How to Read a WCIRB Experience Rating Worksheet, Line by Line

Every California submission worth pricing comes with an experience rating worksheet, or should. It is the one document that tells you, in the bureau's own numbers, how this employer's claims compare to what the bureau expected. Most brokers glance at the mod and move on. Most underwriters read a little further. Very few people can explain every line, and that gap is where mistakes get made in pricing, in proposals, and in conversations with prospects who want to know why their mod went up.

This post walks through the worksheet the way we teach new underwriters to read it. Where we cite a rule, it comes from the California Workers' Compensation Experience Rating Plan published by the WCIRB. The plan is amended every year, so treat the specific numbers below as current to the plan effective September 1, 2024 and check the version in force on the rating effective date in front of you.

Start with the header: who, when, and which years

The top of the worksheet identifies the employer, the bureau number, and the rating effective date. The rating effective date is the single most important field, because it determines everything else: which experience period applies, which expected loss rates and D-ratios are used, and which eligibility threshold governs.

Under the plan, the experience period is three years, beginning four years and nine months before the rating effective date and ending one year and nine months before it. In practice that means the most recent completed policy year is excluded. A mod effective in 2026 is built from policy years that ended in 2022 through 2024, roughly speaking. This is the first thing to explain to a prospect who says "but we had a great year last year." Last year is not in the mod yet. It will show up next time, and the year after, and the year after that.

The worksheet also reflects the eligibility test. An employer is experience rated only if its expected losses over the experience period meet the plan's eligibility threshold, which is adjusted annually. The September 1, 2024 plan set it at $10,400. If a prospect is just above the line, small changes in payroll or class mix can move them in and out of experience rating altogether, which is worth knowing before you promise anyone anything.

The payroll and expected loss section

The body of the worksheet is organized by policy year, and within each year by classification code. For each class code and year you will see the payroll reported by the carrier and the expected loss rate for that class. Multiply payroll by the expected loss rate, divide by one hundred, and you get expected losses for that line.

Expected loss rates are not pure premium rates. They are set specifically for the experience rating plan and reflect the losses the bureau expects to see, at the valuation point used for rating, per hundred dollars of payroll in that class. The numbers will not reconcile to a rate sheet and are not supposed to.

Next to each line is a D-ratio. The D-ratio is the portion of expected losses for that classification that is considered primary, at the primary threshold that applies to this employer. Multiply expected losses by the D-ratio and you get expected primary losses for the line. The plan notes that expected primary losses are shown for information and are not themselves used in the final formula; they are there so you can see how the bureau is splitting the expectation.

Sum the lines and you have the two totals that drive the calculation: total expected losses, labeled E, and total expected primary losses, labeled Ep. Expected excess losses, Ee, are simply E minus Ep.

When you review this section, ask whether the payroll by class looks right against the current census and whether the class codes match what the business actually does. Audited payroll from three years ago is what shows up here. If the employer has grown or changed operations since, the mod is rating a business that no longer quite exists.

The actual loss section

Below the expected losses is the claim listing. Each claim is shown with its policy year, claim number, injury date, status (open or closed), and incurred loss, which is the sum of indemnity incurred and medical incurred as reported by the carrier at the valuation date for that policy year.

This is where the primary threshold does its work. Since 2017, the California plan has used a variable split point: the primary threshold depends on the size of the employer's total expected losses. The WCIRB describes a table of more than ninety values, ranging from $4,500 for the smallest experience-rated employers to $75,000 for the largest. Larger employers get a higher threshold, which means a larger share of each claim counts as primary.

Since 2019 there is also a $250 adjustment on every claim. The plan defines actual primary losses on a claim as zero if the incurred amount is $250 or less; as the incurred amount minus $250 if it falls between $250 and the primary threshold; and as the primary threshold minus $250 if the incurred amount exceeds the threshold. The practical effect is that very small claims drop out entirely, and every other claim is capped at the threshold less $250.

Everything above the threshold is excess. Under the current California plan, actual excess losses receive no weight at all. That is the single most misunderstood feature of the worksheet, and it changes how you talk to a prospect about a large claim.

