Loss run forensics: seven checks before you submit an account to market

A loss run is the first thing an underwriter reads and the last thing most brokers do. The standard move is to request it, attach it, and let the carrier's numbers speak for themselves. But carrier loss runs are an adjuster's working file, not an audited statement — and the errors in them almost always run against the insured, because an over-reserved claim costs the adjuster nothing and costs your client real premium.

These are the seven checks we run on every loss run before an account goes to market. Each one ends in the same place: a dollar figure, because a flag without a dollar value doesn't get acted on.

1. Reserve outliers on open claims

Sort open claims by total incurred and compare each reserve to the paid-to-date and the injury description. A soft-tissue strain, two years old, $4,000 paid, reserved at $45,000 — that pattern is the single most common finding. The reserve may reflect a worst case the adjuster set on day one and never revisited.

Why it matters in dollars: incurred losses (paid plus reserve) are what enter the mod. On the worked account in our mod-calculation post, letting four open claims develop 25% over current reserves moves the mod from 0.975 to 1.008; at +50% it reaches 1.041. Reserve movement of that size on a mid-size account is a five-figure premium swing — which is also the business case for a written reserve-challenge letter to the adjuster, with the claim-by-claim math attached.

2. Claims that should be closed

Any claim with no payment activity in 12+ months and a modest remaining reserve is a closure candidate. A claim closed before the unit-stat valuation date enters the mod at its final cost; one left open enters at its reserve. The question for the adjuster is simple: what is preventing closure, and what would resolve it before the valuation date?

3. Coding to the wrong policy year or entity

Match every claim's date of injury against the policy period it's listed under, and the named insured against your client's actual entities. Acquisitions, FEIN changes, and PEO transitions produce claims that belong to a different entity's experience — or to a policy year outside the rating window. A claim coded one year off can be the difference between it counting in the mod and aging out of it.

4. Subrogation and recovery credits

If a third party caused the injury — a motor-vehicle accident is the classic case — ask whether the carrier pursued subrogation and whether recoveries were credited against incurred. A recovery that reduced the carrier's cost but never flowed back into the loss run inflates the experience the next carrier prices.

5. The description-to-class-code match

Read the injury descriptions against the class codes charged. A "fall from dock" coded to a clerical class, or warehouse injuries on an account that reports mostly office payroll, tells an underwriter one of two things — misclassification or incomplete payroll reporting — and both get priced against the account. Better that you find it first. (This is its own discipline for staffing books; see class code drift.)

6. Frequency patterns worth a narrative

Cluster the claims: by body part, by cause, by location, by tenure at injury. Three lumbar strains in one warehouse in eight months is a story an underwriter will build with or without you. If your submission names the pattern and documents the response — the lifting-equipment purchase, the job-rotation change, the supervisor training — the pattern becomes evidence of management rather than evidence of risk. This narrative is what schedule credits are actually granted on.

7. The valuation date, against the clock

Every loss run has a valuation date. Check it against the unit-stat reporting date for the expiring policy year — the data that will set next year's mod is extracted roughly six months after the policy year ends. Reserve challenges, closures, and coding fixes that land after that date help next year's picture, not this one's. Pull loss runs early enough that the fixes you win actually make it into the calculation.

The submission that results

An underwriter in a tight market triages submissions by completeness and credibility. A loss run forwarded raw says the broker hasn't read it. A submission that arrives with reserves challenged in writing, closures requested, coding corrected, and the frequency story told — that account gets quoted first, and the quote reflects the corrected experience rather than the adjuster's working file.

HALO IQ parses carrier loss runs and runs these checks automatically — every open claim's mod impact in dollars, every coding mismatch flagged. See it on your own book.