Most California restaurants overpay workers' comp — usually on the wrong class code.

Workers' compensation is the largest single insurance line item most California restaurants carry, and it is also the most frequently mispriced. The reason is structural: WC premium is built from three multiplied numbers — payroll, a class-code rate, and the Experience Modification factor (the X-Mod) — and a restaurant can be wrong on any of the three without ever noticing. The payroll can be split across the wrong codes. The class code itself can be the wrong one for the actual operation. And the X-Mod can be carrying claims that were closed years ago or that never should have hit the mod in the first place. Each of these is a lever, and on a typical full-service restaurant payroll, a single misapplied lever moves real money.

The most common error is the class code. Until September 1, 2024, almost every California restaurant was rated on a single legacy code — 9079 — that bundled fast food, full-service, caterers, and bars into one rate. The Workers' Compensation Insurance Rating Bureau (WCIRB) eliminated 9079 and split it into six new codes, each meant to track the loss experience of a narrower segment. Many restaurants are still riding policies that were bound under the old code logic, or that were re-coded mechanically at renewal without anyone checking whether the assigned code actually matches how the restaurant operates. A counter-service taqueria rated as a full-service restaurant, or a full-service restaurant with a real bar rated as a plain restaurant, is paying a rate built for a different risk.

The second common error is payroll allocation. A restaurant owner who also waits tables, a manager who covers the line on a short night, a server who runs food and also tends a small bar — these create payroll that can legitimately sit in more than one code, and the split matters because the codes carry different rates. When the entire payroll is dumped into the highest-rated code that touches the operation, the restaurant overpays. When it is allocated honestly with proper records, it does not.

The third lever is the X-Mod, and it is the one owners understand the least. The mod is supposed to reward a restaurant with better-than-average claims experience and penalize one with worse. But mods are calculated off data the WCIRB receives from carriers, and that data is frequently stale or wrong — open reserves on claims that have actually settled for less, claims attributed to the wrong policy period, medical-only claims that should have been reported at a reduced value under California's medical-only adjustment. A mod built on bad data overcharges a restaurant that actually runs a safe operation.

None of this is exotic. It is the ordinary friction of a high-turnover, high-frequency class of business being rated inside a system that assumes someone is checking the inputs. Most of the time nobody is. The re-rate is simply the act of checking — pulling the code, the payroll split, and the loss runs behind the mod, and seeing whether the number on the policy is the number the operation should actually be paying.

Effective September 1, 2024, the WCIRB retired the single legacy restaurant code 9079 and replaced it with six codes, each intended to capture a distinct food-service operating model. The codes are: 9058 — food and beverage employees of hotels and short-term lodging; 9080 — full-service restaurants (table service, where guests order and are served at the table and pay after eating); 9081 — restaurants not otherwise classified; 9082 — caterers; 9083 — fast food and fast casual (counter-order or limited service); and 9084 — bars and taverns (operations where alcohol service is the primary activity rather than incidental to food).

At the moment of the split, all six codes share a single advisory pure premium rate. The WCIRB separated the codes before it had enough independent loss data to assign each one a different rate, so for now they price the same on the advisory rate. That does not make the code choice cosmetic. Carriers apply their own loss-cost multipliers, schedule credits and debits, and underwriting appetite by code, and a carrier that wants full-service restaurants may price 9080 sharply while pricing 9084 bar risk conservatively. More importantly, the WCIRB is collecting the separate loss data now, and within two to three years the rates will diverge. A restaurant sitting on a code that does not match its operation will be on the wrong side of that divergence when it happens.

The code is assigned on the actual operation, not on the owner's preference or on how the business describes itself. A 'fast casual' concept with table service and a liquor license may underwrite as full-service. A 'restaurant' where bar sales dominate may underwrite as a bar. The carrier's auditor makes this determination at the year-end audit, and an audit re-code can move the rate after the fact — sometimes producing an unexpected additional premium bill, sometimes a credit. Getting the code right at binding, on an honest description of the operation, avoids the audit surprise in both directions.

