Chapter 8 — Key Takeaways
A one-page field card for Information Gathering: Applications, Inspections, Reports, and Building the Risk Picture. The decision is only as good as the information behind it.
The core claims
- You must decide a risk you cannot see. The submission is a few PDFs; the real building, roof, welding floor, drivers, and finances are behind them. Your job in this chapter is to gather the right information — not yet to assess, price, or decide.
- Every source has a purpose AND a limit. Know both for each one, every time.
- Information gathering is where adverse selection attacks first. The applicant and broker curate the submission toward your "yes." Order independent reports and notice what is missing.
- A data source can be statistically valid and ethically contested at the same time. The credit-based insurance score is the emblem: it predicts loss and raises a real fairness/disparate-impact question. Predictiveness is the start of the analysis, never the end.
- The data tells you what is there; the underwriter notices what is not. Reading the silences — the blank field, the missing year, the unmentioned hot-work program — is the part still irreducibly human.
The sources — what each can and cannot tell you
| Source | What it tells you | What it CANNOT tell you |
|---|---|---|
| Application / submission | The risk's self-described exposure; the underwriting answers; the legal representations | Trajectory; verified facts — it is self-reported, a snapshot, and unverified |
| Loss run (5 yrs, all lines) | What actually happened: cause, severity, frequency, open reserves — the story | What it means; it is backward-looking (prior management) and often a thin sample |
| MVR | The official driving record: license status, violations, at-fault accidents | What was done vs. cited — a clean MVR may just mean "not caught" |
| CLUE | Filed property/auto claims across all carriers (catches the carrier-switcher) | Out-of-pocket losses (never filed); fault — a claim ≠ the applicant's fault |
| Credit-based insurance score | A statistically loss-predictive class signal | Anything about this individual; it is a correlation, contested, restricted in some states |
| Inspection report | Verified physical facts; unnoticed hazards; loss-control recommendations | Anything beyond a single day; quality depends on the inspector |
| Financials / public records | Financial health (a leading loss indicator); exposure-base reality; OSHA/liens/news | Audited certainty if unaudited; the future — they show stress before the loss |
The rule of thumb
Predictiveness is the start of the analysis, not the end. A factor that improves loss prediction has cleared one test — not the fairness test, the legality test, or the proxy test.
The compliance line (FCRA)
- Permissible purpose required to pull a consumer report.
- Adverse action (deny, cancel, non-renew, OR charge more) based in whole/part on a consumer report → notice owed. The forgotten case is charging more. Automate it.
- The notice must: say an adverse action was taken; identify the reporting agency; disclose dispute and free-copy rights (and, for credit scores, the key factors).
- State patchwork (McCarran-Ferguson): what you may use varies by state and line. Know your states.
- Bright line: classify by risk; never discriminate by protected class. Beware proxy discrimination (a neutral-looking factor standing in for a protected one).
Key terms
application/submission · motor vehicle report (MVR) · CLUE · credit-based insurance score · loss run · inspection report · third-party data
What you could defend to your manager
"I didn't quote Harbor Steel off the submission as received. I ordered five years of loss runs across all lines, currently valued; an on-site inspection of a \$20M plant with two fires and a 1994 roof; MVRs on all twelve drivers; three years of financials to confirm the revenue and payroll exposure bases; the SOV; and public records. I flagged what's still missing — the causes and corrective actions behind both fires, a written hot-work program, the financial trajectory, and whether the loss runs are complete — and I'm treating the open products claim as a real, uncertain number, not zero. Nothing is graded or priced yet, because I haven't finished seeing the risk."