NCCI Employment Surge Raises New-Worker Frequency Risk for WC Pricing

From tracking NCCI loss cost filings across 38 jurisdictions, the pricing assumption that has held longest and delivered the most rate relief is the frequency trend. Since the early 1990s, lost-time claim frequency has declined nearly every year, averaging roughly negative 3.6% annually from 2004 through 2023. That persistent decline has offset severity increases and underwritten a decade of consecutive rate decreases. But NCCI's May 2026 data releases point to a structural risk that could bend the frequency curve in the opposite direction: the U.S. economy is hiring again, and the workforce composition shift that comes with rapid hiring has a well-documented, quantifiable effect on workplace injury rates.

NCCI's May 2026 Labor Market Insights report, published May 12 from its Annual Insights Symposium in Orlando, shows monthly employment growth accelerating to 76,000 jobs per month through early 2026, up from just 10,000 per month in 2025. The April 2026 employment report was even stronger: the private sector added 123,000 jobs, marking the second consecutive monthly gain and breaking an 11-month streak of alternating employment gains and losses. Healthcare and social assistance led with 53,900 new positions, followed by transportation and warehousing at 30,300 and retail trade at 21,800.

The New-Worker Effect: Quantifying the Frequency Exposure

The relationship between employment tenure and workplace injury risk is one of the most robust findings in occupational safety research. The Institute for Work & Health (IWH) in Toronto, studying Ontario workers' compensation data, found that employees on the job for less than one month had four times the claim rate of workers with more than one year of tenure. By 2007, that ratio had moderated to 3.14 times, still a substantial differential. Men faced five times the first-month risk; women, 3.3 times. Second-month claim rates dropped to roughly double the rate of tenured workers, suggesting the steepest risk gradient occurs in the initial weeks of employment.

Travelers' 2025 Injury Impact Report, drawing on 1.2 million claims from 2021 through 2025, reinforces the pattern from a U.S. carrier perspective. First-year employees now account for 36% of all workplace injuries and 34% of total claim costs, up from 34% and 32% respectively in the prior five-year period. In construction, the concentration is sharper: 44% of injuries and 47% of compensation costs fall on first-year workers. Restaurants see 50% of injuries among first-year staff. These are not marginal effects; they are the dominant driver of frequency variation within a book of business.

The pricing implication is mechanical. When hiring accelerates, the share of the insured workforce in the first zero-to-twelve months of tenure increases. That compositional shift raises the blended frequency rate even if no individual worker's injury probability changes. For a workforce that adds 76,000 net new jobs per month, approximately 900,000 workers per year enter the high-risk tenure window. If these workers carry a lost-time claim frequency three times the rate of tenured workers, the incremental frequency contribution is the equivalent of adding 2.7 million worker-years at the baseline rate, roughly a 1.5% to 2% upward frequency shift against a workforce base of approximately 160 million.

Why the Frequency Decline Slowed to Negative 2%

NCCI's 2026 State of the Line report confirmed that lost-time claim frequency declined only 2% in 2025, well below the long-run average of negative 3.6% per year. For context, accident year 2024 frequency fell 6%, more than double the historical trend. The 2025 deceleration represents a four-percentage-point swing in a single year.

Several compositional factors contributed. Donna Glenn, FCAS, MAAA, NCCI's Chief Actuary, noted at the symposium that healthcare frequency "ticked up," driven partly by newer employees with less experience and partly by rising workplace violence, with assault rates in healthcare growing 5.3% annually from 2011 through 2022. Office-sector frequency also increased, attributable to motor vehicle accidents among sales personnel. Patrick Coate, PhD, NCCI's Senior Economist, added a demographic overlay: work injury rates have increased every year since 2015 for workers aged 65 and older, while declining for other age groups. As retirements accelerate and new hires replace experienced workers, the blended frequency composition shifts toward both ends of the tenure and age spectrums where injury rates are highest.

