How Your EMR Is Actually Calculated

Your experience modification rate (EMR, or e-mod) is the multiplier applied to your workers' compensation manual premium. A 1.00 is exactly average for your class of business. A 0.85 means you pay 15% less than average; a 1.25 means you pay 25% more. General contractors routinely require a sub-1.00 EMR to bid, so for many companies the number is not just a cost — it's a gate on revenue.

The three-year window (and the year that isn't in it)

EMR is calculated from a rolling three-year experience period that excludes the most recent policy year. If your rating date is January 2026, the experience period is roughly 2022–2024. That lag has two consequences people consistently get wrong:

• A bad claim year keeps hurting you for three more rating periods, long after the program has been fixed.
• Improvements you make today don't show up in the number for over a year. This is why 'we'll fix safety when the rate goes up' is always too late.

Expected losses vs. actual losses

The rating bureau (NCCI in most states; independent bureaus in states like California, Pennsylvania, and Michigan) calculates what a company of your size and class is expected to lose, using your payroll by class code times published expected loss rates. Your actual losses are then compared to that expectation. Simplified, the formula is:

The ballast and stabilizing values are credibility adjustments — they keep a single bad claim from wrecking a small employer's rate, and they make a large employer's number more responsive to actual experience. The practical takeaway is that larger payrolls have more 'credibility,' so their EMR moves further, faster, in both directions.

Primary vs. excess losses — why frequency is the killer

Each claim is split at a primary loss threshold (commonly around $15,000–$19,000 depending on state and year). The portion below the split counts at full weight. The portion above it is 'excess' and is heavily discounted in the formula.

This is the single most misunderstood fact about EMR. The rating system is explicitly designed to punish frequency, because frequency is what predicts future severity. A company with ten strains and sprains a year has a worse e-mod than a company with one catastrophic but genuinely freak accident.

What counts as a loss

• Incurred losses — paid amounts plus reserves. Open reserves count at their reserved value, not the eventual settlement.
• Medical-only claims are typically discounted (often to 30% of value) under ERA rules in most states — one reason keeping a claim medical-only matters so much.
• Indemnity (lost-time) claims count at full value and drag the number hardest.

The four levers that actually move it

1. Reduce frequency. Target the small recurring claims — strains, slips, lacerations, eye injuries — not just the catastrophic scenarios.
2. Keep claims medical-only. A functioning return-to-work / modified-duty program converts indemnity claims into medical-only claims, which are discounted.
3. Manage reserves. Review the loss run quarterly, challenge stale reserves, and close what should be closed before the filing date.
4. Verify class codes and payroll. Misallocated payroll to a higher-rated class inflates both premium and expected losses. Audit it annually.

A realistic timeline

Because of the lag, a serious program overhaul typically shows a first measurable improvement at the second rating period after it starts, and a full effect at the third or fourth. Anyone promising a lower EMR this quarter is describing a reserve correction or a payroll reclassification, not a safety improvement.