What insurance contracts can reveal about risk aversion

The crucial comparison is alignment

The researchers call their approach “risk–insurance parity”: an equivalence between a particular form of risk aversion and a preference for insurance across a specified collection of contracts.

Imagine a firm choosing between coverage tied directly to its own losses and a payment tied to another index. Both options cost the same and have the same distribution of possible payments. What differs is when the money arrives. Properly aligned coverage pays more when the firm's losses are greater; the alternative need not.

The paper calls a consistent preference for the aligned option “insurance propensity.” This is not a preference for buying insurance rather than remaining uninsured. It compares two payment arrangements, holding their individual payout distributions fixed while changing their dependence on the firm's loss.

That distinction allows the analysis to avoid requiring an actuarially fair premium—the average expected payout. It does not establish whether either option is worth buying.

Two meanings of disliking risk

Weak risk aversion means preferring a certain outcome to an uncertain one with the same average. Strong risk aversion goes further: it means preferring less dispersion even when both alternatives remain uncertain and have the same average.

The mathematical idea behind the latter is a “mean-preserving spread.” Some outcomes become better and others worse, with the changes balancing so that the average stays unchanged. A strongly risk-averse decision-maker prefers the situation before that extra spread.

These definitions coincide within standard expected utility theory, but can differ in other models. The authors therefore work with a broader framework rather than assuming one particular utility formula.

What one contract really establishes

The headline result concerns proportional coverage: insurance that reimburses a fixed fraction of a loss, greater than zero but less than the whole amount.

The authors prove that one such contract, with a fixed fraction and a fixed premium, is sufficient to characterize strong risk aversion. Earlier results discussed in the paper required varying both the coverage fraction and the premium across entire families of proportional contracts.

The qualification is substantial. “One contract” means one fixed contract rule applied across every possible underlying risk. The decision-maker must consistently prefer its aligned payments over every alternative arrangement with the same payout distribution. One observed choice—or one policy purchase—does not establish the result.

For weak risk aversion, full coverage behaves differently. With fixed-price contracts, the characterization requires full-indemnity contracts whose premiums can approximate every real number. A single full-coverage pricing rule can suffice if its price varies with the risk distribution and meets the paper's stated range condition.

How the results are obtained

This is mathematical research, not a survey, experiment or analysis of insurance customers. Its results are equivalence theorems supported by proofs and counterexamples, not measured effects.

The initial analysis uses risks with finitely many possible outcomes. Preferences depend on outcome probabilities rather than the identities of events, are transitive, and satisfy a continuity condition. Continuity lets conclusions survive suitable limiting approximations. The framework even permits a decision-maker to leave some alternatives unranked.

A key proof technique rearranges losses between equally probable events. Comparing aligned and rearranged payments produces reductions in dispersion without changing the average. The authors show how these comparisons establish the broader risk-aversion property. They also extend their main results to wider classes of random losses under additional technical conditions.

Deductibles reveal a different middle ground

Deductible-only and payout-limit-only contracts lead to two proposed intermediate notions of risk aversion, each stronger than weak risk aversion but weaker than strong risk aversion.

“Right-handle” risk aversion concerns spreads that worsen an existing maximum loss while balancing that change with improvements elsewhere. It corresponds to insurance propensity across suitably extensive deductible-only contract sets. “Left-handle” risk aversion concerns spreads extending an existing minimum loss, with balancing changes elsewhere, and corresponds to suitably extensive limit-only contract sets.

These are precise restrictions on preferences, not evidence that customers fall into newly discovered psychological categories. Within expected utility theory, the distinctions collapse.

What remains untested

The supplied source is arXiv preprint 2512.09208v2, posted on November 9, 2026; the manuscript labels its current version November 10. The evidence does not establish peer-reviewed publication.

The practical promise is a simpler theoretical connection between contract design and risk preferences. Whether finite, noisy observations can reliably identify those preferences remains untested here. The framework also allows negative losses and contract parameters to cover general financial outcomes, so its mathematical comparisons extend beyond ordinary retail policies.