From tracking NCCI loss cost filings across thirty-plus states, patterns emerge in how frequency and severity trends interact during periods of economic transition. The 2024 accident year data presents one of the widest such divergences in recent memory: lost-time claim frequency fell 6%, nearly double the long-term average annual decline of 3.6% measured from 2004 through 2023, while medical severity rose 6% and indemnity severity rose 5%. The resulting pure premium trend is roughly flat, but that single number conceals the opposing forces underneath. Getting the component-level trend selection right matters far more than the combined answer, because frequency and severity respond to different economic drivers, move on different time horizons, and have different implications for whether current rate levels will remain adequate through the next policy period.

86%
CY 2024 private carrier combined ratio, the 8th consecutive year below 90
99%
AY 2024 combined ratio, a 13-point gap driven by favorable prior-year reserve development
$41.6B
Net written premium in 2024, down 3.2% from prior year despite payroll growth

How NCCI Selects Frequency and Severity Trends

NCCI develops separate frequency and severity trend factors using exponential (log-linear) regression on countrywide accident-year data developed to ultimate. The model form is straightforward: ln(metric) = a + b × t, where t indexes accident years and the slope coefficient b represents the average annual percentage change. A negative slope on frequency reflects long-term workplace safety improvements, industry-mix shifts toward lower-hazard employment, and the cumulative effect of claims management practices. A positive slope on severity reflects medical price inflation, wage growth, and utilization changes in healthcare delivery.

NCCI fits these curves over multiple window lengths, typically ranging from ten to twenty years, and presents the resulting trend projections alongside the countrywide average selections. The fitting period matters because shorter windows are more responsive to recent acceleration or deceleration but more vulnerable to single-year noise. Longer windows produce more stable estimates but can underweight structural shifts. In practice, NCCI's countrywide frequency trend has hovered near negative 3% to negative 4% per year for the past decade. The 2024 data point, at negative 6%, falls well below the fitted line but a single year of acceleration, however large, may not meaningfully shift a log-linear slope fitted over ten-plus years of data.

Medical and indemnity severity trends require separate treatment because they respond to fundamentally different cost drivers. Medical severity is a function of medical price inflation (as measured by the Workers' Compensation Weighted Medical Price Index, or WCWMI), utilization changes in treatment patterns, and shifts in the mix of services across the fee schedule. Indemnity severity tracks wage inflation and claim duration, specifically how long injured workers remain on temporary total or permanent partial benefits before return to work or settlement. NCCI's 2024 data illustrates the distinction: the WCWMI rose only 2.8% and general medical CPI stood at 2.9%, but actual medical severity per lost-time claim increased 6%, implying that utilization growth and treatment mix shift added roughly three percentage points above pure price inflation. Indemnity severity's 5% increase aligned more closely with the 5.6% wage growth reported by the Bureau of Economic Analysis for 2024.

The Pure Premium Trend Formula

Frequency and severity trends combine into an overall pure premium trend through multiplication:

Pure Premium Trend = (1 + Severity Trend) × (1 + Frequency Trend) − 1

When the severity trend is positive and the frequency trend is negative, these components partially or fully offset each other. Using the 2024 observed values as illustration: if severity runs at +5.5% (blending the medical and indemnity components) and frequency at negative 6%, the pure premium trend works out to (1.055) × (0.94) − 1 = negative 0.83%. That near-zero result looks benign. But the actuary selecting trends for a loss cost filing must decide whether to give full weight to the latest year's movements or rely more heavily on the longer-term fitted slopes.

A selected frequency trend of negative 4% (closer to the long-term average) combined with a selected severity trend of +4% (below the latest year's spike but above the prior three-year average) produces a pure premium trend of negative 0.16%. The difference between negative 0.83% and negative 0.16% might seem trivial, but when projected from the experience period midpoint to the prospective policy period midpoint over a two-year or longer trend period, the cumulative impact on the indicated rate change can reach one to two percentage points, enough to shift a filing from a decrease to flat or from flat to a modest increase.

The selected trend feeds directly into the standard rate indication formula:

Indicated Rate Change = [(Projected Ultimate Losses × LCM × ULAE Loading) / (Earned Premium at Current Rate Level)] − 1

Projected ultimate losses reflect the trended and developed loss experience. The loss cost multiplier (LCM) and unallocated loss adjustment expense (ULAE) loading translate pure losses into the full cost basis. When the trend selection changes the numerator by even a small percentage, the indicated rate change shifts accordingly.

Calendar Year vs. Accident Year: What the 86% Combined Ratio Masks

The 86% calendar year combined ratio that headlines NCCI's 2024 results is the number carriers and regulators cite most frequently. It represents the eleventh consecutive year of underwriting profitability for private workers compensation carriers. But the accident year 2024 combined ratio is 99%, a thirteen-point gap explained almost entirely by favorable prior-year reserve development.

For pricing actuaries, the distinction matters enormously. Calendar year results include adjustments to reserves established for prior accident years. When older claims close for less than their case reserves, or when bulk IBNR factors prove conservative, the resulting favorable development flows through the calendar year income statement and compresses the combined ratio. NCCI's data indicates roughly $4 billion in annual prior-year reserve reductions have benefited calendar year results in each of the last seven years. Estimated industry reserve redundancy stood at $16 billion as of year-end 2024, down from $18 billion a year earlier.

The erosion from $18 billion to $16 billion is worth watching. If accident year 2024 truly runs at a 99% combined ratio, the current-year business is producing near-breakeven underwriting results before investment income. The 86% headline depends on a cushion that is gradually depleting. Pricing actuaries evaluating whether current rate levels are sufficient should anchor to the accident year result, not the calendar year number, and should consider how long the favorable development tailwind will persist as older, more reserved accident years cycle out of the development period.

