From tracking NCCI loss cost filings across dozens of jurisdictions, we have seen how sensitive the indicated rate change is to the wage trend assumption. A quarter-point shift in the annual wage growth selection, compounded over a two-and-a-half-year trend period, can move the filing indication by a full percentage point. The ECI wages-and-salaries series for private industry workers has been the default benchmark for that assumption for years, and the BLS just announced two simultaneous changes that will break the continuity of the series. Starting with the January 2027 data release covering December 2026, workers' compensation insurance costs will be permanently removed from both the ECI and the Employer Costs for Employee Compensation (ECEC) measures, and the 2021 fixed employment weights that have been in use since December 2022 will be replaced with new 2025 weights derived from updated Occupational Employment and Wage Statistics (OEWS) and Quarterly Census of Employment and Wages (QCEW) data.

3.4%
Q1 2026 ECI: private industry wages and salaries, year-over-year
3.6%
Q1 2026 ECI: employer benefit costs, year-over-year
5.7%
Q1 2026: employer health insurance cost growth, outpacing wages for fifth straight quarter

What BLS Is Changing

The April 30, 2026 ECI release included a formal notice: beginning with the December 2026 reference period (published January 2027), workers' compensation insurance costs will no longer appear in either the ECI benefits component or the ECEC compensation tables. BLS cited two reasons. First, workers' compensation costs have become a diminishing share of total employer benefit costs, making them less consequential to the headline index. Second, collecting accurate WC cost data at the establishment level has grown more complex as carriers have shifted to retrospectively-rated and loss-sensitive programs that make the per-employee cost difficult to pin down in a quarterly survey.

Simultaneously, BLS will introduce 2025 fixed employment weights to replace the 2021 weights that have been in effect since the December 2022 reference period. The ECI is a fixed-weight Laspeyres index: it holds occupational and industry employment shares constant so that changes in the index reflect pure compensation growth rather than shifts in the employment mix. When the weights are updated every few years, the composition of the index changes, and any sector that has grown or shrunk disproportionately since the prior weight year will shift the implied compensation trajectory. The transition from 2021 to 2025 weights captures four years of post-pandemic labor market restructuring, including the accelerated growth of healthcare, logistics, and technology employment relative to manufacturing and extractive industries.

How the ECI Feeds Into WC Loss Cost Filings

The standard NCCI loss cost trending framework projects ultimate losses from the historical experience period to the prospective policy period using a trend factor that decomposes into frequency trend and severity trend. Frequency trend captures changes in claim occurrence rates per unit of exposure. Severity trend captures changes in the average cost per claim, and for indemnity benefits, severity is driven primarily by wage growth because statutory benefit formulas (temporary total disability, permanent partial disability) are tied to the injured worker's pre-injury average weekly wage, which in turn tracks economywide wage inflation.

The pricing actuary selects an annual wage growth assumption, typically benchmarked to the ECI wages-and-salaries series for private industry workers, and applies it over the trend period. The trend period runs from the midpoint of the historical experience period to the midpoint of the prospective policy period, usually spanning two to three years. The mechanics are straightforward:

Indemnity Severity Trend Factor = (1 + w)n

where w is the selected annual wage trend and n is the trend period in years. Using the Q1 2026 ECI reading as an example: if the actuary selects w = 3.4% and the trend period is 2.5 years, the indemnity severity trend factor is (1.034)2.5 = 1.0865, meaning indemnity severity is projected to grow 8.65% over the trend period. If the wage trend selection were instead 3.0%, the factor drops to (1.030)2.5 = 1.0762, a difference of roughly one percentage point in projected losses. On a $500 million loss cost base, that difference is $5 million in indicated premium.

The Discontinuity Problem

When BLS removes WC costs and re-weights the index in December 2026, the pre-transition and post-transition ECI series will no longer be directly comparable. Workers' compensation costs represented roughly 1.4% to 1.6% of total employer compensation costs in recent years (per ECEC data), so their removal from the benefits component introduces a small but non-trivial level shift. The weight update has a larger potential impact: if the 2025 occupational distribution assigns more weight to higher-wage or faster-growing sectors than the 2021 distribution did, the implied compensation growth rate could shift upward even if underlying wages in each sector remain unchanged.

