Workers' Compensation Experience Modification Rate Calculator

How is the experience modification rate actually calculated?

At the top it's simple: actual losses divided by expected losses. Everything interesting happens in how both sides get adjusted before the division.

Expected losses come from your audited payroll. Each class code's payroll is divided by 100, multiplied by an expected loss rate, and the result multiplied by a D-ratio to split out the expected primary portion. The calculator's defaults run three class codes across $3,500,000 of payroll, producing $108,148 of expected losses, of which $30,867 is expected primary.

Actual losses are split claim by claim at the primary threshold. Anything below it is primary; anything above is excess. At a $18,500 split point, a $42,000 claim contributes $18,500 primary and $23,500 excess. Across six claims totalling $87,700, that's $56,000 primary and $31,700 excess.

Then both sides get the same two adjustments. Ratable excess is excess losses multiplied by the weighting factor — 30% here — which discounts large claims. Stabilizing value is expected excess losses multiplied by one minus the weighting factor, plus the ballast. That value is identical on both sides of the fraction; it exists to stop a single claim from throwing the mod far from 1.00.

Numerator: actual primary + stabilizing value + actual ratable excess = $161,607. Denominator: expected primary + stabilizing value + expected ratable excess = $150,148.

Mod: 1.076. On $285,000 of manual premium, that's $21,750 of surcharge.

Why do most EMR calculators give a different number?

Because the formula that circulates online omits the stabilizing value, and that changes the answer.

The version you'll find on most calculator sites reads: actual primary, plus the weighting factor times actual excess, plus ballast, all over expected primary, plus the weighting factor times expected excess, plus ballast. It looks symmetrical and it isn't the bureau's formula.

The real structure adds a stabilizing value to both sides, and that value contains the undiscounted remainder of expected excess losses — the portion the weighting factor doesn't ratably include. Leave it out and the denominator shrinks by that remainder, which inflates the mod.

At the calculator's defaults the correct mod is 1.076. The shortcut formula produces 1.119 — overstated by 0.043 mod points, which on $285,000 of manual premium is $12,250 of premium that isn't owed.

There's a clean structural test for whether a formula is right. Because the stabilizing value and the expected ratable excess add back to the full expected excess, the denominator must always collapse to expected losses plus ballast — $108,148 plus $42,000 equals $150,148 here, which the calculator's reconciliation line confirms. Any formula that doesn't collapse that way isn't the NCCI formula.

Why is my mod above 1.00 when my losses are below expected?

This is the result that confuses people most, and the calculator's defaults reproduce it exactly.

Total incurred losses are $87,700 against $108,148 expected — a raw ratio of 0.811. Losses are 19% under budget. And the mod is 1.076, a surcharge.

The reason is that the rating formula doesn't care much about your total. It cares about the primary portion, because primary losses reflect frequency and frequency is what predicts future losses. Actual primary is $56,000 against expected primary of $30,867 — 81% over. Meanwhile the excess losses, where you're well under, get multiplied by a 0.30 weighting factor and largely disappear.

Six claims, two of which broke the split point, is a frequency profile. The formula reads it as a company that has accidents often, and prices accordingly, regardless of the fact that none of them were catastrophic.

The practical consequence runs against instinct. A single large claim is less damaging to your mod than several moderate ones, and a scatter of small claims is the worst outcome of all. Which means the highest-leverage safety work is usually preventing the routine strains, slips and lacerations that never make it into an incident review — not the rare severe event that everyone remembers.

Check your own worksheet for this. If your actual-to-expected ratio looks good and your mod doesn't, frequency is your problem.

How much does claim distribution matter, holding total losses constant?

Enormously — the calculator quantifies it directly, and the range is larger than most loss-control budgets. Take the same $87,700 of incurred losses and redistribute it three ways.

• All in one claim: primary is capped at $18,500 and the remaining $69,200 rides the weighting factor. Mod: 0.901.
• As actually distributed across six claims: 1.076.
• None exceeding the split point — say a dozen claims all under $18,500: every dollar is primary. Mod: 1.224.

That's a 0.323 mod-point swing on identical loss dollars, worth $91,945 of premium at this premium level. Nothing about the money changed; only how it was distributed across claims.

Two things follow. First, the split point is the single most important number on your worksheet after the losses themselves, and it moves — NCCI has raised it repeatedly, so check the current value rather than reusing last year's. The calculator takes it as an input for that reason.

Second, claim count is a lever you can pull. Aggressive early medical management that keeps a strain from becoming a lost-time claim doesn't just reduce that claim's cost, it can keep the claim out of the frequency count entirely. Return-to-work programmes work on the same mechanism.

What does this calculator not do?

Three limits, and the first is the important one.

• It assumes NCCI methodology. A number of states run independent rating bureaus with their own formulas — Washington, for example, uses a credibility-weighted structure where credible actual primary loss is actual primary times primary credibility plus expected primary times one minus that credibility. That is a genuinely different calculation, not a variant. California, Pennsylvania, Delaware and others also operate independently. If you're rated by an independent bureau, use this to understand the concepts, not to reproduce your worksheet.
• It doesn't source the rating values. Expected loss rates, D-ratios, weighting factors, ballast and the split point all come from bureau tables that vary by state and change annually. They're inputs here because hardcoding them would make the calculator wrong within a year. Pull them from your own experience rating worksheet, which shows every one of them.
• It doesn't handle the finer adjustments. Medical-only claim reduction, where a proportion of medical-only losses is discounted before rating, applies in most NCCI states and isn't modelled. Neither is loss limitation, the treatment of large deductible policies, or combination of experience across commonly owned entities.

What it does do is show you the mechanism — which side of the fraction each number lands on, and why frequency moves the answer more than severity. If your broker can't explain your worksheet in those terms, that's worth knowing before renewal.