The National Academy of Social Insurance released its 28th annual workers’ compensation report on August 11, 2026, showing employer WC costs fell to $0.98 per $100 of covered payroll in 2023, the first sub-$1.00 reading in the modern data series. Total system-wide WC benefits reached $64.1 billion that year. Self-insured employers paid approximately $16.4 billion of that total, accounting for 25.6% of all benefits paid.
The headline read like good news. Forty-three of 50 jurisdictions saw declining employer costs between 2022 and 2023; Idaho led with a 17% decline. Nevada was the clearest outlier, posting a 3.1% increase across cash and medical benefits.
The Benchmark Problem
The three-year publication lag means the most favorable WC cost environment on record is now the baseline that actuaries and risk managers will cite in 2026 reserve discussions.
That matters because system-level cost-per-payroll figures from NASI feed directly into the expected loss pick used in Bornhuetter-Ferguson (BF) reserve methods. The BF method blends developed triangle data with an a priori expected loss ratio. When that a priori pick is calibrated to NASI benchmark data, the pick reflects the cost conditions of the data year, not the accident year being reserved.
Current severity conditions are substantially different from 2023. NCCI and WCRI data through 2025 show lost-time indemnity severity approaching $30,000, a level near a 20-year high. Both indemnity and medical components have been rising at roughly 6% annually since 2022. Medical costs have crossed 50% of total WC claim expenditure, a structural shift the 2023 system data does not capture.
Who It Affects
Self-insured employers in manufacturing, construction, healthcare, and public entities whose actuaries use NASI cost-per-payroll as a benchmarking input when setting expected loss rates for 2024 and 2025 accident years. The effect is not yet visible in the development triangle, because recent accident years are still immature. That is precisely the situation where the BF expected loss complement carries the most weight, and where a low a priori pick does the most damage to reserve adequacy.
For self-insured WC programs, the risk is reserve inadequacy on recent accident years: the expected claim cost is anchored at a historically low level while actual emerging costs are running well above it.
What This Means for Your Next Review
Ask your actuary two questions before the next reserve update:
- What data year does the expected loss pick reference, and is the NASI sub-$1.00 figure being used as a direct input or a sanity check?
- If 2024 and 2025 indemnity severity is running 6% or more above the 2023 baseline, by how much could the BF expected loss complement be understating IBNR on those accident years?
If your actuary does not have a ready answer to the second question, that is a signal to look harder at recent accident year adequacy. A diagnostic review of reserve drivers for accident years 2023 through 2025 is a reasonable step before year-end.
Watch Next
NASI’s 2024 data release (expected late 2027) will be the first opportunity to see whether the 2024 severity acceleration appears in system-level benchmarks. Until then, any reserve work relying on the sub-$1.00 NASI figure is working from a baseline that is at minimum two severity inflection years behind current conditions.