Experience Modification Rate (EMR) for Manufacturers: What It Costs and How to Lower It

Your experience modification rate is a multiplier applied to your workers’ compensation premium, based on how your claims history compares to the average employer in your class. A 1.00 is average. Below 1.00 you pay less than average; above 1.00 you pay more.

What it costs in dollars: a $10 million manufacturer with $2.5 million in shop payroll and a manual rate of $4.50 per $100 typically has a manual premium around $112,500. At a 1.00 mod that is what they pay. At a 1.25 mod they pay $140,625, $28,125 a year more, for the same payroll, the same work and the same coverage.

Over five years that is $140,000, and that is only the insurance cost. The bigger number is usually the work you did not win. Workers’ compensation is normally the largest line in a manufacturing insurance program, and the mod is the one input on it you can change.

How your EMR is calculated: split point, D-ratio, ballast and weighting factor

The NCCI experience rating formula compares what you actually lost against what an employer your size, in your class, would be expected to lose.

Mod = Adjusted Actual Losses ÷ Adjusted Expected Losses

NCCI’s own worked example: $243,120 ÷ $242,971 = 1.00. Five components do the work.

1. Expected losses. Calculated from your payroll and an Expected Loss Rate assigned to your class code:

Expected losses = ELR × (payroll ÷ 100)

NCCI’s example: an ELR of 1.413 against $3,125,350 of payroll gives $44,161.

2. The primary and excess split. Every claim is split at a dollar figure called the split point. The portion below it is primary; the portion above is excess.

At an $18,500 split point:

A state accident limitation caps how much of any single claim counts. NCCI’s illustration uses $200,000, and states move it. Illinois cut its per-claim limitation from $485,500 in 2023 to $206,000 in 2024.

3. Primary losses are weighted heavily; excess losses are not. This is the mechanical heart of the formula. The primary portion enters at close to full weight. The excess portion is heavily discounted through the weighting factor (W). Frequency drives your mod. Severity mostly does not.
   
4. Medical-only claims are discounted to 30% of value before they enter the calculation. A $6,000 medical-only claim counts as $1,800.
5. Ballast and weight stabilise the result. The ballast value (B) and the weighting factor stop one large loss from swinging a small employer’s mod. Both grow with employer size, so a large employer’s mod moves freely and a small employer’s is damped. NCCI’s example uses a weight of 0.16 and a ballast of $45,900. The D-ratio converts total expected losses into expected primary losses.

The experience period: three years, with a one-year lag. Data comes from policies effective 21 to 57 months before the rating effective date, reported 60 to 90 days before renewal. The policy you are on right now does not affect the mod you are about to receive.

Eligibility. You have to be big enough to be experience rated at all. NCCI’s example thresholds are $14,000 in audited premium within the most recent 24 months, or $7,000 in average annual premium across the three-year period. Figures vary by state.

Why frequency hurts more than severity in the experience rating formula

This is the single most useful thing to understand about the mod, and it is counter-intuitive. Compare two manufacturers with identical payroll and identical total losses of $150,000, at an $18,500 split point.

Same money out the door. Manufacturer B’s mod will be dramatically worse, because all $150,000 landed in the heavily weighted primary layer while Manufacturer A’s severe claim was mostly discounted into the excess layer.

What this means in practice. Preventing one catastrophic injury is a moral obligation. Preventing ten minor injuries is a financial strategy. A program that eliminates strains, minor lacerations and slips moves your mod faster than one aimed only at catastrophic risk. Do both.

It also means claim management matters as much as claim prevention. A $9,000 medical-only claim counts as $2,700. The same claim, once a single lost-time indemnity payment is made, counts at full value. Getting an injured employee back on modified duty before the claim converts is worth several times the cost of accommodating them.

NCCI class codes for manufacturing, and how misclassification distorts your mod

Expected losses come from the Expected Loss Rate attached to your class code. Put payroll in the wrong code and the denominator of the mod is wrong before a single claim is filed. Too low, and an average loss year produces an above-average mod. Too high, and you pay manual premium you never owed.

