What is a credit escalation and why does it matter for model training?

A credit escalation happens when a decision sits above the authority of the person holding it. A credit officer reviewing a business loan application finds that the requested exposure exceeds their delegated limit, or that a policy exception is needed, so the case moves to a senior reviewer or committee.

In a working business this is rarely a single click. The first reviewer prepares a recommendation, attaches the evidence, states which limit or rule triggered the referral, and hands over. The senior reviewer may approve, decline, approve with conditions, or send the case back for more information. Each of those outcomes changes the state of the case and sets different follow-up work.

This matters for model training because the interesting behaviour is in the judgement about when to stop deciding and hand over, and in what to write so the next person can act without re-reading everything. Models that only see final outcomes miss the delegated-authority structure, the handover note, and the exceptions.

For model labs, a dataset or environment built around this workflow can teach and test authority awareness, handover quality, and policy-grounded reasoning. The output MirrorFirm generates is synthetic, shaped by how such workflows operate, and is not a copy of any organisation's records.