NCCI presented the CY2025 State of the Line in Orlando on May 12, 2026. Private carriers posted a 91% calendar-year combined ratio, five points worse than CY2024's 86%, extending the line's underwriting profit streak to twelve years. The accident-year combined ratio was 102%. The industry did not earn an underwriting profit on the business it wrote in 2025. It earned one on reserves booked years earlier.

Key Takeaways

  • 91% calendar year against 102% accident year. Current-year business is unprofitable before investment income; the calendar-year result is carried by development on older accident years.
  • Frequency improvement fell to 2% from 5% in CY2024, against a long-run range of 3 to 5%, which removes most of the offset that has been holding pure premium trend near zero.
  • Medical and indemnity severity both ran at 4%, so the two components stack rather than partially cancel as they have historically.
  • Reserve redundancy fell to roughly $14 billion from $16 billion, a $2 billion drawdown in a single year and the buffer the calendar-year result is drawn against.
  • Net written premium contracted 0.2%, the first decline since the line recovered from the 2008 to 2012 soft market, while rate decreases continue to be filed.

What the Two Combined Ratios Measure

Calendar year captures everything recognized in the accounting period, including development on prior accident years. Accident year strips that out and reports only the estimated ultimate cost of claims from the year in question.

At 102%, private carriers expect to pay roughly $1.02 in losses and expenses for every premium dollar written on 2025 policy years, before investment income. At 91%, the reported result is nine points better, and the difference is prior-year development credit.

An accident-year result running above calendar year is not itself unusual in a line drawing down redundancy. What makes CY2025 a different reading is the combination underneath it:

MetricCY2024CY2025Direction
Calendar-year combined ratio86%91%Worse (5 pts)
Accident-year combined ratio~98%102%Worse (>100)
Net written premium growthPositive-0.2%Contraction
Lost-time frequency trend-5%-2%Decelerating
Medical severity trend+6%+4%Moderating but elevated
Indemnity severity trend+4%+4%Stable, elevated
Reserve redundancy estimate~$16B~$14BDrawdown (-$2B)

Every line in that table moved the same way. Frequency improvement slowed, both severity components stayed elevated, premium contracted, and the redundancy stock funding the calendar-year advantage shrank by $2 billion.

The Frequency Offset Is What Actually Changed

Frequency has done the arithmetic work in workers compensation pricing for fifteen years. CY2025 is where it stopped.

Lost-time claim frequency declined 2%, against 5% in CY2024 and a long-run range of 3 to 5%. Set that beside severity. With medical severity at 4% and frequency improving at 4%, net pure premium trend sits near zero, which is how carriers filed flat or declining rates through a profitable decade. With the same 4% severity and only 2% frequency improvement, net pure premium trend is roughly 2%.

That is the specific recalibration the release forces. A trend selection fitted to CY2022 through CY2024 experience, where frequency was improving at 4 to 5%, overstates the offset and understates the indicated rate change. The correction is not a judgment call about the future; it is what the CY2025 data already says about the most recent year.

The second change compounds it. Medical and indemnity severity both ran at 4%. These components are driven by different things: medical by healthcare unit cost, utilization and pharmacy trends, indemnity by wage growth against statutory maxima. They have historically diverged, so softness in one partially absorbed hardness in the other. Running at the same rate, they stack. A blended severity selection that assumes partial cancellation is assuming a relationship that did not hold this year.

Medical at 4% is in fact moderating, down from 6% in CY2024, and indemnity at 4% reflects post-pandemic wage growth decelerating toward its 3 to 4% range. Neither is accelerating. The problem is that neither is falling to meet a frequency trend that is no longer moving.

The Buffer the Calendar Year Is Drawn Against

Redundancy is the reason CY and AY can differ by nine points, and it is finite.

Carriers built reserves through roughly 2009 to 2014, when frequency fell faster than trend models expected and medical inflation came in below assumption. Those reserves proved more than adequate, and the releases have flowed into calendar-year results ever since. NCCI now estimates roughly $14 billion of industry redundancy, down from $16 billion.

The last time this line ran out of buffer, the correction was not gradual. The cycle from roughly 1997 to 2003 produced a reserve deficiency that took years of rate increases and strengthening to resolve, and the 2000 to 2004 period shows the shape: modest erosion first, then a sharper strengthening wave.

Two billion dollars is not a large number against a multi-hundred-billion-dollar reserve base. The arithmetic of the drawdown is what matters: at that annual pace the buffer runs about seven years, and the releases must shrink as the stock does. If AY2025 and the years after it develop worse than initially estimated, the prior-year development line stops being a credit and becomes a charge. Calendar-year combined ratios would then deteriorate faster than the accident-year trend alone implies, because the gap that has been flattering the reported result inverts.

The exposure side is moving against that at the same time. Net written premium fell 0.2% while wages ran above long-term averages and employment held broadly stable through 2025, which points at rate compression rather than exposure loss. Carriers are filing decreases earned on past redundancy while the accident year runs above 100 and the redundancy that justified the decreases is being consumed.

The twelve-year streak is intact and was genuinely earned. It is now being funded from a stock that is $2 billion smaller than it was a year ago, against current-year business that does not pay for itself.

Further Reading

Sources