Governing classifications and standard exceptions complicate this further. Clerical office staff, outside salespeople, and drivers can sometimes be split into their own standard-exception codes at much lower rates, but only with proper payroll records that segregate their hours. A bookkeeper who works in a back office and never touches the floor does not belong in the restaurant code. Capturing these splits legitimately is one of the most reliable ways a re-rate lowers premium without changing anything about how the restaurant runs.

The premium itself is built the same way for every restaurant: the carrier's filed rate per $100 of payroll, multiplied by your payroll divided by 100, multiplied by your X-Mod. California is an open-rating state — the WCIRB publishes an advisory pure premium rate (the pure loss cost), and each carrier files its own rate above that to cover expenses and margin. The advisory pure premium approved for policies incepting on or after September 1, 2025 averaged $1.52 per $100 of payroll across all industries, an 8.7% increase over the prior year, and the WCIRB has filed for a further 10.4% average increase for September 2026. Restaurants sit above that all-industry average — they are a high-frequency class — so the restaurant rate per $100 runs higher than the statewide mean, and the trend is upward. That is precisely why the two controllable inputs, the class code and the X-Mod, are where the savings live: you cannot negotiate the filed rate down, but you can make sure payroll sits in the correct codes and the mod reflects accurate loss data.

What a re-rate cannot do is invent a lower rate out of nothing. The advisory pure premium rate is set by the WCIRB, and the carrier's loss-cost multiplier is filed. The re-rate works by putting payroll in the correct codes, confirming the operation is on the right code, and making sure the X-Mod reflects accurate loss data — not by negotiating a rate that does not exist. What yours costs depends on your payroll, your code, and your loss history; we shop it across our carrier appointments and quote the real number same-day on a complete submission.

The Experience Modification factor is the single largest controllable input to a restaurant's workers' comp premium. It is a multiplier applied to the manual premium that compares the restaurant's actual claims experience to the expected claims experience for similarly-sized operations in the same class code. A mod of 1.00 is exactly average. Below 1.00 is a credit that lowers premium; above 1.00 is a debit that raises it. The WCIRB calculates the mod annually using three prior policy years of payroll and loss data — specifically, it excludes the most recent (still-developing) year and uses the three years before that.

Restaurants are a high-frequency class, and frequency drives the mod harder than severity does. The X-Mod formula deliberately weights the number of claims more heavily than the dollar size of any single claim, on the theory that frequent small losses predict future losses better than one large fluke does. The practical consequence for a restaurant is counterintuitive: three small slip, strain, or burn claims hurt the mod more than one larger claim of the same total dollar value. A kitchen that produces a steady drip of minor injuries — cuts on the line, slips on a wet floor, lifting strains in the walk-in — can carry a worse mod than a kitchen that had one bad accident and an otherwise clean record.

Because frequency is the driver, the controllable response is claims-frequency management: non-slip footwear programs, cut-glove discipline at prep stations, proper lifting and walk-in procedures, prompt incident reporting, and — critically — a return-to-work program that brings injured employees back on modified duty quickly. California's WC system reduces the loss value that hits the mod when an injured worker returns to light duty rather than staying out, so a real return-to-work program both helps the employee and protects the mod.

Mod errors are common and worth auditing. Because the mod is built on carrier-reported data, it frequently carries open reserves on claims that have actually closed for far less, claims booked to the wrong policy period, or medical-only claims reported at full value rather than at California's reduced medical-only valuation. Each of these inflates the mod above what the true loss experience justifies. Pulling the loss runs that feed the mod, comparing reserves to actual paid-and-closed amounts, and correcting reporting errors with the WCIRB can lower a mod that is overstated — which lowers premium on every subsequent policy until the data ages off.

