonstrate how component work and significant risks were challenged. Audit investigations remain investigations until resolved; disciplinary findings remain limited to the people and work decided. Criminal inquiries are reported as inquiries. Closed regulatory investigations are updated promptly, including the fact that Mirabella's case ended without breaches requiring further action.

No impact total mixes funds invested, deposits, guarantees, claims, provisions, recoveries and losses. Each measure has a date and legal owner. Employees and suppliers receive continuity plans; investors receive asset and recovery reporting; taxpayers receive guarantee status; creditors receive dated administration estimates. Different stakeholders can then see the evidence relevant to their remedy.

These requirements are not anti-innovation. They are the infrastructure that permits financial technology and private capital to serve smaller suppliers without turning speed into opacity. Automation is especially valuable when it preserves provenance, aggregates concentration and blocks incomplete assets. It becomes dangerous when it makes an originator's unsupported field look authoritative merely because it appears in a polished system.

Greensill Capital's failure became an accountability test because the same economic relationship crossed so many gates. Receivables moved toward funds and a bank; insurance supported confidence; public schemes relied on accreditation; government services tested private financing; auditors examined different entities; and regulators saw different slices. The answer is not to pretend one institution could see everything. It is to require every institution to own its decision, preserve its evidence, share material risk appropriately and stop when proof is missing.

Conclusion

The durable lesson from Greensill is a distinction between financing an evidenced obligation and financing confidence in a story. Supply-chain finance can improve supplier cash flow. It cannot be governed safely by labels, platform scale, insurance expectations or the reputation of connected institutions. The underlying debt, concentration, coverage, funding and authority to refuse must be visible independently at every gate.

The March 2021 administration was the trigger through which dependencies became undeniable. The root was broader: outside-investor and insurance reliance, limited asset visibility, prospective-receivable risk, concentrated relationships, cross-border fragmentation, regulatory-perimeter gaps and limited public evidence independent challenge. The impacts spread through funds, creditors, employees, industrial borrowers, depositors, public guarantees and services, but each impact requires its own measurement.

Official records also impose discipline on accountability itself. FINMA made findings against Credit Suisse. BaFin published a limited finding about specified bank receivables. Australian authorities resolved defined auditor duties, while the UK audit investigation remains current. The SFO inquiry remains an investigation. The FCA's Mirabella inquiry closed without identifying breaches requiring further action. Those differences are not caveats to hide; they are the structure of an honest account.

Repair will be credible when evidence survives pressure: obligors are independently confirmed, prospective claims are explicit, connected exposure is aggregated, insurance failure is stress-tested, fund managers retain control, public accreditors can pause guarantees, auditors challenge across entities, and boards can see unresolved exceptions. Until those operating results are available, a revised policy is a promise. Accountability begins when institutions can prove, transaction by transaction and decision by decision, why the promise should be believed.