Pricing actuaries who use multiyear ECI history to fit a wage trend face a choice at the transition point. Two options dominate:

Option 1: Splice the old and new series. Calculate a linking factor at the December 2026 overlap point by comparing the last value of the old-weight series (including WC costs) with the first value of the new-weight series (excluding WC costs). Apply this linking factor to the historical series so the entire time series is expressed on a consistent basis. This preserves the full length of the historical data for trend fitting but introduces estimation error in the linking factor itself, particularly if the WC cost removal and the weight change interact in ways that are not purely additive.

Option 2: Re-estimate from the revised series only. Discard the pre-transition data and begin fitting the wage trend from the January 2027 release forward. This avoids the splicing issue entirely but produces a much shorter consistent time series. A log-linear trend fitted to two or three years of quarterly data will have wide confidence intervals, and the resulting wage trend selection will be heavily influenced by the particular economic conditions of that short window. If the post-transition period happens to coincide with a labor market slowdown or acceleration, the estimated trend could be temporarily unrepresentative.

Neither option is clearly superior. In practice, most state filing actuaries will likely splice the series for the first two to three years after the transition, gradually shifting weight toward the revised series as it accumulates sufficient history. The key is to document the linking methodology in the filing support and ensure the approach is transparent to regulators and rate review actuaries.

Impact of 2025 Occupational Weights

The weight update is potentially more consequential than the WC cost removal for certain industry groups. The ECI holds employment shares fixed so the index measures pure compensation changes within occupations and industries. When those shares are updated, the index rebalances to reflect the current structure of employment rather than the structure that prevailed in 2021.

Between 2021 and 2025, several shifts reshaped the employment landscape with direct relevance to workers' compensation:

  • Healthcare and social assistance employment grew from 20.4 million to approximately 22.1 million, an 8.3% increase that outpaced total nonfarm growth of roughly 5%. Higher weight on this sector, where WC frequency is moderate but medical severity is elevated, could push the ECI-derived wage trend upward.
  • Construction employment grew from 7.5 million to approximately 8.0 million. Construction is among the highest-hazard WC sectors, and its wage growth has outpaced the national average due to chronic labor shortages. More weight on construction wages would further elevate the index.
  • Manufacturing employment remained roughly flat at 12.8 million. With slower growth than services, manufacturing's share of the ECI weight declines, potentially reducing the influence of a sector where WC claim frequency has been declining faster than the national average.
  • Professional and business services continued expanding, with information technology and consulting subsectors growing at double-digit rates. These are low-frequency WC sectors but high-wage sectors, so their increased weight pulls the national compensation average upward.

The net effect of these shifts is likely to increase the ECI-implied national wage growth rate modestly, perhaps by 0.1 to 0.3 percentage points relative to what the 2021 weights would have produced. For pricing actuaries using the ECI for countrywide indemnity severity trending, this means the post-transition series may run slightly hotter than the historical series even after accounting for the WC cost removal, particularly for goods-producing industry groups where the weight reduction works in the opposite direction.

Alternative Wage Trend Benchmarks

The ECI transition is an opportunity to reassess whether it should remain the sole wage trend benchmark. Several alternative data sources offer advantages in geographic granularity or industry specificity:

Quarterly Census of Employment and Wages (QCEW). Published by BLS from state unemployment insurance tax records, the QCEW provides average weekly wages by state, county, and 6-digit NAICS industry. It covers virtually all employees covered by state UI programs, making it the most comprehensive measure of actual wages paid. Its primary limitation is a six-month publication lag, and it measures average wages rather than wage growth at constant employment mix, so shifts in the occupational composition of a state's workforce can distort the implied wage trend.

State Average Weekly Wage (AWW) series. Many state labor departments publish their own AWW series, often calibrated to WC benefit calculations. These series have the advantage of directly reflecting the wage base that drives statutory indemnity benefits. However, coverage definitions vary by state, publication frequencies differ, and data revisions can be substantial.

NCCI unit statistical report data. NCCI's own payroll data from carrier unit statistical reports captures actual payroll reported for WC premium computation. This is the most directly relevant measure for WC pricing because it reflects the same payroll base on which premiums are calculated. The limitation is that it takes 18 to 24 months for unit stat data to become sufficiently mature for trending purposes.

Bureau of Economic Analysis (BEA) compensation series. The BEA publishes quarterly estimates of compensation of employees by industry as part of the National Income and Product Accounts. These series are comprehensive but heavily revised and are better suited to macro-level analysis than to state-specific WC pricing.

