California's $20 fast-food minimum wage: what every study found

On April 1, 2024, California tried something new: a minimum wage for one industry, fast food, set at $20 an hour, four dollars above the general state minimum. Opponents predicted closures, layoffs and kiosks where cashiers used to stand.

What followed was a research arms race. A Berkeley labor center found no job losses. Economists publishing through the National Bureau of Economic Research found about 18,000. An industry-retained consulting firm found 10,700. A Hoover Institution piece claiming roughly 10,000 lost jobs was retracted. Each number circulated on its own, stripped of who produced it and how.

This page puts them side by side. We think wage floors are a poor way to help low-wage workers, and we say so. But a scorecard is only useful if it is honest, so where the evidence cuts against us we say that too.

What AB 1228 did

Governor Gavin Newsom signed Assembly Bill 1228 on September 28, 2023. It came out of a deal: unions dropped an effort to make fast-food corporations liable for their franchisees' conduct, and the industry agreed to pull a wage referendum from the 2024 ballot.

The law covers limited-service restaurants that belong to a national chain of at least 60 establishments, franchisees included. It exempts several categories, among them bakeries that make and sell their own bread, restaurants inside large grocery stores, and outlets in airports, hotels, event centers and theme parks. Small chains and independent restaurants were not covered. Neither are full-service sit-down restaurants.

AB 1228 also created a Fast Food Council inside the Department of Industrial Relations, with industry, franchisee, worker and advocate seats, one public member and two non-voting state members. Starting January 1, 2025, the council may raise the wage each year by the lesser of 3.5 percent or the change in the CPI for urban wage earners. As of spring 2026 it had not approved any increase, so the covered wage is still $20.

The scorecard

Here is every study we could verify against its original source. "Funder / source" names who produced or commissioned the work. That is not a verdict on quality, but readers should know it.

The Berkeley studies, and their moving numbers

The most-cited work comes from Michael Reich and Denis Sosinskiy at UC Berkeley's Institute for Research on Labor and Employment. Their first report, released September 30, 2024, used more than 11,000 Glassdoor salary reports and menu prices from over 1,500 California restaurants, plus a similar number in comparison states. It found average hourly pay up 18 percent, employment holding steady, and prices up 3.7 percent, about 15 cents on a $4 burger. Newsom cited it within weeks.

Two revisions followed. The September 2025 version added QCEW data through the end of 2024. It put the wage gain at 10 to 11 percent in weekly earnings and the price increase at 2.1 percent. The March 2026 version brought in new data: Square payroll records, Advan's phone-location data on how many workers are present at each restaurant, and prices from more than 2,000 restaurants. It found weekly wages up about 11 percent, prices up about 1.5 percent, and no employment decline. Reich called it "the most definitive employment analysis" of the policy to date.

Still, readers should notice what moved. The headline wage gain fell from 18 percent to about 11. The price effect fell from 3.7 percent to 1.5. Some of that reflects a switch from hourly to weekly measures and to new data sources, so the versions aren't strictly comparable. Still, a weekly earnings gain well below the hourly one would leave room for some reduction in hours, a possibility the Shift Project survey did not find but which the later papers don't settle. The Berkeley team also reads its results as consistent with monopsony, the idea that employers have enough wage-setting power that a mandated raise need not cost jobs. That is a real model with real evidence behind it, and we take it seriously in our piece on monopsony.

The case for job losses

The strongest evidence on the other side is an NBER working paper from July 2025 by Jeffrey Clemens of UC San Diego and Olivia Edwards and Jonathan Meer of Texas A&M. They used the Quarterly Census of Employment and Wages, the BLS dataset built from unemployment-insurance filings that covers nearly every employer, and compared California's fast-food sector with fast food in the rest of the country from September 2023, when the bill was signed, to September 2024.

In unadjusted data, California fast-food employment fell 2.7 percent relative to the rest of the country. Adjusting for pre-existing trends raised that to 3.2 percent, and netting out changes in industries the law didn't touch raised it further. Across specifications the estimates ran from 2.3 to 3.9 percent. The median works out to about 18,000 jobs below where the sector would otherwise have been. Other California industries, the authors noted, tracked national trends over the same period, which makes a statewide downturn a less likely explanation.

The newest entry, released in September 2026, may be the most important. Vitor Melo, David Neumark, Liam Sigaud and Gary Wagner went to establishment-level data so they could identify which restaurants were actually covered, the problem that defeats the aggregate series. They found negative employment effects, concentrated among small chains, where the policy also slowed new openings and increased closures. Their summary: the primary effect was "to disadvantage small chains, albeit with modest overall employment effects thus far."

