How a workers compensation premium is assembled
Workers compensation is rated on payroll, not on headcount, revenue or property values, because payroll is the best available proxy for exposure to injury. The unit is $100 of payroll, a convention inherited from an era of manual rating, and it survives because it makes rates readable: a $4.50 rate means $4.50 of premium for every $100 of wages, or 4.5% of payroll.
Every dollar of payroll is assigned to a class code describing the work being done. In most states those codes and their rating rules come from the National Council on Compensation Insurance; California, New York, Pennsylvania, Delaware, Michigan, Minnesota, New Jersey, Texas and Wisconsin run independent bureaus with their own code sets. The code follows the business operation rather than the individual job, with specific exceptions — clerical office employees, outside sales staff and drivers are separately rateable in most jurisdictions precisely because their injury exposure differs so sharply from the shop floor.
Multiply each class payroll by its rate and sum, and you have manual premium: what an average employer in your industry would pay. Everything after that adjusts for you specifically. The experience modification factor compares your own claim history against that average. Schedule rating lets an underwriter recognise safety programmes and controls the ex-mod does not capture. A premium discount reflects that expenses do not scale with account size. The expense constant covers the fixed cost of issuing and auditing a policy. State assessments fund second-injury funds and administration.
The result is provisional. Policies are written on estimated payroll and audited afterwards, so the number this calculator gives is a budgeting figure, not a bill.
What the experience mod actually measures
The ex-mod is the part employers argue about, and it is widely misunderstood. It is not a safety score and it is not a discount. It is a ratio comparing your actual losses over a three-year window against the losses expected for a business of your size and class mix, credibility-weighted so that a small employer's random bad luck does not swing the answer as far as a large employer's genuine trend.
Three features of its construction have practical consequences.
Frequency counts more than severity. The experience rating formula splits each claim into a primary portion, counted in full, and an excess portion, heavily discounted. Ten $5,000 claims therefore raise a mod far more than one $50,000 claim, even though the dollars are identical. That is deliberate — frequency predicts future losses better than severity does — and it is why controlling small recurring injuries moves a mod faster than avoiding one catastrophe.
It lags. The experience period is typically the three policy years ending one year before the rating effective date, so the most recent year is excluded entirely. Improvements you made this year will not appear in your mod until two renewals from now. Equally, a bad year keeps costing you for three rating periods after it ends.
Reserves count, not just payments. Open claims enter the calculation at their reserved value. A claim reserved at $80,000 that eventually settles for $20,000 has been inflating your mod for years. Reviewing open reserves before the unit statistical filing date is one of the few genuinely effective levers an employer has, and it costs nothing but attention.
Arithmetically the mod is simply multiplicative: a 1.15 mod adds 15% to manual premium, a 0.85 mod removes 15%. On the $25,050 manual premium in the example below, the difference between those two is $7,515 a year on identical payroll and identical rates.
Below a state-specific premium threshold an employer is not experience rated at all and is assigned 1.00. Growing past that threshold is therefore a genuine risk event: a business with poor claim history can see its premium jump on the renewal at which experience rating first applies, with no change in payroll or rates.
Worked example: $800,000 of payroll across two class codes
A small manufacturer has $500,000 of shop payroll at a $4.50 rate and $300,000 of clerical payroll at $0.85. Its experience mod is 1.15, the underwriter has applied a 10% schedule credit, there is no premium discount, the expense constant is $200 and the state adds 5% in assessments. There are 12 employees.
- Class 1 manual premium. 500,000 ÷ 100 = 5,000 units × $4.50 = $22,500.00.
- Class 2 manual premium. 300,000 ÷ 100 = 3,000 units × $0.85 = $2,550.00.
- Total manual premium. 22,500 + 2,550 = $25,050.00.
- Apply the experience mod. 25,050 × 1.15 = $28,807.50. The mod costs $3,757.50.
- Apply schedule rating. 28,807.50 × (1 − 0.10) = $25,926.75.
- Premium discount. None, so the figure is unchanged.
- Add the expense constant. 25,926.75 + 200.00 = $26,126.75.
- Add state assessments. 26,126.75 × 1.05 = $27,433.09.
- Cost per $100 of payroll. 27,433.09 ÷ 8,000 units = $3.4291.
- Cost per employee. 27,433.09 ÷ 12 = $2,286.09.
Two things in that chain are worth noticing. The all-in rate of $3.4291 is below the class 1 rate of $4.50 and above the class 2 rate of $0.85, because it is a payroll-weighted blend adjusted by everything downstream — which is why comparing a quoted class rate against a competitor's all-in rate compares two different quantities.
And the split between codes is doing enormous work. Move the $300,000 of clerical payroll into class 1 by mis-coding it and the manual premium becomes 800,000 ÷ 100 × 4.50 = $36,000, which flows through to a final premium of (36,000 × 1.15 × 0.90 + 200) × 1.05 = (37,260 + 200) × 1.05 = $39,333.00. That is $11,899.91 more, a 43.4% increase, from a classification error rather than any change in the business. Class assignment deserves more scrutiny than the rate itself.