The remote work dividend, which drove a 40% decline in WC claims frequency among office workers from 2019 through 2022 according to NCCI research, also appears to be plateauing. NCCI's preliminary 2023 data shows slight claim upticks among some remote-friendly categories, suggesting that partial return-to-office mandates are clawing back a portion of that frequency benefit. Remote-friendly occupations generate only about 11% of total WC premium despite representing half of insured payroll, so the frequency effect of office re-densification is modest in dollar terms but directionally unfavorable.

Frequency Trend Selection Under a Potential Structural Break

NCCI's standard approach to frequency trend projection uses an exponential trend regression model applied to the latest 5, 8, and 15 observations of lost-time claim frequency data. The three estimates are aggregated using actuarial judgment or, more formally, through Bayesian trend selection that averages across the three fits using a double-exponential likelihood and minimizes the sum of absolute forecast errors across overlapping holdout periods. Only lost-time claims are included; medical-only claims, which represent over 60% of total claims but under 4% of losses, are excluded to avoid distorting the trend signal.

This framework performs well under stable trend regimes. When the frequency curve is steadily declining at negative 3% to negative 4%, the 5-point, 8-point, and 15-point fits converge and the selection is straightforward. The problem arises when the most recent data points diverge from the longer history, as they are now. A 5-point fit covering accident years 2021 through 2025 captures the post-COVID frequency rebound (2021 uptick), the sharp 2024 decline (negative 6%), and the 2025 deceleration (negative 2%). That series is noisy and produces a fitted trend that depends heavily on how the actuary weights the endpoints. A 15-point fit extending back to 2011 smooths the noise but embeds a decade of steeper frequency declines that may no longer be representative.

Pricing actuaries facing this kind of potential inflection have several diagnostic tools. A Chow test for structural break evaluates whether the regression parameters differ significantly between two sub-periods, for example, 2011 to 2019 versus 2020 to 2025. CUSUM and CUSUM-of-squares statistics track whether the cumulative sum of recursive residuals exceeds a significance boundary, signaling parameter instability. Change-point detection algorithms (Bai-Perron, PELT) identify the most likely breakpoint(s) in the time series without requiring the actuary to pre-specify the break date.

If the diagnostics confirm a structural break around 2020 or 2021, the appropriate response is to down-weight or exclude the pre-break data and select the frequency trend primarily from the post-break window. A trend fitted to 2021 through 2025 data would likely produce a frequency projection in the range of negative 1% to flat, compared to the negative 3% to negative 4% that a longer window supports. That difference, while seemingly small, compounds through the rate effective period and cascades into the net loss cost indication.

Net Loss Cost Arithmetic: When Does the Indicated Change Turn Positive?

The components underneath NCCI's accident year 2025 combined ratio of 102 tell a clear severity story. Medical claim severity grew 4%, and indemnity claim severity grew 4%. Against a frequency decline of negative 2%, the net loss cost trend works out to approximately positive 2%: the combined severity increase exceeds the frequency decrease. For the WC line that has operated with a negative or flat net trend for most of the past 15 years, a positive net trend is a pricing regime change.

Now consider the employment growth scenario. If hiring continues at 76,000 jobs per month through the remainder of 2026, the new-worker effect could erode the frequency decline from negative 2% toward negative 1% or flat. With severity holding at positive 4% on both medical and indemnity sides, the net loss cost trend shifts to positive 3% to positive 4%. Applied prospectively through a two-year rate effective period, this implies an indicated loss cost increase of 6% to 8%, a figure that stands in stark contrast to the average 5% rate decrease NCCI has filed for the current cycle.

The approved filings projecting a further 5% premium reduction from 2025 to 2026 were built on experience periods and trend selections that predate the employment acceleration. Those filings used frequency trends fitted through 2024, a year in which frequency declined 6% and employment growth was minimal. The lag between when the employment surge materializes in the frequency data and when it flows into filed loss cost indications is at least two years, because the development triangle and trend regression need two to three data points at the new level before the fitted trend shifts meaningfully.