The mod also matters beyond premium. Many California landlords, lenders, franchisors, and government or institutional contracts require a mod at or below 1.00 as a condition of doing business. A restaurant bidding for a catering contract with a hospital, a university, or a municipal venue can be disqualified by a mod over 1.00 regardless of price. The mod is, in that sense, a business credential as much as a pricing input.

A workers' compensation policy is bound on estimated annual payroll and trued up to actual payroll at a mandatory audit when the policy term ends. The estimate at binding is just that — an estimate — and the audit reconciles it against what the restaurant actually paid in wages over the year. If actual payroll came in higher than the estimate, the audit produces an additional premium bill. If it came in lower, the audit produces a return. For restaurants, where headcount swings with the seasons and wage floors keep rising, the audit is rarely a non-event.

The single most common budgeting trap in restaurant WC is the conservative payroll estimate. An owner who lowballs the estimated payroll at binding gets a lower deposit premium and feels good about the number — and then gets a large, unbudgeted audit bill twelve months later when the actual payroll is reconciled. The bill is not a penalty or a carrier trick; it is the premium that was always owed on the payroll that was actually run. Honest payroll estimates at binding cost slightly more up front and eliminate the year-end surprise entirely. We estimate payroll to reflect known wage increases and seasonal staffing rather than to produce a flattering deposit number.

California minimum-wage increases flow directly into WC premium because premium is a percentage of payroll. The statewide minimum, the fast-food-specific wage under AB 1228 (applicable to covered national fast-food chains), and the local minimum-wage ordinances in cities like Los Angeles, West Hollywood, San Diego, and others all raise restaurant payroll — and WC premium rises at the same percentage, at the same rate and mod. A 10% wage increase is, all else equal, roughly a 10% WC premium increase. The audit captures it whether or not the estimate anticipated it.

What gets counted as payroll for WC matters and is frequently mishandled. Reported tips, in most California WC treatments, are included in payroll for premium calculation, which surprises operators who assume tips are outside the wage base. Overtime is generally included at the straight-time portion only — the premium-portion of overtime (the extra half on time-and-a-half) is typically excluded, but only when the payroll records separate it, which many restaurant payroll systems do not do by default. Owner and officer payroll has specific inclusion rules and elective-coverage options. Getting these right at audit, with records that support them, is the difference between an audit that confirms the estimate and one that inflates it.

Audit disputes are winnable but require records. When an auditor reclassifies payroll into a higher code, includes payroll that should have been excluded, or applies the wrong code to the operation, the result can be challenged — but only with documentation: segregated payroll by job function, time records that support standard-exception splits, and a clear operations description. A restaurant that keeps clean payroll records by job class is in a position to contest a bad audit; one that does not is at the auditor's discretion. Part of the re-rate is making sure the recordkeeping will hold up when the audit comes.

A 'ghost policy' is a minimum-premium workers' compensation policy written for a business that certifies it has no employees subject to coverage — typically a sole proprietor or a partnership whose only workers are the owners, who can elect out of their own coverage in California. The policy exists primarily to produce a certificate of insurance for a client, a landlord, or a general contractor who requires proof of WC before they will do business. It covers, in effect, nobody. For a genuine no-employee operation, that is legitimate and inexpensive. For a restaurant, it is almost never appropriate — and relying on one is a serious exposure.

California requires workers' compensation for any business with employees, and 'employees' includes essentially everyone a restaurant pays to work: cooks, servers, dishwashers, bussers, hosts, and managers, whether full-time, part-time, or seasonal. There is no waiver. A restaurant cannot decline WC for its staff, cannot have employees sign away their right to coverage, and cannot substitute a health plan or an indemnity agreement for statutory WC. The only people who can elect out are qualifying owners and corporate officers, and only of their own coverage — not their employees'.