The manufacturing codes that come up most often:

One correction, because it is misapplied constantly: 2143 is Winery & Drivers. It is not a fruit or vegetable canning code. Canning is classified state by state. California uses WCIRB’s “Fruit or Vegetable Preserving including Canning,” Delaware uses 0113. A cannery sitting in 2143 is misclassified.

The classification errors that actually move a mod:

• Office, sales and driver payroll in a shop code. The most common and most expensive one. It inflates manual premium and expected losses together.
• Multiple operations under one code. A shop that stamps and machines may support both 3400 and 3632 on a payroll split, but bureaus require verifiable records. An estimate will not hold at audit. More in our guide to machine shop and metal fabrication insurance.
• Food plants running several processes. A dairy that also bakes, or a meat processor that also slaughters, is not a single-code operation. See food and beverage manufacturing insurance for how those splits get written.
• Breweries with taprooms and fleets. 2121 covers drivers and distributing stations, but retail and restaurant exposure is classified separately. See our brewery insurance guide.

NCCI assigns each class to a hazard group, A through G, with A the lowest severity potential. The code-to-letter mapping is not published, and neither is any nationwide rate-per-$100-of-payroll table by class code and state. Treat any site publishing one as unverified

How much an EMR costs: premium by payroll and mod

Illustrative, using a $4.50 per $100 manual rate for a mid-hazard manufacturing class. Manual rates vary by class code and state. The mod is a straight multiplier, so the proportional effect holds at any rate.

Read across the $2.5 million row. Moving from 1.25 to 0.85 saves $45,000 a year, a full-time safety coordinator’s salary, funded entirely by the position paying for itself.

For where workers’ compensation sits inside a full program, see how much manufacturing insurance costs.

Typical and standard EMR thresholds: what counts as a good mod

No bureau publishes a grading scale. The thresholds that matter are the ones your customers use.

1.00 is the number to plan around. It appears more often than any other figure in supplier qualification questionnaires, and it works as a pass/fail gate rather than a scored input. A 1.02 and a 1.40 fail the same screen.

On the 1.25 figure. One broker states it plainly: mods above 1.25 often disqualify businesses from bidding on major contracts. That is broker guidance, not a universal rule. Customers set their own cutoffs, and some set none. Treat 1.25 as a widely reported trigger point, not a published standard.

Why the bidding consequence is larger than the premium

Aerospace, automotive and anyone whose people set foot on a construction site get screened on their mod. Your competitor at 0.88 and you at 1.12 are not just paying different premiums. You are looking at different bid lists. The mod is a commercial metric that happens to be calculated by an insurance bureau.

Nine ways manufacturers lower their EMR

1. Build a return-to-work program. The highest-return action available. It converts lost-time claims into medical-only claims, discounted to 30% of value. Write real modified-duty job descriptions in advance, signed off by your occupational health provider, ready before the injury rather than after.
   
2. Manage claims and challenge reserves. Your mod runs on incurred losses, paid plus reserved. An open claim carrying a $60,000 reserve counts as $60,000 even if it settles for $12,000. Review open reserves quarterly with the adjuster. Close claims before the valuation date.
3. Audit your class codes. Payroll in the wrong class distorts expected losses in one direction or the other. Office, sales and driver payroll should not sit in a shop code such as 3632.
4. Run a real safety committee. Monthly, minuted, with actions assigned and closed. Underwriters ask for the minutes, and it is one of the few safety controls with a paper trail.
5. Guard your machines and document it. Machine guarding (1910.212) was the tenth most-cited OSHA standard in FY2025; lockout/tagout (1910.147) was fourth. Manufacturing accounts for roughly half of the 26,000-plus workplace amputations recorded between 2015 and 2024, about seven a day nationally. A written guarding audit with photographs and a corrective action log is both a safety control and an underwriting document.
6. Fix ergonomics. Repetitive strain and material handling injuries are the highest-frequency claims in most plants, and frequency is what the formula punishes.
7. Report near-misses. A near-miss program surfaces the conditions producing the frequent minor claims doing the most damage. It costs nothing and it works.
8. Direct care through a medical provider network where your state permits it. Occupational medicine physicians who understand modified duty return people to work faster than emergency rooms do.
9. Verify your unit statistical report. Covered below, because it is the one nobody does.

