Capital Mispricing and Structural Risk

Chapter 6: The Economics of Structural Opacity

When the financial model cannot see structural investment, a repeating cycle emerges: a platform investment is rejected, its absence produces incidents costing far more, and those costs are attributed to “engineering quality” without being traced to the original decision. The “headcount illusion” means AI-assisted development may be increasing structural fragility invisibly.

A European insurer with five hundred services invests € 1.2 million in a shared integration layer for its claims process: a common event infrastructure that would let the claims, policy, and payments teams exchange information independently, replacing a fragile web of direct connections that causes an average of three incidents per month.

The platform team delivers a working prototype in four months, and the three consuming teams are willing to migrate, with projected incident reduction between 70% and 80%.

The CFO kills the investment in the annual budget review because the platform “does not show ROI in year one.” The decision is structurally rational rather than a failure of imagination: he cannot approve what the accounting system cannot see. The € 1.2 million in salaries and infrastructure is visible; the benefits of fewer incidents, faster change, and reduced coordination are diffuse, and none appears as a line item in any business unit's P&L.

Over the following eighteen months, the direct connections cause nine major incidents, two of them customer-facing and one triggering regulatory scrutiny. A claims payment fails because the policy and payments systems disagree on whether a policy is active, a commercial fleet operator waits nine days for a claim that should have settled in two, and his broker emails the CEO. The CTO traces the failure to the integration point the cancelled platform would have replaced, but the incident log has already recorded it as a “data synchronisation issue.” The broker files a complaint with the industry ombudsman, which reaches the compliance function as a request for a root-cause report. The report describes the technical failure accurately and never mentions the investment decision that made it inevitable, because it follows the incident, not the decision. The CFO who cancelled the platform never sees the letter: it is handled two levels below him, in a function with no reporting line to the person whose decision caused it. The total cost of the nine incidents, including response, remediation, compensation, and regulatory reporting, is approximately € 2.1 million, all of it attributed to “engineering quality” in the quarterly review. Nobody connects it to the cancelled platform.

A year later, the budget includes a line item for “integration resilience,” funded at € 400,000, one third of the original investment and insufficient to do the work properly. The platform team has been disbanded, and its lead engineer has left. Finance occupies a position of unquestioned authority in most software-dependent corporates, presenting as prudence, stewardship, and inevitability. The CFO does not say “I am overruling your technical judgement.” He says “the numbers don't support it,” and the conversation ends. The asymmetry is one of form: the platform team's case was a presentation and a prototype, while the CFO's model was a budget with sign-off authority. When the formal artefact and the informal argument conflict, the formal artefact wins, not because it is right but because it is an artefact.

The condition has a name: structural opacity, in which an organisation's own instruments cannot trace the relationship between its decisions and their systemic consequences.

...

See also: Full contents · Preview chapters · Illusions of Work