One Number, Two Jobs: What EMR Can and Cannot Do for a Construction Executive

Experience modification rate earns the attention it gets. A mod above 1.0 can cost you access to hospital systems, university systems, municipal work, and data center developers who use it as an automatic filter, and a single pursuit lost that way is worth more than a year of premium difference. Any executive treating that number seriously is doing so for sound commercial reasons.

The difficulty is that the same number is also being asked to describe how exposed the business is to a catastrophic loss, and that is a job it was never built to do.

Two lines, two entirely different risk drivers

Workers’ compensation experience rating is weighted toward frequency. The rating system takes three years of claims, weights them against payroll, and produces a number that rewards a company for having fewer of the routine injuries that make up the bulk of construction claims. General liability, umbrella, and excess are more heavily driven by severity, which means they respond to the rare catastrophic event, the litigation that follows it, and the size of recent verdicts in your jurisdictions.

The market makes that division visible. Marsh's US rate index has been reporting US casualty increases in the high single digits, and the same index shows the increase running several points higher once workers' compensation is excluded from the calculation, with umbrella and excess climbing far faster than either. Workers' compensation is the line pulling that average down. WTW's Insurance Marketplace Realities describes the same structural picture, with casualty firming on litigation pressure while other lines soften.

Percentages move with the cycle. The division underneath them does not, because frequency and severity are produced by different work, different decisions, and different controls. Your mod speaks to the first category, and nothing on a standard safety report speaks to the second.

The gap is measurable, and it has been measured

Construction has driven recordable rates down over the past decade while remaining among the most lethal industries in the country. BLS put the private industry recordable rate at 2.3 cases per 100 full time workers in 2024, continuing a long decline, while the Census of Fatal Occupational Injuries recorded 1,032 deaths among construction and extraction workers at a rate of 12.6 per 100,000, roughly four times the all industry rate. Those two measures do not move together, which is the entire problem.

The research says the same thing directly. The statistical analysis led by Matthew Hallowell for the Construction Safety Research Alliance examined roughly 3.26 trillion work hours and found no discernible association between recordable rate and fatalities, along with evidence that recordable injuries occur almost entirely at random and that comparing companies or projects on recordable rate is statistically unsound in most practical circumstances. The follow on work, Moving Beyond TRIR, proposes measuring high energy control instead.

Read that as a scope statement rather than an indictment. Recordable rate and mod describe frequency history in a standardized and auditable way, which is genuinely useful. They were never built to tell you whether the pick scheduled for Thursday has a verified control on it, and the industry's mistake has been treating silence on that question as an answer. We have written about that failure mode before in Zero Recordables Do Not Predict Safe Work and The Myth of Compliance.

What sits in the gap

The measures that speak to severity are not exotic, and most contractors already generate the underlying data without assembling it.

The first is control verification on high energy work, meaning the percentage of tasks involving the energy sources that actually kill people where someone confirmed a direct control in the field rather than assuming it on a plan. This is the approach behind energy based safety and the High Energy Control Assessment, and it is the one measure in common use aimed squarely at the events that end lives.

The second is hazard closure time, both the median and the age of the oldest open item, which indicates whether the organization responds to what its people report and is difficult to manufacture for a submission. Quality matters more than volume here, which is the argument in PTPs May Be Hiding Your SIF Potentials and in Near Misses in Safety.

The third is subcontractor performance tracked at the trade level, because that is where a general contractor's severity exposure concentrates and where most insurance submissions are thinnest. If you want the field level version of what you are looking for, Your Next Fatality Has Already Happened covers how these events announce themselves before they occur.

None of these replace the mod. They describe the exposure the mod is silent about.

Why this is becoming a renewal conversation

Underwriters have started asking for evidence that sits outside the loss run, and Marsh's own commentary on current conditions notes that underwriting outcomes are increasingly driven by exposure quality and risk management rather than by rate alone. A carrier pricing a severity exposed line cannot get what it needs from a frequency based history, so it has to look somewhere else.

For a contractor with a genuinely disciplined operation, that is an opportunity. A submission documenting controlled severity exposure gives an underwriter something to work with on the lines that are actually moving, while a clean loss run and nothing else leaves the carrier to price on assumption.

The practical version

Keep the mod work going, since bid eligibility is real money. Add a short second set of forward looking measures aimed at severity, review them monthly at the leadership level rather than quarterly at the safety level, and bring them to the broker as a documented trend.

The reason to run both is that they answer different questions. One tells you where you stand commercially against a three year history you can no longer influence. The other tells you whether the exposures capable of ending a project, a relationship, or a life are under control today.