Sector-Specific Frequency Signals

The sector composition of current hiring matters for frequency trend selection because WC class-level frequency rates vary by orders of magnitude. Healthcare, which led April 2026 employment growth with 53,900 new jobs, carries a distinct injury profile: higher frequency of musculoskeletal injuries (patient lifting, repetitive strain) and lower average indemnity severity compared to construction or manufacturing. The BLS projects healthcare and social assistance will add approximately 2 million jobs from 2024 through 2034, the fastest growth of any sector. If healthcare dominates the hiring mix, the compositional frequency effect is weighted toward moderate-severity, higher-frequency claims.

Construction and manufacturing, by contrast, show employment gains but at lower volumes. Construction first-year workers carry the highest injury concentration of any sector at 44% of claims and 47% of costs (Travelers data). A construction hiring surge would produce a more severe frequency impact per worker than healthcare hiring, but the current employment data does not support that scenario.

For class-level pricing actuaries building state-specific indications, the actionable step is to disaggregate the frequency trend by industry group. If healthcare frequency is rising while manufacturing frequency continues declining, a single blended frequency trend applied to all classifications understates the exposure in healthcare classes and overstates it in manufacturing. NCCI's published frequency data supports this disaggregation at the industry-group level, and the state-specific filing can incorporate class-group-specific trend selections where credibility permits.

The Reserve Cushion and Timing of the Cycle Turn

The $14 billion industry reserve redundancy, down from $16 billion in 2024 for the second consecutive year of erosion, provides a buffer that masks the emerging loss cost inadequacy. Calendar year favorable development of approximately $4 billion to $5 billion per year bridges the gap between the CY combined ratio of 91 and the AY combined ratio of 102. As long as reserve releases continue at that pace, the P&L looks profitable and competitive pressure maintains rate decreases.

But the buffer is finite. At the current erosion rate of roughly $2 billion per year, redundancy reaches $10 billion to $12 billion by 2027. If the employment surge accelerates frequency deterioration and pushes the AY combined ratio from 102 toward 104 or 105, the redundancy drain accelerates as well, because the newer accident years contributing to the carried reserves develop less favorably. The pricing cycle turn, when filed loss costs begin increasing across a majority of NCCI states, historically follows two to three years after the AY combined ratio persistently breaches 100. The AY ratio has been above 100 for two years running. If the new-worker frequency effect pushes it higher in 2026, the timeline to a pricing inflection shortens.

Why This Matters for Pricing Actuaries

The employment surge documented in NCCI's May 2026 data creates a specific, quantifiable risk to the frequency trend assumption that underpins every WC loss cost filing. Three practical actions follow for actuaries preparing 2027 cycle indications.

First, test for a structural break in the frequency trend. Run formal diagnostics (Chow test, CUSUM, or change-point detection) on the lost-time frequency series using the post-2020 data as the candidate break window. If the test confirms a structural shift, weight the 5-point exponential trend more heavily than the 15-point fit, and consider a flat or negative 1% frequency trend rather than the historical negative 3% to negative 4%.

Second, monitor employment growth as a leading indicator. Payroll growth of 5%, split roughly as 1% employment and 4% wages, supports premium volume. But the employment component, not the wage component, drives the new-worker frequency effect. Track monthly BLS data by sector and map it to the WC class mix in the rated state. If healthcare and construction hiring continue to lead, apply class-group-specific frequency adjustments rather than a single blended trend.

Third, stress-test the net loss cost indication. With severity at positive 4% on both medical and indemnity sides and frequency potentially flattening from negative 2% toward zero, the net indicated loss cost change shifts from negative (supporting continued rate decreases) to positive (requiring rate increases). A scenario where the frequency trend reaches zero and severity holds at positive 4% produces a net indicated increase of approximately 8% on a two-year prospective basis. That represents the first such reversal in over a decade of WC pricing and would mark the effective end of the current soft-market cycle.

The frequency trend is the load-bearing assumption in WC loss cost work. For 30 years it has moved in one direction. The employment surge does not guarantee a reversal, but it introduces a compositional shift with a well-documented, statistically significant effect on injury rates. Pricing actuaries who incorporate that signal into their 2027 trend selections will be ahead of the curve. Those who rely on the 15-year average may find their indications overtaken by events before the rate effective period expires.