Operating a restaurant with staff on a ghost policy — or with no policy at all — is uninsured for the exposure that actually exists. If an employee is injured, the ghost policy does not respond, because it was written on the representation that there are no covered employees. The restaurant then faces the injured worker's medical and indemnity costs directly, plus California's uninsured-employer penalties, plus a potential stop-order shutting down operations until coverage is in force, plus personal liability that can reach the owners. The California Uninsured Employers Benefits Trust Fund may pay the injured worker and then pursue the employer for full reimbursement.

Misclassifying employees as independent contractors is the adjacent trap, and California's ABC test (codified in AB 5 and its successors) makes it very hard to clear. A line cook, a delivery driver, or a dishwasher labeled a '1099 contractor' is almost certainly an employee under the ABC test, which presumes employment unless the worker is free from control, performs work outside the usual course of the business, and is engaged in an independently established trade. Cooking food in a restaurant's kitchen is the usual course of a restaurant's business — prong B fails on its face. Misclassified workers are uncovered for WC, and the exposure surfaces exactly when one of them is injured.

The honest version is straightforward and not expensive relative to the risk: a real workers' compensation policy that covers the actual staff, with owners and officers electing out of their own coverage if they choose. The cost of that policy is a fraction of a single uninsured injury claim, and it produces a certificate that is actually backed by coverage. Part of what a re-rate confirms is that the policy in force is a real policy for the people who actually work there — not a ghost certificate that will fail at the moment it is needed.

Workers' compensation is a no-fault, statutory coverage: when a covered employee is injured in the course of employment, the policy pays regardless of who was at fault, and in exchange the employee generally gives up the right to sue the employer in tort. For a restaurant, that bargain matters, because the alternative — defending negligence lawsuits over kitchen injuries — would be far more expensive and far less predictable. The coverage has two parts. Part One pays the statutory benefits to the injured worker: all reasonable medical treatment for the injury (with no policy dollar limit on medical), temporary disability payments that replace a portion of lost wages while the worker recovers, permanent disability benefits if the injury leaves a lasting impairment, supplemental job-displacement vouchers for retraining where the worker cannot return to the same job, and death benefits to dependents in a fatal case.

Part Two — Employers' Liability — covers the employer for injury claims that fall outside the no-fault WC system: certain third-party-over actions, dual-capacity claims, and consequential injury claims by family members. The standard limits are modest, and any restaurant carrying an umbrella policy will want the umbrella to sit over the employers' liability limit, which requires the underlying limits to meet the umbrella's attachment requirement. This is a frequently-missed coordination point that we check as part of placing the program.

Medical treatment in California WC runs through a Medical Provider Network (MPN) in most cases, which directs injured workers to network physicians and is a meaningful cost-control tool for the restaurant — both for the quality of the treatment and for the speed of return to work. Carriers with strong MPNs and active nurse case management resolve restaurant claims faster and at lower ultimate cost, which protects the mod. The cheapest premium at binding is not always the lowest total cost of risk if the carrier's claims handling lets minor injuries develop into mod-damaging losses.

Palm Trinity Insurance Services, Inc. is a California commercial insurance brokerage based in Chino, founded by Brian Kong in 2013 and run with partner Ron Ng, managing over $5 million in commercial premium across more than 4,900 customers, 97% of them in California. Restaurants are a core vertical, and workers' compensation is the line where the re-rate work pays off most reliably. We shop the WC across our carrier appointments — including carriers with real restaurant appetite and strong MPNs — rather than defaulting to the incumbent's renewal.

The re-rate process is concrete. We pull the current declarations and the WC class codes, three to five years of loss runs, the current X-Mod worksheet from the WCIRB, and a payroll breakdown by job function. We check whether the assigned code matches the operation, whether payroll is allocated correctly across codes and standard exceptions, and whether the mod is built on accurate loss data. Where the code is wrong, the payroll is mis-split, or the mod carries stale reserves, we correct it and re-shop the corrected submission. On a complete submission, we quote the real number same-day for admitted appetite. What yours actually costs depends on your payroll, your class code, and your loss history — we'll tell you the real figure rather than a flattering one.