The Minnesota Department of Labor and Industry published its annual Workers’ Compensation System Report on August 21, 2026. The headline reads well: total paid claims fell 42% relative to full-time-equivalent workers between 2004 and 2024, and total system cost reached $0.83 per $100 of payroll, the lowest in the series’ history.
Self-insured employers should look past that headline at the per-claim data.
Over the same 20 years, average TTD duration per claim rose 33%. Total disability benefits per paid indemnity claim rose 37%. The system produces far fewer claims today, but each claim that enters it runs longer and costs more than its 2004 counterpart. Minnesota’s WC system did not get cheaper; it got smaller and denser.
Who It Affects
Self-insured employers in manufacturing, distribution, healthcare, and construction carry the most long-tail WC exposure and the most open lost-time claims being reserved against this evolving claim population. Public-entity pools and captives writing WC face the same diagnostic question: are the development factors in current ECR calculations calibrated to a claim mix that no longer exists?
The Reserve Mechanism
The problem is the shape of the development pattern, not just its level.
Chain-ladder methods generate link ratios by observing how paid and reported losses age from one maturity to the next. Those ratios reflect the population of claims that produced the historical triangle. When the population was dominated by higher-frequency, shorter-duration claims, development was faster: most claims resolved by 36 or 48 months, tails were modest, and a typical factor at 60-to-ultimate might be 1.03 or 1.04.
Today’s claim population has a different profile. Fewer claims enter the system, but they are weighted toward longer-duration cases with extended medical treatment and slower return-to-work resolution. High-severity claims carry fatter tails and longer expense lags than the moderate-severity claims they replaced in the mix. The development pattern that fits a 2024-era claim population should be wider and slower than one fitted to 2010.
The problem is that actuaries selecting link ratios draw on triangles populated largely by 2010-era claims. Those factors, applied to today’s smaller and more severe claim population, are solving for the wrong claim shape. IBNR will emerge more slowly than the factors imply, and the ultimate will be higher than a straight-line projection of historical development suggests.
NCCI’s 2026 State of the Line confirms the pattern is national: claim frequency has been declining at roughly 2% per year while indemnity severity rose 4%, a sustained divergence that reshapes every development triangle built on the pre-2020 claim mix.
Where This Shows Up in Your Reserves
Look at paid development factors for accident years beyond 48 months. A factor that has been compressing steadily, reflecting faster emergence in a high-frequency period, is worth flagging. If that selection is applied to a recent accident year where fewer but heavier claims are still open, it will underestimate remaining development. The Berquist-Sherman diagnostic tests for exactly this type of population shift: when claim handling or exposure mix changes, factors from the prior period can systematically misstate the emerging development curve.
What This Means for Your Next Review
Two questions for your actuary before the next reserve opinion:
- Were the link ratios in the current analysis selected from a period when Minnesota or portfolio-level claim frequency was materially higher? If the calibration window extends back to 2008-2015, the selected factors reflect a claim mix that is no longer representative.
- Have 2022 through 2024 accident years been tested against a shorter, recent calibration window? A five-year lookback focused on 2020-2024 experience gives development factors shaped by a claim population much closer to today’s.
Minnesota’s data are public and specific. The 33% increase in TTD duration per claim over 20 years is an auditable measure of how differently today’s open claims will develop compared to the historical baseline embedded in most self-insured triangles.