// Institutional exposure · Core distinction

Principal–agent risk versus the Fiduciary Gap™

One describes an agent with different interests. The other describes an agent with no interests at all.

Principal–agent risk describes what happens when an agent acting for a principal has incentives of its own. The Fiduciary Gap™ is the structural gap between what an autonomous system is optimised towards and what the institution is obliged to protect. Formally, it occurs when the agent's executional power exceeds its contextual encoding.

// 01

Three different records

Principal–agent risk

An agent with its own interests. Addressed by contracts, monitoring and incentives.

The Fiduciary Gap™

A system with an objective function and no interests. Incentives do not apply.

The remedy

Encoding, not contracting. Mandate has to reach the execution constraint.

// 02

A concrete example

Deposit optimisation: a retail bank deploys an agent to optimise deposit pricing. The agent is not disloyal, negligent or misaligned in the economic sense. It has no interests to misalign. It optimises precisely what it was given, and what it was given did not encode the institution's obligation to a particular class of customer. Every control reports green. The optimisation is exactly as specified. The obligation was simply never written into the constraint.

No contract would have prevented this, because there was no counterparty to contract with. The failure is one of encoding.

// 03

Why the distinction is not academic

Treating the Fiduciary Gap™ as principal–agent risk leads directly to the wrong remedies: sharper incentives, tighter contracts, closer supervision of a party that cannot respond to any of them. The economics literature assumes an agent capable of self-interest. An automated system is capable only of optimisation. What must change is not the agent's motivation but the fidelity with which institutional obligation is expressed into the constraints the system actually executes against.

Read the canonical definitions →

// Questions people ask

Common questions

Is the Fiduciary Gap™ just principal–agent risk renamed?

No. Principal–agent risk assumes an agent with interests that may diverge from the principal's. An automated system has an objective function rather than interests, so the classical remedies — incentive alignment, contracting, monitoring for shirking — have nothing to act upon.

Can better incentives close the gap?

No. There is no party to incentivise. The gap closes when fiduciary intent is expressed directly into execution constraints, which is a design and governance problem rather than an economic one.

How would an institution know it has one?

Ask what the system is optimised towards, then ask what the institution is obliged to protect, then compare the two in writing. Where the second is not present in the first, the gap is the distance between them.

// The practical test

Test one institutional decision

The ten-day Decision Drift Audit™ maps one material decision across all eight layers, assesses replayability and authority boundaries, and delivers one prioritised board finding.

Test one decision →Canonical definitions →Decision Integrity Chain™ →