Credibility Weighting National and State Data

The ECI transition reinforces a principle that has always applied to WC wage trending: no single national index perfectly represents the wage dynamics of any individual state. A Bühlmann credibility framework can blend the national ECI-derived trend with state-specific QCEW or AWW data to produce a more stable, jurisdiction-appropriate wage trend selection.

The credibility-weighted wage trend takes the form:

wselected = Z × wstate + (1 − Z) × wnational

where Z is the credibility assigned to the state-specific data based on the volume and stability of the state's wage observations. Large states with deep labor markets and stable industry composition (California, Texas, New York, Florida) may warrant credibility weights of 0.6 to 0.8 on their state-specific wage trend. Smaller states with volatile employment bases may receive credibility of 0.2 to 0.4, relying more heavily on the national ECI complement.

This matters more now because the ECI discontinuity will temporarily reduce confidence in the national trend component. During the first few years after the transition, the national series will be fitting on a shorter consistent history. States with strong QCEW or AWW data may find it appropriate to temporarily increase the credibility weight on their state-specific series until the revised national ECI accumulates sufficient post-transition history.

Consider a concrete example. A pricing actuary filing in Ohio observes that the state's QCEW average weekly wage grew 3.8% over the latest four quarters, while the national ECI wages-and-salaries reading is 3.4%. With a credibility weight of Z = 0.5 on the state data, the selected wage trend is 0.5 × 3.8% + 0.5 × 3.4% = 3.6%. The indemnity severity trend factor over a 2.5-year period becomes (1.036)2.5 = 1.0917, compared with 1.0865 using the national figure alone. The difference is small but accumulates across large state filing portfolios.

Implications for Classification Ratemaking

The weight update has an additional, often overlooked, impact on classification-level ratemaking. WC premium is calculated by class code, and each classification code maps to specific industries and occupations. When the ECI rebalances its occupational weights, the implied wage trend for specific class groupings shifts. A construction classification group that was overweighted relative to its actual current employment will see its imputed wage growth rate decrease; a healthcare classification group that was underweighted will see it increase.

Pricing actuaries performing classification relativities analysis, whether for NCCI or for independent bureaus, should evaluate whether their current classification-level wage trend assumptions remain appropriate after the weight transition. This is particularly relevant for class codes where the indemnity severity trend materially diverges from the countrywide average, such as heavy construction (typically above-average wage growth), white-collar professional services (above-average), and retail trade (below-average).

What WC Pricing Actuaries Should Prepare Now

The December 2026 data release is seven months away. Actuaries who will be preparing loss cost filings with effective dates in late 2027 or 2028 should take several preparatory steps:

  1. Document the current ECI baseline. Record the full historical ECI wages-and-salaries time series through the September 2026 release (the last quarter before the transition) to establish the pre-transition baseline for future splicing calculations.
  2. Build the QCEW comparison. Begin compiling state-level QCEW average weekly wage data for the jurisdictions in your filing portfolio. Having two or three years of quarterly state QCEW data alongside the ECI will make the credibility blending framework immediately usable when the transition occurs.
  3. Quantify the WC cost removal effect. Using ECEC data from recent years, calculate what the ECI benefits component would have looked like without the WC insurance cost component. This approximation provides the basis for a linking factor when the actual revised series is published.
  4. Evaluate the weight shift by industry group. Compare the 2021 OEWS employment distribution with available 2025 Current Employment Statistics data to estimate how the weight update will shift ECI readings for your most significant class groupings.
  5. Engage with bureau staff early. For NCCI filings, discuss the transition methodology with NCCI's actuarial staff before the December 2026 data arrives. For independent bureau states (New York, New Jersey, Massachusetts, Delaware, Pennsylvania, and others), coordinate with the respective bureau on their planned approach to the ECI discontinuity in their standard trend exhibits.

The BLS changes are methodological, not economic. Underlying wage growth is unchanged; only the measurement is shifting. But in a line of business where the trend assumption is one of the largest single drivers of the indicated rate change, measurement methodology matters as much as the underlying economics. Getting the transition right is the difference between a trend selection that accurately reflects prospective wage inflation and one that carries a systematic bias from a broken time series.

Further Reading

Sources