NCCI Reserve Redundancy Erosion Signals Workers Comp Pricing Cycle Turn

From tracking NCCI loss cost filings across 38 states, the pattern of the past decade has been reliable: frequency declining, severity contained, rate decreases filed in most jurisdictions, and favorable prior-year development producing calendar year combined ratios well below 100. That pattern started bending in 2024, and the data NCCI presented at the 2026 Annual Insights Symposium in Orlando on May 12 to 14 confirms the structural shift is accelerating.

The headline looks comfortable: a calendar year 2025 combined ratio of 91, marking 12 consecutive years of underwriting profitability for the workers compensation line. The number underneath tells a different story. The accident year 2025 combined ratio is 102, meaning current-year business is generating losses faster than premiums can cover them. The 11-point gap between CY 91 and AY 102 comes entirely from prior-year favorable reserve development. And the reserve pool feeding that development is shrinking.

The Calendar Year / Accident Year Gap: Anatomy of an 11-Point Divergence

Understanding the 11-point gap between CY 91 and AY 102 requires decomposing how prior-year reserve development flows into calendar year financials. When an insurer sets case reserves and IBNR estimates for accident year 2020, for example, those reserves reflect the actuary's best estimate of ultimate losses at that point in time. As claims mature and close, actual payments frequently come in below the original estimate, particularly in workers compensation where long-tail medical claims develop over 10 to 20 years. The difference between the original estimate and the revised estimate flows through the income statement as favorable development in the calendar year when the revision occurs.

NCCI's reserve redundancy estimate measures the aggregate industry-level cushion: the amount by which carried reserves on NAIC Annual Statement Schedule P exceed NCCI's actuarial estimate of ultimate losses across all open accident years. That cushion stood at approximately $18B in 2023, declined to $16B in 2024, and contracted further to $14B in 2025. The $14B represents roughly 12% of total carried reserves, a margin that has compressed meaningfully over two years.

The implied annual reserve release is the arithmetic bridge between the accident year and calendar year results. With an AY combined ratio of 102 and a CY combined ratio of 91, the annual favorable development flowing into the P&L is approximately $4B to $5B on a net written premium base of $41.6B. Simultaneously, the redundancy stock is declining by $2B per year. This means the industry is drawing down its reserve bank faster than favorable development from older accident years can replenish it, because the more recent accident years (2020 forward) are developing less favorably than the AY 2010 to 2018 cohort that built the cushion in the first place.

How Reserve Redundancy Feeds Back Into Filed Loss Costs

NCCI's loss cost filing methodology uses historical accident-year experience developed to ultimate via standard loss development methods. For each state filing, reported losses (paid losses plus case reserves) by accident year are organized into development triangles, with each column representing an additional year of maturity. Age-to-age development factors are selected from these triangles, and a cumulative development factor (CDF) converts reported losses at each maturity to an estimated ultimate. Trend factors for frequency, medical severity, and indemnity severity are then applied prospectively through the rate effective period.

The Bornhuetter-Ferguson framework provides a useful lens for understanding the interaction between development and initial estimates. Under BF, the ultimate loss estimate is a credibility-weighted blend of two inputs: the actual reported losses at the current evaluation point and an a priori expected loss ratio multiplied by earned premium. As the accident year matures, the credibility weight on actual experience increases and the weight on the a priori estimate decreases. For a mature accident year (say, AY 2015 at 10 years of development), the actual experience dominates and the BF estimate converges on the chain-ladder result. For an immature year (AY 2024 at one year of development), the a priori expected loss ratio carries substantial weight.

The reserve redundancy problem enters this framework through the tail. When older accident years consistently develop more favorably than the selected development pattern, the implied tail factors are overstated, which means carried reserves exceed what will actually be paid. As those redundant reserves are recognized, they flow into calendar year income. But when the development patterns shift, specifically when more recent accident years develop less favorably or in line with the selected factors rather than beating them, the annual release shrinks. Filed loss costs, which rely on development patterns from the historical triangle, continue to reflect the older, more favorable development experience for several years after the shift occurs.