The worksheet sums actual primary losses across all claims into a total labeled Ap.

The formula, and why frequency matters more than severity

The California experience modification is calculated as:

Mod = (Ap + Ee) / E

Actual primary losses plus expected excess losses, divided by total expected losses. Read it slowly, because it explains most of the behavior that confuses people.

The excess portion of the calculation is entirely expected. Whatever the employer actually experienced above the primary threshold is replaced with what the bureau expected to see above the threshold. So a single catastrophic claim affects the mod only up to the primary threshold less $250. A $400,000 claim and a $40,000 claim can contribute the same primary loss for an employer whose threshold is below $40,000.

What drives the mod, then, is frequency. Ten claims at $8,000 each contribute far more primary loss than one claim at $80,000 for most employers. The WCIRB said as much when it introduced the variable split point: the formula was designed to put more weight on frequency than on severity, on the theory that frequency is the better predictor of future losses and the thing an employer can most directly control.

For a PEO or broker, this is the core of the conversation with a prospect. The large claim they are worried about is not the problem. The eleven small claims from the same warehouse are.

Credibility, in the older sense of a weight applied to excess losses, is no longer a separate line. Before 2017 a portion of actual excess losses counted, scaled by employer size. The variable split point replaced that: size now enters through the primary threshold.

Take a hypothetical California employer with total expected losses of $60,000 over the experience period, expected primary losses of $18,000, and therefore expected excess losses of $42,000. Suppose the bureau's table assigns this employer a primary threshold of $12,000.

Now suppose the claim listing shows five claims with incurred amounts of $200, $3,000, $9,000, $15,000, and $90,000. Applying the plan's rule: the $200 claim contributes zero; the $3,000 claim contributes $2,750; the $9,000 claim contributes $8,750; the $15,000 and $90,000 claims each contribute the threshold less $250, or $11,750. Actual primary losses total $35,000.

Mod = (35,000 + 42,000) / 60,000 = 1.28.

Notice that the $90,000 claim and the $15,000 claim counted identically. Notice also that if the two smallest claims had been avoided, actual primary losses would drop to $32,250 and the mod to roughly 1.24. Frequency moves the number; severity above the threshold does not. These figures are invented for illustration and are not from any real worksheet.

What to check before you rely on the number

The worksheet is only as good as the data the carrier reported. Before a mod goes into a proposal or a pricing model, we check a few things every time.

First, reconcile the claim listing to the loss runs. A claim on the loss run that is missing from the worksheet, or a claim on the worksheet with a different incurred value than the loss run shows at the same valuation date, is a reporting problem that may be correctable through the carrier and the bureau.

Second, look at open claims. The incurred value on an open claim includes the carrier's reserve. If a claim has since closed for less, the mod may be revised when the carrier reports the updated value, depending on the plan's rules for revisions. If a claim is still open and the reserve looks thin, the next worksheet may be worse.

Third, confirm the payroll and class codes against the current census and the ACORD 130. A payroll figure that is too low inflates the mod; a class code with the wrong expected loss rate distorts both the expectation and the threshold.

Fourth, consider ownership history. The plan has specific rules on ownership changes and combinations of entities. If the prospect recently reorganized, the worksheet in your hand may not be the one that will apply.

Why this matters for PEO pricing

A PEO on a large-deductible master policy may find that the carrier does not apply the client's mod at all; the PEO's own experience governs the policy. Even then, the worksheet is essential. It is the best available evidence of the client's loss frequency, which is exactly what a PEO's loss fund has to absorb. A client with a 1.28 mod driven by eleven small claims is a different risk from a client with a 1.28 mod driven by one bad surgery, and the worksheet tells you which one you are looking at.

This is why ModIQ in HALO IQ parses the worksheet rather than just storing it. The payroll, expected loss rates, D-ratios, claim listing, and primary threshold come out as structured data, reconcile against the parsed loss runs, and feed the mod projection and the loss pick. When the WCIRB worksheet is not in the submission, ModIQ can prepare the request for it. The goal is not to replace the bureau's number; it is to make sure every line of it has been read.