How long it takes to lower an EMR

Longer than anyone wants, and the reason is the one-year lag.

The practical implication: start now, and expect nothing for two years. Companies that abandon safety programs at eighteen months because “the mod has not moved” abandon them exactly one renewal before they would have worked. The corollary holds too: a bad year takes three years to leave the calculation, so one terrible year prices into three renewals.

How to audit your unit statistical report for EMR errors

Your mod is calculated from data your carrier reported to the bureau on a unit statistical report, the loss and payroll filing due roughly 18 months after policy inception, then updated annually. That data carries errors more often than anyone assumes. Correcting them is free money.

Request the unit statistical report worksheet from the bureau or your broker. Then check four things.

1. Open reserves that should be closed or reduced. The most common and most valuable error. A claim reserved at $75,000 that settled for $9,000 two years ago, never updated, is inflating your mod today.
2. Claims in the wrong policy period. Injury date versus report date confusion moves claims into experience periods where they do not belong.
3. Subrogation and recovery not credited. If a third party reimbursed the carrier, the net figure should be reported, not the gross.
4. Claims that are not yours. Common where a company runs multiple entities, has acquired a business, or shares a name with another employer.

Check the payroll side too. A classification or payroll error moves expected losses, which moves the mod just as surely as a loss error does.

Corrections require the carrier to file a corrected unit statistical report, after which the bureau recalculates and reissues the mod. Most states limit corrections to within 24 months of the original report date, so do this annually rather than when you finally notice.

Check combinability. Multiple entities under common majority ownership may be required to combine into one mod, or may be improperly combined when they should be separate. Both errors happen and both are correctable.

Which states use independent rating bureaus instead of NCCI

NCCI administers experience rating in most states. Eleven run their own bureaus with their own formulas, split points and rules:

Four states run monopolistic funds with no private market: North Dakota, Ohio, Washington and Wyoming. Texas is not an independent-bureau state. It uses NCCI loss costs under a competitive filing framework, and is distinctive instead for being the only state where workers’ compensation is optional for private employers.

On split points. The countrywide uniform split point was $18,500 before NCCI’s 2023–24 methodology update. That update replaced the single countrywide figure with state-specific split points and recalibrated the weight and ballast values, unrevised for over two decades. Broker sources report the current range running roughly $9,500 to $38,000 by state, most NCCI states between $15,000 and $25,000. Treat those as broker-reported rather than bureau-confirmed, and get the figure for your state from your broker.

Multi-state employers get one interstate mod covering all NCCI states, plus separate mods in each independent-bureau state where they operate. Those numbers will differ, sometimes substantially.

Eight ways machine shops lower premium

Split your payroll by class code. The fastest win available, frequently worth 10% to 20% of the comp line.

Report overtime correctly. Only straight-time is rateable in most states. Shops reporting gross overtime overpay until an audit catches it, and audits correct in the carrier’s favour more often than yours.

Document machine guarding. A written audit against 1910.212, with photographs and a corrective action log. Underwriters know the amputation data. Show them you do too.

Document lockout/tagout. A written 1910.147 program with machine-specific procedures and annual employee verification.

Build a return-to-work program. The most reliable way to control the primary loss portion of your mod: it converts lost-time claims into medical-only claims, and medical-only claims are discounted to 30% of value in the NCCI formula.

Restructure deductibles. No property claim under $25,000 in a decade means you are paying a carrier to insure losses you already absorb.

Schedule your tooling accurately. Underscheduled tooling is an uninsured loss waiting; overscheduled tooling is a straight overpayment. Most shops are wrong in one direction and have not checked in years.

Verify your unit statistical report. The loss data behind your mod comes from the carrier and contains errors more often than anyone assumes: open reserves that should be closed, claims coded to the wrong period, subrogation not credited. Challenging them costs nothing but time.

