Skip to content

Workers Comp Loss Runs and the Experience Mod

Workers comp loss runs show the claims that become the unit statistical data behind the experience mod, so reading them early lets you catch errors before the mod is published.

Machine shop floor with CNC machines, a workers compensation account

/8 min read

Loss runs for workers comp list every injury claim under the policy with medical and indemnity amounts paid and reserved. The same claims, reported by the carrier to the rating bureau as unit statistical data, drive the experience modification factor, so reviewing workers comp loss runs early is how brokers catch errors before they raise the insured's premium.

Workers compensation is the line where loss runs have the most direct effect on price. A general liability underwriter uses loss history as one input among many. In workers comp, past losses feed a formula that multiplies the manual premium, and that formula is published. This guide explains what a workers comp loss run shows, how the experience mod uses it, and what to check before renewal. It describes the NCCI approach used in most states; states with independent rating bureaus, such as California, New York and Pennsylvania, use their own plans with similar principles.

What a workers comp loss run shows

A workers comp loss run has the same skeleton as any loss run, with a few fields specific to the line.

FieldWhat it tells you
Claim typeMedical only (no lost wages paid) or lost time (indemnity paid for missed work)
Injured workerUsually masked or initials only
Date of injury and date reportedLate reporting tends to raise claim cost
Injury description and body partShows patterns such as back strains or lacerations
Class codeThe NCCI classification the worker was assigned to
Medical paid and reservedTreatment costs
Indemnity paid and reservedWage replacement benefits
Expense paid and reservedAdjusting and legal costs, usually not included in the mod
StatusOpen, closed or reopened

If the fields themselves are new to you, start with what a loss run report is and how to read one.

How the experience mod works

The experience modification factor, usually called the mod or EMR, compares an employer's actual losses with the losses expected for an employer of the same size in the same classifications. A mod of 1.00 is average. A mod of 0.85 is a 15 percent credit on premium, and a mod of 1.25 is a 25 percent debit.

The experience period

Under NCCI rules the mod generally uses three completed policy years of data, ending one year before the rating effective date. For a mod effective January 1, 2027, the experience period is typically the policies that started in 2023, 2024 and 2025. The most recent completed year is left out because its claims are still too immature to value reliably.

Unit statistical data

The mod is not calculated from the loss run itself. Carriers report each policy's payroll and claims to the bureau as unit statistical data. The first report is valued 18 months after the policy effective date, with later reports at 12-month intervals while claims remain open. The claim values in those reports are what the mod uses. A loss run lets you see the same claims, often before or after the valuation the bureau uses, which is why it is the best tool for spotting problems in advance.

Primary and excess losses

NCCI splits each claim into a primary portion, up to a dollar threshold called the split point, and an excess portion above it. Primary losses count in full. Excess losses are weighted down and blended with expected excess losses, with smaller employers getting less weight on their own experience. The split point is set and adjusted by NCCI over time, so check the current figure for the rating year you are reviewing.

The effect is that frequency matters more than severity. Five claims of $10,000 each usually hurt a mod more than one claim of $50,000, because more of the total sits below the split point.

Claim pattern (fictional)Total incurredPortion counted in fullTypical mod impact
One lost time claim$50,000Only up to the split pointModerate
Five lost time claims of $10,000$50,000Nearly all of it, if the split point is above $10,000Higher
Ten medical only claims of $1,500$15,000Reduced further in states using ERALower

The experience rating adjustment

Most NCCI states apply the Experience Rating Adjustment, which counts only 30 percent of medical only claims in the mod. It rewards employers who report minor injuries promptly and keep them from becoming lost time claims. Whether a claim is coded medical only or lost time therefore matters, and a miscoded claim is one of the errors worth looking for.

What to check on workers comp loss runs before renewal

The window to fix an error is before the bureau publishes the mod. Review loss runs for the experience period with these questions.

  1. Are closed claims carrying reserves? A claim that closed with reserves still attached inflates incurred losses.
  2. Are claims coded correctly? A medical only claim reported as lost time loses the ERA reduction.
  3. Are subrogation recoveries reflected? If the carrier recovered from a third party, incurred should drop.
  4. Are claims assigned to the right employer and policy? Mistakes happen with related entities and mid-term changes.
  5. Are reserves reasonable? A large reserve on a claim where the worker returned to work months ago is worth a question to the adjuster.
  6. Are there duplicates? The same injury reported twice counts twice.
  7. Is the payroll right? Payroll is outside the loss run, but it sets expected losses. Compare audited payroll with the experience rating worksheet.

When you find an error, raise it with the carrier first, since the carrier is the one that reports corrected unit statistical data to the bureau. Do this well before the rating date, because corrections take time to flow through.

Using loss runs to forecast next year's mod

Because the experience period rolls forward each year, you can see next year's mod coming. When the next mod is calculated, the oldest year drops out and the next completed year comes in. Compare the two years on the loss runs. If a heavy year is dropping out and a clean year is coming in, the mod should improve. If it is the other way around, warn the client early and start the loss control conversation now, not at renewal.

The same reading tells you which open claims to watch. An open lost time claim in a year that will be in the experience period is worth managing closely, because its value at the next unit statistical valuation is what the mod will use.

Workers comp loss runs in the submission

For the submission itself, markets want five years plus the current term of workers comp loss runs, the current mod worksheet, and payroll by class code. Present the loss summary by policy year with claim counts split into medical only and lost time, and medical, indemnity and expense shown separately. Our workers comp loss runs page goes into the submission layout, and how many years of loss runs you need covers the history requirement.

LossRunReports reads workers comp loss runs in PDF, scanned, XLSX and CSV formats into one claims schema, keeps medical, indemnity and expense separate, labels claims as medical only or lost time, and flags closed claims that still carry reserves, duplicates and valuations older than 90 days. The five-year summary with loss ratio is built automatically. You can see how it handles one of your own reports with analyze a loss run, or read more about our loss run software.

Summary

  • Workers comp loss runs separate medical, indemnity and expense, and medical only from lost time.
  • The mod generally uses three completed years ending one year before the rating date.
  • Primary losses below the split point count in full, so frequency weighs heavily.
  • Check reserves on closed claims, claim type coding, recoveries and duplicates before the mod is published.
  • Compare the year dropping out with the year coming in to forecast the next mod.

See your own loss run read in under a minute

Upload one carrier report or use our sample. No account needed for the first read.