The experience modification rate is the single number on a workers' comp quote that clients most often ask about and least often understand. A 0.82 gets celebrated, a 1.35 gets blamed on the agent, and neither reaction says much about what the number measures. If you write or service commercial workers' comp, you need to be able to explain the mod in plain terms, read the worksheet line by line, and know which parts of it the client can influence and which parts they cannot.
This guide walks through the mechanics as NCCI describes them in its published material, flags where independent rating bureaus differ, and ends with a practical way to walk a client through a worksheet. It deliberately avoids promising any result. Nobody, including the carrier, can tell a client what next year's mod will be until the rating organization calculates it.
Manual rating groups employers by classification and charges each class an average rate per $100 of payroll. Experience rating refines that. In NCCI's words, it recognizes differences among qualifying employers "by comparing the experience of individual employers with the average employer in the same classification."
The result is a factor applied to the manual premium:
Two points clients routinely miss:
The mod is also not the only pricing tool. Schedule rating, premium discounts, deductibles and carrier-specific programs all sit on top of or alongside it. When a client says "my rate went up," check whether it was the mod, the class rates (loss costs and carrier multipliers), the payroll estimate, or something else before you start explaining the mod.
Every worksheet compares two things: what the employer's losses were expected to be, and what they actually were.
Expected losses come from payroll and class codes. Each class has an Expected Loss Rate (ELR), the expected losses per $100 of payroll for that class. The worksheet multiplies the ELR by the payroll divided by 100. NCCI's own example uses an ELR of 3.50 on $250,000 of payroll: 3.50 × 2,500 = $8,750 in expected losses for that line.
Each class also has a D-ratio (discount ratio), the share of expected losses that is expected to be primary loss. Multiply expected losses by the D-ratio and you get expected primary losses.
This is why payroll and classification accuracy matter so much. If an employer's payroll is assigned to the wrong class, the expected losses on the worksheet are wrong, and so is the comparison. Misreported payroll on a single policy year can distort the mod for three years.
Actual losses come from the unit statistical reports that insurers file with the rating organization for each policy, under NCCI's Statistical Plan or the bureau's equivalent. Each claim is reported with its incurred value, meaning what has been paid plus what is reserved for future payment. On the worksheet, indemnity and medical are combined into one figure per claim, and each claim is marked open or closed.
Two practical consequences:
Very large claims are capped for rating purposes by a state per-claim accident limitation, which differs by state. The portion above the limit is non-ratable and does not enter the mod at all. NCCI's plan also has a multiple-claim limitation for a single accident involving several workers, and a maximum debit modification that caps how high a mod can go.
This is the part of the formula agents most need to understand, because it explains why ten small claims can hurt more than one large one.
Each ratable claim is divided at a split point:
Primary losses enter the formula at full weight. Excess losses enter at only partial weight, through a "weighting value" (shown as Wt on the worksheet) that is small for small employers and rises with employer size. The larger the employer, the more its own excess experience counts.
The split point used to be one national number in NCCI states. Since NCCI's 2024 plan changes, each NCCI state has its own split point approved as part of that state's filing, and the worksheet summary shows the state's approved split point. Read the figure on the client's worksheet instead of quoting a number from memory.
NCCI's own illustration compares two similar employers in the same class: one with a single $50,000 loss, one with ten losses totaling $50,000. Statistically, the employer with ten claims is the one likely to cost more in the future, because any one of those ten accidents could have been the expensive one. The plan reflects this by counting every dollar of each small claim as primary, while a large claim contributes only its first slice at full weight.
So when a client says "we only had one bad claim," that may be good news for the mod. When they say "nothing serious, just a lot of little stuff," that is the pattern the formula is built to penalize.
Most NCCI states have approved the Experience Rating Adjustment (ERA). Under it, the mod includes only 30% of the primary and excess portions of each medical-only claim, which NCCI describes as a 70% reduction. On the worksheet, medical-only claims are injury type 6; the claim detail shows the full amount and the summary carries the reduced value.
The stated purpose of ERA is to reduce the incentive for employers to pay small medical claims out of pocket rather than report them. That is worth repeating to clients: unreported claims create coverage, compliance and employee-relations problems, and under ERA the rating payoff from hiding them is smaller than many owners assume. ERA is a state-approved feature, so confirm it applies in the client's state and on the client's worksheet.
The formula also adds a ballast value to both the actual and expected sides. It is a stabilizer that keeps a single loss from swinging a small employer's mod too far. Ballast rises with expected losses. You do not need to calculate it by hand, but it explains why a small employer's mod moves less than its raw loss ratio would suggest.
NCCI states that the experience period is "generally based on three years of payroll and loss data," though it can contain less than 12 months or as much as 45 months of data depending on policy terms. A policy is included if its effective date falls between 57 months and 21 months before the rating effective date.
In practice, for an annual policy renewing January 1, the mod usually uses the three policy years that started four, three and two years earlier. The policy that is currently in force, and the one that just expired, are not in it. NCCI explains why: insurers do not have to report a policy's data until 18 months after its inception, which gives them time to value the claims.
This lag is the key point for client conversations. A safety improvement or a run of good claim results this year will not show in the mod for roughly two years, and a bad year will stay in the calculation for three. Clients who expect next year's mod to reward this year's changes will be disappointed, and agents who imply otherwise create an expectation they cannot control.
