Chapter 14 — Key Takeaways

A one-page field card for personal-auto underwriting. Pair it with the rating-factor logic in §14.2 and the combined-ratio discipline in §14.7.

The core claims

  • Personal auto is the most-purchased, most-regulated, least-reliably-profitable line. Roughly half of P&C premium is personal lines, and auto is the largest slice. The underwriting question is usually "what does this risk cost, and may I legally charge it?" — not "should I write it?"
  • The premium is not one number; it is the sum of priced coverage parts. Liability and UM/UIM are third-party, people-driven (injury + litigation environment); collision and comprehensive are first-party, car-driven. Different parts → different factors → different price movements.
  • The driving record is a real but weak predictor. Most drivers — including most who will crash next year — have clean records. The strongest factors (territory, prior-insurance history, and where permitted credit) have nothing to do with how the applicant drives. That is the chapter's title.
  • Rating is multivariate. Factors are estimated holding the others constant (the GLM, Chapter 32); univariate intuition ("that factor's too expensive") is usually the thing that's wrong.
  • Credit-based insurance scores are among the strongest predictors and the most contested. Predictive (validated against loss) yet entangled with income and, through history, race → the proxy-discrimination problem (Chapter 35). Heavily restricted, state by state; FCRA requires adverse-action notice.
  • Telematics/UBI price measured behavior, not proxies. They rescue good risks the static factors mislabel and turn adverse selection in the carrier's favor (good drivers self-select in) — but they measure a window, raise privacy questions, and can re-create territory/demographic proxies.
  • The line is regulated factor by factor, state by state. California Prop 103 (record/miles/experience prioritized; credit & territory restricted); Michigan no-fault (the mandate drove price). Society negotiates actuarial-vs-social fairness here, in public.
  • Nonstandard auto is where adverse selection bites hardest. Required drivers who can't get standard coverage; the craft is classifying within a wide distribution; the residual market is the backstop.
  • The combined ratio is the verdict. Loss costs (medical, repair, litigation) trend up continuously; rate increases lag approval and competition is fierce → the line rides 100%. Discipline, not growth, is what survives it.

The rule of thumb

Rate the household into the plan; price the trend, not the moment; and never let a factor that is banned be mistaken for not predictive, or a factor that is predictive be mistaken for permitted.

The key calculation

A multiplicative rate: Premium = base rate × Π(filed relativities). Then judge the book by the combined ratio = loss ratio + expense ratio; above 100% is an underwriting loss before investment income. (Worked in §14.2 and §14.7.)

The key terms

personal-auto policy (PAP) · driver class · vehicle symbol · rating territory · usage-based insurance (UBI) · telematics · nonstandard auto

What you could defend to your manager

"I rated this household to the filed plan — driver class, vehicle symbol, territory, use, prior insurance, and, as our state permits, the credit-based insurance score, with the required adverse-action notice on file. The clean driving record is the factor the applicant fixates on, but the price is built mostly from factors that don't touch the record, and every one of them is permitted in this state as filed. We're offering the telematics option, which could lower the price for the careful young driver in the household and helps us win the good risks our competitors can only see as 'young.' The rate is set to the forward loss trend, not last quarter — because in this line, the carriers that chase volume by holding rate down are the ones whose combined ratio blows up two years later."