Setting a premium normally starts with historical claims. For a risk that has no history, insurers construct an estimate from components and price the uncertainty explicitly.
Exposure analysis replaces claims data
Without loss records, underwriters begin by describing what could go wrong: what is at stake, what sequence of events would cause a loss, and how large it could be.
Engineering assessments, comparable industries and near-miss reports all provide inputs, even where no insured loss has occurred in that specific form.
The result is a structured estimate of frequency and severity built from parts, rather than a figure extracted from a table of past outcomes.
Analogous risks are used where they exist, so cover for a new technology may be priced initially from the loss patterns of the closest established equivalent.
Models turn assumptions into distributions
Catastrophe and scenario models simulate large numbers of possible years, producing a distribution of losses rather than a single expected value.
Insurers price against points in that distribution, since the capital they must hold depends on the tail rather than on the average outcome.
Because the model rests on assumptions rather than observations, the width of the distribution reflects genuine uncertainty about the assumptions themselves.
Uncertainty is charged for directly
Where the expected loss is poorly known, the premium includes a loading above the modelled estimate to compensate for the possibility that the estimate is wrong.
This is why cover for emerging risks looks expensive relative to the losses that eventually occur, at least in the early years.
Policy terms carry the same caution, with lower limits, tighter definitions and specific exclusions narrowing what the insurer has actually agreed to pay.
Structure limits exposure while data accumulates
Insurers often write small lines across many accounts rather than large lines on a few, which keeps any single misjudgement survivable.
Aggregate limits cap total payouts across a portfolio, and sunset clauses restrict how long claims can be reported after a policy period ends.
These devices allow participation in an unfamiliar risk without committing capital that a modelling error could exhaust.
Prices correct as experience arrives
Every year of claims data narrows the uncertainty, and premiums adjust as the estimate firms up in either direction.
Competition accelerates the process, since insurers with better data can price more finely and win the business that has been overcharged.
Cyber cover followed this path, moving from broad and cheap to restricted and expensive as losses emerged, then towards finer differentiation as understanding improved.
The pattern is consistent enough that an unusually cheap price for an unfamiliar risk is generally a sign the loss experience has not yet arrived rather than that the risk is small.