Insurance Is the Only Party Measuring Availability

This desk has filed repeatedly that availability harm is unmeasured: no notification, no register, no statistic — at 26-0209, at 25-0902, at 25-0606. That is true of the public record. It is not true absolutely.

A business-interruption claim requires the insured to establish what stopped, for how long, and what it cost, in a form an adjuster will accept and a court could review. That is a more rigorous quantification of availability harm than anything in the public domain.

The insurer then aggregates across its book, prices next year’s premiums on it, and reinsures the tail. Somewhere in that process sits the closest thing that exists to a real dataset on what outages cost.

The data is proprietary, and it is the basis of competitive advantage in underwriting. It covers only insured organisations — excluding the uninsured, the self-insured, and the public bodies at 25-0918 that could not buy cover if they wanted to.

And it is filtered by claims behaviour rather than by incidence: losses below deductible do not appear, and organisations that fear a premium increase may not claim at all. It is a measurement of insured, claimed, accepted loss, which is a specific and non-random subset.

The reinsurer analysis this desk cited at 25-0728 is drawn from precisely this data, and it is genuinely among the more useful material in the corpus for exactly that reason.

The observation to hold onto is that an entire category of harm — the category containing the largest incidents in this database — is measured only as a byproduct of a commercial risk-transfer market. Graded medium: the structure is clear, and this desk cannot see inside the data to characterise it.