Craft Ceiling Hits EMR: What the AI Construction Boom Means for Your Mod
ENR's 2026 Top 400 survey shows most contractors can't fill craft positions. When you hire fast to keep pace, your injury frequency rises. So does your mod.
The craft ceiling in construction, where more than 80% of Top 400 contractors can't fill skilled positions (ENR, May 2026), directly inflates the EMR. Workers in their first six months on the job carry roughly 46% higher injury frequency than those with two or more years of tenure. Those injuries feed primary losses into the experience rating worksheet, pushing the mod higher for three full policy years.
ENR published its 2026 Top 400 list this week. The headline: $671.4 billion in construction revenue, up 11.8% year over year (ENR, May 2026). The real story sits three tables deeper. More than 80% of surveyed contractors report they can't fill craft positions. That craft ceiling isn't a staffing inconvenience. It's an EMR (Experience Modification Rate) problem.
The work isn't the issue. Data center construction starts hit $77.7 billion in 2025, a 190% increase over the prior year (ConstructConnect, 2026). Another $88 billion in projects sits in the pipeline for the next six months alone. Cloud providers have committed roughly $7 trillion in AI infrastructure spending over the coming decade. The demand is real. The experienced crews to meet it safely aren't.
New tenure, new frequency
When contractors can't find journeyman electricians and seasoned carpenters, they fill spots with whoever's available. NCCI (the National Council on Compensation Insurance) research on workforce composition shows that short-tenure workers carry roughly 50% higher injury frequency than the broader workforce (NCCI, 2019). A peer-reviewed study of more than 58,000 workers found that those with less than six months on the job had an injury rate of 10.2 per 100 person-years, compared to 7.16 for workers with more than two years of tenure (Breslin et al., 2007). For workers under 30, that gap widened to a 46% higher relative risk.
These aren't abstract public health numbers. Every one of those injuries generates a workers' comp claim. And every claim exceeding the medical-only threshold lands on the experience rating worksheet as a primary loss.
How green crews shift the mod math
The experience rating formula splits losses into primary and excess components. Primary losses carry full weight in the formula. Excess losses get dampened. That split means frequency matters far more than severity when calculating the EMR.
Consider a concrete scenario. A crew of 10 experienced framers might generate one lost-time claim per year. Replace three of them with first-six-month hires, and expected frequency shifts. One additional lost-time claim at $22,000 contributes roughly $18,500 in primary losses. Two additional claims in a single year add $37,000 in primary losses to the worksheet. For a mid-size specialty trade contractor running $4 million in annual payroll, that kind of frequency increase can move the mod six to ten points.
That higher mod doesn't reset at the next renewal. It stays on the worksheet for three full policy years after the claim's valuation date, multiplying against every dollar of premium in that window.
The data center corridor is where this hits hardest
The states absorbing the most data center work are the same states where craft labor was already thin before the AI boom. Virginia. Texas. Georgia. The Carolinas. Specialty trade contractors working NCCI class codes 5190 (electrical wiring within buildings), 5183 (plumbing), and 5403 (commercial carpentry) face expected loss rates that already price in elevated frequency for those classifications. When actual loss experience outpaces expected losses, the mod climbs.
BLS data for 2024 puts the total recordable incidence rate for specialty trade contractors at 3.4 per 100 full-time workers (BLS SOII, 2024). That's the industry average. Contractors staffing up with workers who carry 46% higher injury frequency are running well above that baseline. Most won't see the damage until the next mod worksheet arrives.
In our reviews of Southeast contractor worksheets, the pattern is consistent: contractors that ramped headcount fastest over the prior 18 months showed the steepest primary loss increases. The 65% of Top 400 firms expecting data center work to grow this year (ENR, May 2026) should be asking what that growth does to their loss experience, not just their backlog.
What an audit would check
An audit examines whether claim activity on the worksheet reflects the actual workforce composition during the experience period. It checks whether classification assignments match the work crews perform, not just the contract scope. It verifies that primary loss values correspond to current claim reserves and that reserve levels reflect realistic outcomes rather than stale carrier estimates from early in a claim's life.
If you've grown headcount by 20% or more in the past two years, your mod may already be reflecting that decision. Request a free worksheet review and we'll show you where.