Post 022 — Risk Management · Practitioner

2026

Preventing deep organisational collapse: a new paradigm for risk management

Three structural interventions that address the failure patterns — and why each carries the risk of becoming exactly what it was designed to prevent.

Post 021 introduced the four structural lenses — GMM, RBM, RSM, CPM — through the cases of Lehman Brothers and KDDI's G-Plan. This post goes further: it asks what organisations and their leadership can actually do to prevent the failure patterns those lenses reveal.

Three concrete interventions emerge from the analysis. Each addresses a different layer of the structural failure. And each carries its own risk of becoming exactly the thing it was designed to prevent: a formal compliance requirement that is fulfilled without substance.

Intervention 1: making "small death" the rational choice

Both KDDI's Person A and Lehman's Richard Fuld faced a moment where honest disclosure was available — and declined it, because the cost of disclosure was higher than the cost of concealment.

For Person A: reporting the initial small deficit would have meant business closure. Concealing it meant a chance of survival. The calculation was structurally rational. The fraud was the cost-minimising choice.

For Fuld: accepting the KDB acquisition terms would have meant acknowledging that Lehman's assets were worth what the market said. Narrative lock made that impossible.

The structural intervention at this level is not moral instruction. It is cost restructuring. The organisation must make honest early failure — the "small death" — structurally less costly than concealment.

What this looks like in practice:
  • Early disclosure of business difficulty is explicitly rewarded, not just not-punished. The leader who says "this isn't working" at month three is treated differently from the leader who surfaces it at year three.
  • Withdrawal from a new business is framed as competence — reading the environment correctly — not as failure.
  • Performance evaluations do not treat a business closure as a career-ending event. They treat concealment as one.

This is the CPM intervention: restructuring the survival anxiety calculation so that Layer A (career/organisational survival) points toward disclosure rather than concealment. Without this restructuring, every other governance measure operates in an environment where the incentive to circumvent it is stronger than the incentive to comply.

Intervention 2: mandating substance verification (bouching)

The Phantom Auditor state in both cases was enabled by a specific structural absence: the requirement to verify that the things being reported actually exist.

In KDDI's G-Plan, internal auditors completed their procedures without requesting physical evidence from third parties — actual invoices, actual counterparty confirmations, actual delivery records. The audit was formally complete and substantively empty.

In Lehman, the complexity of the financial instruments served the same function. Understanding whether a CDO's underlying assets were actually worth what the model said required expertise that oversight functions did not have — so they did not verify. Formally compliant. Substantively blind.

Phantom Auditor IndicatorWhat it looks likeStructural background
No substance checkAudit concludes "effective" based on interviews alone, without matching invoices to third-party evidence.Procedure completion treated as the objective. The form replaces the substance.
Information asymmetry exploitationTechnical jargon creates a "wall of expertise" that leaves management feeling unable to challenge.Architect Isolation: the reviewer cannot evaluate what they cannot understand.
Psychological deferenceThe "revenue-generating unit" is treated as untouchable. Meeting requests are refused and the refusal is accepted.Production loop capture of the correction loop.
The bouching requirement: for any transaction above a defined threshold, a party with no interest in the outcome must confirm, through direct contact with the counterparty (not through the transaction originator), that the counterparty exists and the transaction has economic substance. This is not a new concept. It is an old concept that was abandoned in exchange for efficiency — and needs to be reinstated as a structural requirement, not a discretionary option.

The crucial extension: any explanation that cannot be independently verified must be treated as a red flag requiring external escalation — not as a sufficient answer. "They prefer to work by phone" is not a valid response to a request for counterparty confirmation. The audit function must have a structural mandate to pursue the signal, not the discretion to accept a plausible-sounding deflection.

Intervention 3: institutionalising the red flag trigger

KDDI Chairman Takahashi's "this growth rate frightens me" was the statement that initiated the real investigation — seven years into the fraud, after every formal governance mechanism had failed.

The intervention is not to depend on this statement appearing at the right moment. The intervention is to institutionalise the conditions that produced it — and to create a formal trigger that activates when the kind of structural anomaly he identified exceeds a defined threshold.

The red flag institutionalisation:
  • Growth/infrastructure ratio threshold: when any subsidiary or division's growth rate exceeds the group's core business growth rate by a defined multiplier (e.g., 3×) over a sustained period, an automatic "substance investigation" trigger activates — independent of management reporting.
  • Business model plausibility gate: when a business unit's declared revenues cannot be explained by a coherent market-based narrative — "how is this unit capturing this much of this market?" — the question must be formally asked and formally answered before financing continues.
  • Cross-referencing declared revenue against external market data: in digital advertising and similar complex domains, declared transaction volumes can be compared against industry benchmarks. A unit claiming market share that does not exist in the independent data is a structural anomaly.

These triggers do not replace human judgement. They create a structural moment in which human judgement — the "this frightens me" intuition — is formally invited and formally consequential. The goal is not to automate the CEO's instinct but to ensure that the structural conditions for expressing it are not suppressed by the production loop's narrative.

What structural resilience actually looks like

The organisations that fail most dramatically are rarely the ones with no governance. They are the ones with governance that is comprehensive in form and empty in substance — where every box is ticked, every audit is complete, and the correction loop has been fully captured by the production loop's success narrative.

A structurally resilient organisation has three properties that cannot be produced by policy alone:

  1. Honest failure is the cheaper option. People under pressure can report early difficulty without career consequences. The structural cost calculation points toward disclosure.
  2. The correction loop cannot be deactivated by the production loop's narrative. The right to scrutinise is structurally decoupled from whether the unit is performing. A revenue-generating unit has no immunity to audit.
  3. The leadership capacity to say "this frightens me" is actively preserved. Senior leadership models the expression of discomfort. The systems that would filter it out — physical isolation, narrative lock, conformity pressure — are actively countered.
"Governance is not the completion of checklists. It is the continuous updating of narrative — and the preservation of a leadership capacity to say, out loud, that something does not feel right."
The final paradox: the most dangerous signal that an organisation needs this analysis is that its governance appears to be working. Comprehensive formal control systems are the most reliable environment for Phantom Auditor states to develop — because the system's apparent completeness removes the urgency to look more closely. The organisations that believe they are safe are the ones that most need to ask whether their correction loops are actually functioning, or merely formally present.
← Post 021Post 023 →