The Actuarial Twin: Underwrite Health with Causes, Not Correlations

The thesis

Actuarial models run on correlation — and correlation quietly launders historical bias into future prices while telling you nothing about what to do. Knowing that a ZIP code predicts cost is not knowledge; it is prejudice with a p-value.

Move to causal twins

Build a causal digital twin of the member: a model that estimates how outcomes and costs would change under intervention — if this patient started this medication, joined this program, saw this specialist. Now the payer can price risk and act on it, because the model answers "what happens if we help?" not merely "who looks expensive?".

Why causal beats correlational

• Interventions become visible. The twin ranks the actions that most reduce future risk for this specific person.
• Fairness is testable. Causal models let you audit whether a variable is doing work through a legitimate pathway or laundering a protected attribute.
• Trust. You can explain a decision as a mechanism, not a black-box score.

The obligation

With causal power comes accountability: publish the graph, test for disparate impact, and let regulators inspect the assumptions. Opacity is no longer defensible when the tools for transparency exist.

Provocation

An industry that prices human risk owes society more than a correlation. Causal underwriting is how insurance earns back the right to be trusted with that job.

A first-principles provocation from the Anxya Health Futures desk. Directional and informational — not medical, legal, financial or regulatory advice. The point is to move the debate, then do the hard validation work.