This creates a structural lag. NCCI's filed loss cost decreases for the 2025 to 2026 period average 5.0% across approved filings, reflecting the favorable development that dominated the experience period used in those filings. But the AY 2025 combined ratio of 102 indicates that current pricing is already inadequate on an undiscounted basis. The lag between the actuarial signal (AY above 100) and the filed loss cost response (decreases flipping to flat or positive) is typically two to three years, because the development triangle needs to accumulate enough data at sufficient maturity to shift the selected factors.

Severity Acceleration Versus Frequency Deceleration: The Net Loss Cost Trend

The components underneath the AY 102 combined ratio reveal the source of the deterioration. NCCI reported lost-time claim frequency declining 2% in 2025, a deceleration from the long-term average annual decline of approximately 3% to 4%. Medical claim severity increased 4%, and indemnity claim severity increased 4%. The net loss cost trend, the combined effect of frequency and severity changes, works out to approximately +2% (negative 2% frequency plus positive 4% severity, simplified).

A positive net loss cost trend in workers compensation is significant because the line has operated with a negative or flat net trend for most of the past 15 years, driven by frequency declines that more than offset severity increases. That frequency tailwind powered the sustained rate decreases. With frequency deceleration slowing toward negative 2% and severity running at positive 4% on both the medical and indemnity sides, the crossover to a positive net trend implies upward pressure on loss costs even before accounting for any reserve development adjustments.

Donna Glenn, FCAS, MAAA, NCCI's Chief Actuary, noted in her State of the Line presentation that frequency decline has moderated and does not appear to be signaling a systemic reversal. The slowdown reflects compositional shifts: tourism and hospitality employment, which carries higher frequency, grew disproportionately in certain states. But the practical consequence for pricing work is the same regardless of cause. A negative 2% frequency trend provides less offset to a positive 4% severity trend than a negative 4% frequency trend did, and the filed loss cost indications need to reflect that arithmetic.

Payroll grew approximately 5% in 2025, driven almost entirely by wage increases rather than employment gains. Only the healthcare sector showed meaningful employment growth. For premium base adequacy, this means exposure growth reflects inflation rather than incremental insured risk. Earned premium rises with payroll, but so does indemnity severity, because temporary total disability and permanent partial disability benefits are tied to pre-injury wages. The payroll growth offsets some of the premium shortfall in absolute dollars but does not improve the loss ratio because the benefit payments inflate in lockstep.

California at AY 129: The Bellwether State

California's accident year 2025 combined ratio of 129 is the most visible illustration of where the national trajectory could head if reserve releases stop covering the gap. California accounts for roughly 20% of national workers compensation premium, so its results exert an outsized pull on the countrywide accident year combined ratio. The state has reported accident year combined ratios above 100 for the past five years, meaning the national AY 102 figure actually understates the deterioration in California while overstating it for the rest of the country.

Several California-specific dynamics amplify the loss cost pressure. Cumulative trauma claims represent approximately 26% of lost-time claims in the state, compared to 1% to 6% in NCCI-administered jurisdictions. Attorney involvement in California WC claims runs significantly higher than in other states, increasing defense and cost containment expenses. The WCIRB approved an 8.7% advisory pure premium rate increase effective September 2025, the first meaningful rate hike in a decade, signaling that even California's independent rating organization has recognized the loss cost inadequacy.

For actuaries working outside California, the state functions as a leading indicator. California's scale means its reserve development (or lack thereof) directly affects the national redundancy estimate. If California's AY 2020 through 2024 cohort develops adversely, the draw on national redundancy accelerates, pulling the $14B figure lower and compressing the CY/AY gap faster than the current trajectory implies.

State Filing Variation: From Negative 15.6% to Positive 21.6%

The range of approved NCCI loss cost filing changes for the 2025 to 2026 period spans from a decrease of 15.6% to an increase of 21.6% (Nevada). This 37-point spread illustrates why national averages can mislead. The 5.0% average decrease masks states where loss costs are already increasing and states where large decreases continue based on local experience.