First-half 2025 data already hints at the shift. NCCI's preliminary figures show the direct loss ratio at 50.1%, up two points from the first half of 2024. Chief Actuary Donna Glenn has projected a full-year 2025 net combined ratio in the range of 85% to 93%, the widest projection band NCCI has issued in recent years, reflecting genuine uncertainty about whether the recent medical severity acceleration is transient or persistent.

State-Level Credibility and Trend Blending

Countrywide trends provide the foundation, but individual state loss cost filings require state-specific trend selections. The challenge is that smaller states may have only a few thousand lost-time claims per accident year, making single-state trend fitting statistically unreliable. A state with 2,000 claims per year has enough volatility in its frequency and severity data that the exponential curve fit for that state alone might produce slope estimates with wide confidence intervals.

NCCI addresses this through credibility-weighted blending. The state's own trend data receives a credibility weight based on its volume of experience, and the complement of credibility is assigned to the countrywide trend. For large states with tens of thousands of annual claims, the state-specific trend receives high credibility and the countrywide complement has minimal influence. For smaller states, the blend leans heavily toward the countrywide figure. This framework follows Bühlmann credibility principles, balancing the responsiveness of state-specific data against the stability of the broader pool.

State-level divergence from countrywide trends can be substantial. Colorado's January 2026 loss cost filing produced a negative 6.9% average change, marking twelve consecutive years of decreases in that state. The Colorado Division of Insurance's independent actuarial review (conducted by Davies Group) confirmed that NCCI's loss development and trend selections were reasonable, with one adjustment: the medical loss trend was revised from negative 6.0% to negative 6.5%. Connecticut's 2026 filing produced a negative 3.8% change. Both states illustrate the current environment where ongoing rate decreases reflect the accumulated years of favorable frequency and the offsetting severity pressures have not yet been large enough to reverse the direction.

Three-Decimal Precision: A Quiet Change with Real Filing Impact

Effective with filings for January 1, 2026 and subsequent effective dates, NCCI extended loss costs, rates, and expected loss rates from two decimal places to three. Intermediate values (indemnity and medical pure premiums) moved from three to four decimal places. The underlying ratemaking methodology did not change; this is purely a precision enhancement.

The practical significance is concentrated in low-rate classification codes. Consider a class code with a current loss cost of $0.10 per $100 of payroll. Under two-decimal rounding, the smallest possible adjustment was $0.01, a 10% swing in either direction. Any indicated change smaller than 5% would round to zero, creating a dead zone where legitimate trend-driven adjustments simply vanished. With three-decimal precision, adjustments of $0.001 are possible, representing a 1% change for that same $0.10 loss cost. For the roughly 20% of NCCI class codes where manual rates fall below $0.20, this change enables more responsive rate adjustments that actually reflect the indicated actuarial change rather than being masked by rounding constraints.

Colorado's January 2026 filing was among the first to use the three-decimal format, and the granularity was immediately visible in the class-level exhibits where codes that had shown zero change for several consecutive filings now reflected small but actuarially indicated adjustments.

Payroll Growth as a Hidden Premium Offset

Filed rate decreases tell only part of the premium story. Workers compensation premium equals payroll multiplied by rate, and payroll exposure has been growing steadily. NCCI's first-half 2025 data shows payroll up approximately 5.0% year-over-year, decomposed as 1% employment growth and 4% wage inflation. The Bureau of Labor Statistics Employment Cost Index for March 2026 measured private-industry compensation costs up 3.4% over the prior year.

This means that a filed rate decrease of 6% does not translate into a 6% decrease in the employer's actual premium cost. If payroll grows 5%, the employer's premium declines only about 1% on a dollar basis. Net written premium for the industry declined just 3.2% in 2024 to $41.6 billion, a much smaller drop than the cumulative rate-level decreases would suggest in isolation. For pricing actuaries, the payroll offset is relevant to the premium volume forecasts that underpin expense ratio assumptions and the earned premium denominator in the rate indication formula. It also explains why carrier profitability has remained strong despite a decade of filed decreases: the exposure base keeps growing even as the rate per unit of exposure declines.

What to Watch at AIS 2026

NCCI's Annual Insights Symposium opens May 11 in Orlando, with Donna Glenn's State of the Line presentation anchoring the agenda. The 2025 full-year data that Glenn will present builds on the foundation reviewed here, and several questions are worth tracking:

First, whether medical severity acceleration persisted into accident year 2025 or reverted toward the long-term average. The gap between the 2.8% WCWMI and the 6% observed medical severity in 2024 implies that utilization, not price, is the marginal driver. If utilization growth proves persistent, selected medical severity trends will need to shift upward.

Second, whether the loss ratio uptick visible in first-half 2025 data (50.1%, up two points) held through the full year. If the full-year 2025 accident year combined ratio comes in above 100%, it would be the first time in over a decade that current-year business failed to cover its own costs before investment income.

Third, the updated reserve redundancy estimate. The decline from $18 billion to $16 billion in a single year, while still large, indicates that the prior-year development cushion is finite and contracting. Pricing actuaries relying on calendar year results to justify further rate decreases should consider how many more years of favorable development the remaining reserve margin can support.

The frequency-severity split in NCCI's 2024 data is not a crisis. A near-zero pure premium trend is consistent with a stable, mature line of business. But the components underneath that flat composite are moving faster and in opposite directions than at any point in the past decade. The trend selections that pricing actuaries make in the next round of loss cost filings will determine whether workers compensation continues its extended profitable run or begins the gradual turn that the accident year results already suggest is closer than the calendar year headline implies.

Further Reading

Sources