The rating effective date is the date the mod can first be applied, usually the policy anniversary. NCCI usually calculates the mod 60 to 90 days before that date. A mod can be issued as preliminary (state's new rating values not yet approved), final, or contingent (audited payroll or loss data still missing from a carrier). A contingent mod is a prompt to chase the missing unit report or audit with the prior carrier.
Not every state uses NCCI. According to NCCI's published guide:
|
Arrangement |
States |
|
NCCI calculates the mod under NCCI's plan |
Most states, including Florida and others where NCCI is the rating organization |
|
Independent bureau, but participates in NCCI interstate rating |
Minnesota, New York, Wisconsin |
|
NCCI plan applies, but the state's own bureau produces intrastate mods |
Indiana, Massachusetts, North Carolina |
|
NCCI plan does not apply; state's own plan and bureau |
California, Delaware, Michigan, New Jersey, Pennsylvania |
|
Monopolistic state fund with its own rating |
North Dakota, Ohio, Washington, Wyoming |
Rules in independent-bureau and monopolistic states can differ from NCCI's in the experience period, split treatment and eligibility. Always check the bureau that rates the client's state before explaining a specific worksheet. Texas, for example, has adopted the NCCI plan with state-specific exceptions set by the Texas Department of Insurance.
An employer with exposure in only one participating state gets an intrastate mod. An employer with exposure in two or more participating states gets one interstate mod that blends the experience from all of them. Exposure in a non-participating state (such as Pennsylvania or California) is rated separately by that state's bureau.
Once issued, a mod is revised only in limited circumstances, such as a reporting error by a carrier, a claim later found non-compensable, a subrogation recovery reported by the carrier, a classification correction, or an ownership change. A reserve that simply drops after the valuation date generally affects the next mod through the next unit report rather than revising the current one. Confirm the specific rule with the carrier or the rating organization before telling a client a revision is possible.
Ownership rules are where agents most often get surprised, especially with contractors that reorganize, buy competitors, or start new entities.
For agents, the practical rule is simple: when a client tells you about a merger, sale, new LLC, or change in partners, get the ERM-14 handled promptly. A new entity formed by the same owners will not start fresh at 1.00 just because it has a new name and FEIN, and finding that out at audit is far worse than finding it out at binding.
Get the current worksheet from the carrier or the rating organization's portal (NCCI's is Riskworkstation). Then take the client through it in this order.
The mod's mechanics are covered in depth in ABTC's experience mods, classifications and audit course, and the wider program design questions (deductibles, dividends, retros) in our article on workers' comp policy design. For broader product knowledge, see ABTC's workers' compensation training.
Agents cannot change the formula, but they can help clients understand the operational inputs behind it: how injuries get reported, how quickly employees get appropriate care, whether there is modified duty available, how the employer communicates with the adjuster, and whether payroll records support the class codes. Those are employer processes, and they are exactly what the Workers' Compensation Training & Certification program covers for the people who run them.
If your client's HR or operations staff are the ones reporting injuries, talking with adjusters and managing restricted duty, the Workers' Compensation Training & Certification program gives them a structured, self-paced course on injury reporting, the claims process, medical restrictions and return to work, safety, and how premiums are determined. It includes an optional exam for the Certified Workers' Compensation Administrator designation and team enrollment with a manager progress view. It is professional training, not state insurance CE.
Anything below 1.00 is a credit mod, meaning the employer's losses were better than expected for its classes and size. A 1.00 means the experience matched expectations or the employer is not experience-rated at all. Above 1.00 is a debit. "Good" also depends on context: many construction owners and general contractors look at the mod during prequalification and set their own cutoffs, so the number that matters for bidding is the one in the owner's requirements. Read the client's bid documents rather than assuming a threshold.
The rating organization compares the employer's actual losses with expected losses for its payroll and class codes over the experience period, usually three policy years. Each claim is split into primary and excess portions at the state's split point. Primary losses count fully; excess losses count partially based on employer size. A ballast value is added to stabilize the result. The mod is adjusted actual losses divided by adjusted expected losses. In NCCI states, NCCI calculates it; some states use their own bureau.
Because the good year is probably not in the calculation yet. Under NCCI's plan, the experience period generally covers three policy years ending about a year before the rating effective date, and insurers are not required to report a policy's data until 18 months after it starts. The policy currently in force and the one that just expired are usually excluded. A strong year typically shows up in the mod about two years later and then stays for three ratings.
Yes, but in most NCCI states they count less. Under the Experience Rating Adjustment, only 30% of the primary and excess portions of a medical-only claim enter the calculation, a 70% reduction. The full claim still appears on the worksheet detail. Lost-time claims are not reduced this way and also tend to be larger. Confirm whether ERA is approved in your client's state, since it is a state-approved plan feature.
Often, yes. Under NCCI's rules, the experience of a sold business generally transfers to the new owner, and businesses with more than 50% common majority ownership are combined into one mod. Ownership changes must be reported to the insurer in writing within 90 days, usually on the ERM-14 form. NCCI can revise the current mod and up to two prior mods after an ownership or combination change. A new entity does not automatically start at 1.00.
In most states NCCI does. California, Delaware, Michigan, New Jersey and Pennsylvania have their own plans and bureaus. Indiana, Massachusetts and North Carolina apply NCCI's plan, but their bureaus produce intrastate mods. Minnesota, New York and Wisconsin have independent bureaus that participate in NCCI interstate rating. North Dakota, Ohio, Washington and Wyoming are monopolistic state-fund states with their own systems. Check the client's state bureau for its rules.
Your Name
Copyright ABTrainingCenter.com 2026
