EMR rating explained

It is not an OSHA number

The experience modification rate is a workers’ compensation insurance factor. It does not appear in 29 CFR part 1904, OSHA does not calculate it, OSHA does not publish it, and there is no OSHA standard you can violate by having a bad one.

That matters practically, because the two datasets are different. The 300 log records injuries and illnesses under the six criteria of 1904.7. The mod is built from claims — losses actually paid and reserved by your carrier.

Contractors who assume the two move together are surprised twice a year.

Source

What the mod compares

In outline, and this part is uncontroversial: the plan compares the losses your business actually ran up over a past experience period against the losses a business of your size, in your classification, in your state would be expected to run up. Perform as expected and the modifier lands at 1.00. Better than expected and it falls below 1.00 and cuts your premium. Worse and it rises above 1.00 and multiplies it.

Two structural features are worth knowing before you argue with a broker about them:

• Frequency is weighted more heavily than severity. Plans split each claim into a primary portion and an excess portion so that many small claims move the mod harder than one large one. Five $4,000 claims are worse for you than one $20,000 claim.
• The experience period lags. It excludes the most recent policy year, so the safety programme you started this spring cannot appear in the number a GC reads this autumn. Nothing you do today shows up for over a year.

What actually moves it

Three levers, in the order of how much they change the number per hour of effort.

1. Stop small claims becoming claims. Because frequency dominates, the highest-value work is the first-aid station, the return-to-work programme and the culture that reports a strain on the day rather than three weeks later. A modified-duty assignment that keeps a case off the indemnity side is the single best-paid hour in the office.
2. Close reserves. An open claim sits in the calculation at its reserve, not its final cost. A reserve that has not been reviewed since the year it opened can be holding your mod up on money nobody expects to pay. Ask your carrier for a loss run and go through the open files line by line before the unit statistical date.
3. Check the classification codes. Expected losses depend on class and payroll. Payroll booked into a higher-hazard class than the work justifies raises expected losses — which, counter-intuitively, lowers the mod but raises the premium the mod multiplies. Both errors cost money, in different directions.

Everything upstream of those is ordinary jobsite work: the JHA, the weekly walk, and writing down the near misses that would otherwise have been next quarter’s claims.

Frequency

The one document that is authoritative

Your experience rating worksheet. It is issued for your business, for your state, for a specific rating effective date, and it shows every claim used, the payroll by class code, the expected losses, and the arithmetic that produced the modifier. Your agent can get it; so can you, from the rating bureau.

Read it before you accept a mod, and check three things: that every claim listed is yours, that the payroll by class matches what you reported, and that no claim appears twice. Errors in all three are routine, and a correction is worth more than a year of good luck.

How the mod sits beside the two numbers that are derived from the 300 log is compared in TRIR vs DART vs EMR, and those two you can calculate yourself with the TRIR calculator.

The rule text, section by section

Every paragraph cited above, in the words of the standard itself. Open a section number to read it; the wording is quoted from the current eCFR text of title 29, read on 2026-10-18; “…” marks text left out.