Nevada's 21.6% increase represents the most dramatic outlier. The state's loss cost filing reflects a sharp deterioration in loss experience that required a substantial correction after years of inadequate pricing. Nevada serves as a microcosm of the broader cycle dynamic: sustained decreases compressed loss costs to a level that could not absorb a severity shift, and the correction, when it came, was abrupt rather than gradual.

For pricing actuaries preparing state-level loss cost indications, the national reserve redundancy figure and the countrywide AY combined ratio provide context, but the state-specific development triangle is the binding constraint. States with thin books, high cumulative trauma exposure, or recent fee schedule adjustments may be closer to the inflection point than the national data suggests. States with robust fee schedules and stable frequency profiles may have several years of runway before loss cost inadequacy surfaces.

Historical Precedent: When Does the Pricing Cycle Turn?

Workers compensation has experienced three major pricing cycle inflections in the past 30 years. The early 2000s turn followed several years of accident year combined ratios above 100 through the late 1990s, with the pricing correction arriving in 2002 to 2003 after reserve deficiencies became undeniable. The post-financial-crisis period saw AY combined ratios deteriorate through 2010 to 2012, with reserve deficiencies peaking at approximately $10B before the sustained favorable development era of 2013 to 2022 reversed the position.

The pattern from these cycles suggests a two-to-three-year lag between when accident year combined ratios first breach 100 persistently and when filed loss costs begin increasing across a majority of states. The current data shows the AY combined ratio at 102 for 2025, up from 99 in 2024 (based on the prior State of the Line). If the AY ratio remains at or above 100 through 2026 and 2027, and if redundancy continues to erode at approximately $2B per year, the reserve cushion reaches $10B to $12B by 2027, and the annual release narrows to a level that no longer holds the CY combined ratio below 100.

At that point, the CY combined ratio breaches 100, reported financials show underwriting losses, and the competitive and regulatory dynamics shift from decreases to increases. Based on the historical lag, a pricing inflection arriving by 2028 to 2029 is consistent with the trajectory, assuming no exogenous shock (pandemic, legislative expansion, judicial benefit increase) accelerates the timeline.

Why This Matters for Pricing Actuaries

The practical implications for WC pricing work fall into three areas. First, development factor selection. Actuaries building loss cost indications should evaluate whether their selected development patterns still reflect the tail behavior embedded in the current triangle. If the most recent accident years are developing less favorably than the AY 2010 to 2018 cohort, the selected tail factors may be too high, which paradoxically delays the recognition of loss cost inadequacy by overstating the ultimate for immature years. Running the indication with tail factors based only on the AY 2018 forward cohort provides a sensitivity test that may show a materially different loss cost level.

Second, trend selection. A net loss cost trend of approximately +2% (negative 2% frequency, positive 4% severity on both medical and indemnity) should be the starting point for 2026 and 2027 rate effective periods. Actuaries who select a negative net trend based on longer historical averages risk understating the loss cost need. The severity acceleration is the dominant signal, and it should carry more weight in the trend selection than the decelerating frequency decline.

Third, reserve adequacy monitoring. The calendar year combined ratio will remain below 100 for at least one to two more years as the $14B redundancy cushion continues to release. Actuaries responsible for reserve opinions should stress-test their carried reserves against a scenario where redundancy erodes to $10B by 2027, annual releases shrink from $4B to $5B to $2B to $3B, and the CY combined ratio rises from 91 to 96 to 98 over the next two to three years. That scenario is not a tail risk; it is the baseline trajectory embedded in the current NCCI data.

The WC pricing cycle has not turned yet. The 2026 State of the Line shows a line still nominally profitable on a calendar year basis, still supported by prior-year releases, still filing rate decreases in a majority of states. But the accident year data, the severity trends, the decelerating frequency, and the eroding reserve cushion collectively describe a system approaching exhaustion. The pricing actuary who builds that forward view into today's loss cost selections positions the book for the turn rather than scrambling to catch up after it arrives.