Your Frequency Is Down. Your EMR Isn't. Here's Why.

Fewer workers got hurt on your jobsite last year. You know it. You can feel it in the claim count. And yet your mod hasn't moved the way you expected.

NCI reported at its Annual Issues Symposium in May 2026 that lost-time claim frequency fell 2% in 2025, below the historical trend of 3 to 4% annual declines. Medical and indemnity severity rose 4% (NCCI AIS 2026, May 2026). The ones that do file claims are costing more. This is the pattern that makes the frequency-severity story counterintuitive for contractors trying to read their mod.

How the NCCI formula splits losses

The Experience Modification Rate (EMR) formula doesn't treat all losses equally. NCCI divides each claim's cost into two buckets: primary losses and excess losses. The split point varies by state and year, but currently sits in the range of $17,000 to $18,500 for most NCCI-administered states.

Primary losses are the portion of each claim's cost up to the split point. They enter the formula dollar-for-dollar. Excess losses, the amount above the split point on any individual claim, are also included, but they're weighted down by a factor called the D ratio. The rationale: primary losses reflect a contractor's ongoing safety environment more accurately than catastrophic individual events, which can happen even to well-run operations.

Expected losses are calculated the same way, splitting into expected primary and expected excess based on the ELR for each classification and the split point. The mod is roughly the ratio of actual primary to expected primary, adjusted for excess. Primary losses drive most of the movement.

Why severity pushes the mod up even when claim count falls

Here's where the divergence matters. Suppose a contractor had 10 claims in a policy year, each costing $8,000. Total actual losses: $80,000. All 10 claims are below the split point, so all $80,000 is primary losses, dollar-for-dollar in the numerator.

Now suppose frequency falls: the same contractor has 7 claims. But severity rose, and each now costs $13,000. Total losses: $91,000. Still all primary losses. Claim count dropped 30%. Actual primary losses went up.

The mod formula compares actual primary losses to expected primary losses for that contractor's payroll and classifications. If expected primary losses fell with the industry frequency trend but actual primary losses rose because of severity, the ratio moves in the wrong direction. Fewer claims, worse mod.

It gets more complicated when claims cross the split point. A claim that previously cost $14,000, all primary, now costs $22,000 because severity rose. Primary losses: $17,500 (capped at split). Excess losses: $4,500, discounted by the D ratio. The claim's contribution to actual primary losses rose from $14,000 to $17,500. Expected primary losses for that classification may have fallen as ELRs adjusted to the industry's declining frequency trend. The ratio worsens even though the contractor had fewer total claims.

The reserve timing problem

Open reserves make this worse. A severe claim that remains open at 18 months, which is more common with high-severity cases, sits on the worksheet at the carrier's estimated final cost, not the amount paid to date. If severity is running 4% above historical expectations, carriers are setting reserves on open cases to reflect that. Those reserves inflate actual losses on the worksheet before the claim is even closed.

In our reviews of Southeast contractor worksheets, the most common pattern for contractors who expected their mod to fall and it didn't is exactly this: fewer claims, but one or two open claims with reserves set above what the current treatment trajectory would justify. The claim count is down. The reserved cost on the remaining open claims is up, and the mod reflects it.

The ELR timing mismatch

There's a secondary mechanic worth knowing. NCCI updates Expected Loss Rates periodically based on industry claim experience. When frequency has been declining at 3 to 4% per year for years, ELRs adjust downward to reflect that lower expected claim frequency. Lower expected primary losses in the denominator means a contractor needs fewer actual primary losses to maintain the same mod.

When frequency then slows to 2%, as it did in 2025, ELRs may have been set anticipating continued rapid frequency improvement that didn't fully materialize. The expected losses assume a lower claim rate than what actually occurred. That timing mismatch can show up in the mod before NCCI recalibrates.

What an audit would check

An audit examines open reserves on claims in the experience period against the specific injury, treatment history, and current prognosis, rather than a trend-indexed estimate set at claim inception and never updated. It checks whether any claims have closed below their reserved amount, creating favorable development that isn't yet reflected on the current worksheet. When frequency falls and severity rises, the audit focus shifts from claim count to the accuracy of what each remaining claim is carrying, and to whether the ELRs applied to the worksheet match the current published NCCI rates for each classification and state. Both inputs matter when the two trends are moving in opposite directions.

If your claim count has been falling but your mod hasn't followed, send us your NCCI worksheet and we'll show you where the formula diverges from the count.