An insurance premium is not a forecast about the person paying it. It is the expected cost of the group that person has been sorted into, plus expenses and a margin, which is why identical-seeming applicants are quoted differently.
Pooling only works if the groups are homogeneous
Insurance functions by collecting predictable amounts from many policyholders to pay unpredictable amounts to a few. That arithmetic requires the group's aggregate losses to be estimable in advance.
If a pool mixes very different risks at one price, the lower risks are overcharged relative to their expected cost and the higher risks are undercharged for theirs.
The overcharged group tends to shop elsewhere or go without cover, leaving the pool worse than it started. Underwriting exists to prevent that unravelling by sorting before pricing.
Rating factors are chosen for predictive power
Insurers analyse historical claims against applicant characteristics and retain the variables that correlate reliably with loss cost across large populations.
Some are causally obvious, such as the construction and age of a building or the driving record attached to a licence. Others are statistically robust without an obvious mechanism behind them.
Because they are selected for correlation rather than explanation, several are contested and some are restricted or banned outright by regulators in particular jurisdictions.
Individual experience is credibility-weighted
A single policyholder generates too few claims for their own history to be statistically meaningful on its own, particularly in personal lines.
Insurers therefore blend individual experience with the class average, weighting the individual portion according to how much data supports it.
That is why one claim can move a premium noticeably for a small account while a large commercial account is rated much more heavily on its own record.
Information asymmetry drives the questions asked
The applicant knows more about their own risk than the insurer does, and people who expect to claim have more reason to buy cover.
Application questions, medical evidence, inspections and waiting periods all exist to narrow that gap before the insurer commits to a price.
Misrepresentation provisions sit behind them, allowing the insurer to adjust or void cover where material information was withheld, which is what makes the questions enforceable.
Filed rates constrain what an underwriter can do
In most regulated markets, insurers must file their rating plans with the supervisor, and prices have to follow the filed formula rather than individual judgement.
The underwriter's discretion therefore lies mainly in whether to accept the risk at all, and in which of the insurer's several rating tiers it belongs.
Declining an applicant and pricing them accurately are commercially similar decisions, which is why availability tightens in regions where regulators hold rates below expected losses.