Manual premium on $500,000 of payroll, and the effect of the mod
| Rate per $100 | Manual premium | At 0.75 mod | At 1.00 mod | At 1.25 mod |
|---|---|---|---|---|
| $0.50 | $2,500 | $1,875 | $2,500 | $3,125 |
| $1.00 | $5,000 | $3,750 | $5,000 | $6,250 |
| $2.50 | $12,500 | $9,375 | $12,500 | $15,625 |
| $5.00 | $25,000 | $18,750 | $25,000 | $31,250 |
| $7.50 | $37,500 | $28,125 | $37,500 | $46,875 |
| $10.00 | $50,000 | $37,500 | $50,000 | $62,500 |
| $15.00 | $75,000 | $56,250 | $75,000 | $93,750 |
The gap between the 0.75 and 1.25 columns is always exactly half the manual premium, because 1.25 − 0.75 = 0.50. On a $10.00 rate that is $25,000 a year on identical payroll.
Comparing quotes and reading the cost per $100
Compare quotes on the all-in cost per $100 of payroll, not on the manual rate and not on the total premium. The all-in figure absorbs every difference in class assignment, schedule credit, discount and expense loading into one number that is directly comparable, and it is the figure to track year on year.
When it moves, decompose the change before reacting. Only four things can drive it: the rates filed for your class codes, your ex-mod, the underwriter's schedule credit, or a change in the mix of payroll between codes. Payroll growth alone changes the premium but not the rate per $100, so a rising all-in rate on flat operations always points at one of those four.
Treat schedule rating as the negotiable part. Rates are filed and the ex-mod is computed by a bureau, so neither is arguable in a quoting conversation. Schedule credits are underwriter judgement within a filed range — commonly up to 25% either way — and they respond to documented safety programmes, return-to-work policies, training records and management commitment. An employer with real controls and no evidence of them is leaving credit on the table.
Beyond premium, note the two adjacent costs a comp calculation does not include. The employer's own payroll tax burden runs alongside it, at 7.65% below the Social Security wage base, and the FICA payroll tax calculator sizes that; unemployment insurance is a third payroll-based charge on much smaller wage bases, covered by the FUTA and SUTA calculator. Adding the three together gives the real loading on a wage dollar, which for a hazardous class can exceed 15%.
If you are on the carrier side rather than the buyer side, the same premium feeds the underwriting ratios: the loss ratio calculator and the combined ratio calculator measure whether that premium was adequate.
What this estimate does not capture
- State payroll definitions. Overtime is commonly reported at straight-time value only, severance and some fringe benefits are excluded, and executive officers are subject to minimum and maximum payroll limits that vary by state.
- Loss cost multipliers. In loss cost states the bureau files pure loss costs and each carrier applies its own multiplier, so two quotes on the same class code can differ before any credit is applied.
- Deductible credits and retrospective rating. Large accounts frequently use deductible programmes or retro plans in which final cost depends on actual losses, not on the guaranteed-cost formula shown here.
- Terrorism and catastrophe charges. Usually separate per-$100 loadings rather than a percentage of premium, so folding them into the assessment field is an approximation.
- Multi-state operations. Each state has its own rates, rules, assessments and often its own bureau; a multi-state policy is several calculations, not one.
- Minimum premium. Every policy has a floor, and a very small account will be charged that rather than the computed figure.
- Monopolistic states. North Dakota, Ohio, Washington and Wyoming require coverage from a state fund with its own rating method, and the NCCI sequence does not apply.
The audit, not the quote, decides what you pay
Workers compensation policies are written on estimated payroll and audited after expiry. The auditor recalculates premium on actual payroll and actual class assignments, and any difference is billed or refunded. Two things routinely produce a large audit bill. The first is underestimating payroll at inception, which produces a bill for the shortfall on top of the following year's deposit. The second is uninsured subcontractors: where a subcontractor cannot produce a certificate of insurance, their payroll is typically added to yours and rated at your class code, which can add a substantial premium for work you did not think you were insuring. Keep certificates on file for every subcontractor and keep clerical payroll separately recorded, because an auditor who cannot verify a split will assign the whole amount to the governing class.
Why the system is built this way
Workers compensation is the oldest form of social insurance in the United States, and it rests on a bargain struck state by state in the 1910s: employees give up the right to sue their employer for workplace injury, and in exchange receive medical care and wage replacement without having to prove fault. Premium rating exists to fund that promise and, secondarily, to price safety.
The pricing signal is the reason the system has the shape it does. If every employer in a class paid the same rate, a business with a poor safety record would be subsidised by its careful competitors. Experience rating converts each employer's own history into a multiplier, which turns injury prevention into a measurable financial return. That return is often larger than it looks, because the ex-mod effect compounds across three rating years — a single claim influences the mod for three consecutive periods, so its cost is roughly three times its effect on any one year's premium.
The corollary is that the direct premium is only part of the cost of an injury. Production disruption, retraining, overtime to cover the absence and administrative time are all uninsured, and industrial safety literature has long argued that these indirect costs are a multiple of the insured ones. The multiple varies enormously by industry and injury type and is not a number to quote as fact, but the direction is not in dispute, and it means a premium-based estimate understates what an accident actually costs.
For an employer building a fully loaded labour rate, the useful practice is to express comp as its all-in cost per $100 of payroll and add it alongside FICA and unemployment taxes to the wage rate. The salary to hourly rate calculator gives the wage base to load, and the burdened figure is what belongs in a job estimate or a bid.