Frequently asked questions

Below 1.00 is better than average. Below 0.90 is strong. Many supplier qualification programs use 1.00 as a hard cutoff, so that is the practical target regardless of what your premium says. Between 1.00 and 1.25 you pay a penalty and answer questions on prequalification forms.

There is no single national figure any more. $18,500 was the last countrywide uniform split point, before NCCI’s 2023–24 methodology update made it state-specific. Broker sources report a current range of roughly $9,500 to $38,000, most NCCI states between $15,000 and $25,000, broker-reported, not bureau-confirmed.

Each claim divides at the split point. The portion below is primary and enters at close to full weight. The portion above is excess, heavily discounted through the weighting factor. A $100,000 claim at an $18,500 split contributes $18,500 primary and $81,500 excess. A state accident limitation caps the total.

Adjusted actual losses divided by adjusted expected losses. Expected losses come from your class code’s Expected Loss Rate times payroll per $100. The D-ratio splits expected losses into primary and excess; ballast and the weighting factor stop one claim swinging a small employer’s mod.

Three years of data, pulled from policies effective 21 to 57 months before the rating effective date. The lag exists because claims need time to develop and carriers need time to report. The policy you are on now does not affect the mod you are about to receive.

Eleven: California, Delaware, Indiana, Massachusetts, Michigan, Minnesota, New Jersey, New York, North Carolina, Pennsylvania and Wisconsin. Four more run monopolistic funds with no private market: North Dakota, Ohio, Washington and Wyoming. Texas uses NCCI loss costs and is not an independent-bureau state.

Only employers meeting the eligibility threshold are experience rated. NCCI’s example figures are $14,000 in audited premium within the most recent 24 months, or $7,000 in average annual premium across the three-year period. Thresholds are set state by state.

Request the worksheet from the bureau or your broker, identify the error (overstated open reserve, wrong policy period, uncredited subrogation, a claim that is not yours) and document it. The carrier files a corrected report and the bureau recalculates. Most states allow 24 months.

It follows the process, not the product. Common ones: 3632 machine shop, 3400 metal stamped goods, 2003 bakery, 2095 meat products (processing, not slaughter), 2065 milk products, 2121 brewery, 2501 mass-production clothing, 2503 custom dressmaking or tailoring. 2143 is winery, not canning.

Around 24 months before improvement appears, and 36 to 48 months for the full effect, because of the one-year reporting lag and the three-year experience period. A bad year takes three years to roll off, so it prices into three renewals.

Less than most people expect. Only the portion below the split point enters the primary layer, and a state accident limitation caps the rest. Ten small claims do more damage than one large one.

Yes, and often more. Aerospace, automotive and construction-adjacent customers screen suppliers on it, and a mod above 1.00 removes you from bid lists regardless of price or capability. The lost revenue usually outweighs the premium penalty.

Get an EMR review from Alliance Risk

Your experience modification rate is calculated from data your carrier reported to a rating bureau, and that data is wrong more often than anyone assumes. Open reserves on claims that settled years ago. Claims coded to the wrong policy period. Subrogation recoveries never credited. Claims belonging to a company with a similar name. Every one of them inflates a number you are paying on and bidding against.

Fixing the number is half the work. The other half is operational: a return-to-work program that keeps claims medical-only, near-miss reporting that surfaces the frequent small injuries doing the real damage, and reserve reviews with your adjuster every quarter. The formula rewards frequency control, not heroics. Insurance pays the claim. The program you run sets your multiplier.

We audit unit statistical reports, challenge what’s wrong, and model the path back below 1.00 honestly, which means telling you that nothing moves for about 24 months, and the full effect takes 36 to 48. Anyone promising faster hasn’t understood the lag.

Send us your mod worksheet and your loss runs. We’ll check them for the four common errors, tell you what a corrected mod would be worth, and lay out the timeline. No charge, and it doesn’t require you to move the account.

Send us your mod worksheet →

Part of the full manufacturing insurance program.