The data fight, and the Hoover retraction

Much of the public argument was about monthly jobs numbers, and it turned on a dull technical point: seasonal adjustment. Restaurant employment rises and falls with the calendar, so a raw January count is usually lower than a raw September count in any year, under any law. Compare unadjusted winter numbers with unadjusted autumn numbers and almost any year looks like a loss.

That is what happened in the spring of 2024. A Wall Street Journal report used figures showing California fast-food employment falling between September 2023 and January 2024. Lee Ohanian built on those figures in a piece on the Hoover Institution's website on April 24, 2024, headlined "California Loses Nearly 10,000 Fast-Food Jobs After $20 Minimum Wage Signed Last Fall." He followed it on June 4 with a second piece claiming another 1,250 jobs lost.

Hoover retracted both pieces. The notices are the same in substance: the author had "interpreted those data as being seasonally adjusted," and after publication the data "were identified as not being seasonally adjusted." To be precise about what this was: two commentary articles on Hoover's website, not a peer-reviewed or NBER paper. It has nothing to do with the Clemens, Edwards and Meer paper, though the two are often confused online.

Even the careful studies argue over data. The monthly jobs survey (CES) is a sample; the QCEW is a near-census but arrives with a lag of several months. Both use the same industry code, which includes many restaurants AB 1228 doesn't cover. The Melo team states the point directly: in aggregate data the 60-establishment threshold "cannot be identified," and there is "measurement error in the industry codes." That is why the establishment-level and phone-location studies matter most. Notably, they still disagree.

Industry reports and surveys

Three other pieces of work get cited often enough to deserve a fair hearing and a clear label.

Berkeley Research Group (February 2025). No relation to UC Berkeley: BRG is a private consulting firm, and its authors were retained by Save Local Restaurants, an industry coalition, though they said the research was conducted independently. Using QCEW data, it found limited-service restaurants lost 10,700 jobs, or 1.9 percent, between June 2023 and June 2024, and that local restaurant prices rose 14.5 percent against 8.2 percent nationally. The main criticism is timing: the window starts about ten months before the wage took effect, so part of the measured decline predates the raise, and a simple before-and-after count can't separate the law from everything else that happened that year.

Employment Policies Institute survey (summer 2024). The group surveyed more than 180 limited-service operators in mid-2024. It reported that 98 percent had raised prices, 89 percent had cut hours, 73 percent had limited overtime or shifts, and 70 percent had reduced staff or consolidated positions. Operator surveys capture something administrative data miss: what owners say they are doing. But the respondents chose to answer and opposed the law, and some questions asked about plans rather than measured outcomes.

UC Santa Cruz (Owen et al.). Led by economics lecturer Stephen Owen with undergraduate co-authors, this working paper combines labor-market data with interviews of franchise and independent restaurant managers, using Santa Cruz as its case study. It reports higher prices, fewer hours, lost overtime and benefits, faster automation, and wage pressure on independents the law doesn't cover. One case describes a Central Valley McDonald's franchisee with 18 locations whose employee hours fell by more than 11.5 percent. Reich has dismissed the paper for making "no attempt to validate their motivated responses with interviews with workers, a control group or objective government data." That critique is fair as far as it goes. Interviews can't estimate a statewide effect. But they can show the margins, like hours, overtime and opening times, where employers adjust first.

What happened to prices

On prices there is more agreement than on jobs. Prices rose; the dispute is by how much.

• Berkeley IRLE: 3.7 percent in the 2024 version, 2.1 percent in the 2025 revision, about 1.5 percent in the March 2026 version. That is about six cents on a $4 item, with roughly half the cost passed through.
• Clemens, Edwards, Meer & Nguyen: 3.3 to 3.6 percent for restaurant meals in California's four CPI metro areas against 17 controls, with no matching rise in groceries or other placebo categories. They attribute part of the effect to spillovers into full-service restaurants and to changes in product quality.
• Berkeley Research Group: 14.5 percent since the law was signed, against 8.2 percent nationally. This compares totals over a long window, not a policy effect against a control.

The CPI study measures something broader than the Berkeley price scrape, so if the effect spilled into restaurants the law didn't cover, a sample of covered chains would miss it. Either way, the customer pays for part of the raise.

What it means for the case against wage floors

It would be easy to pick the NBER paper and declare victory. We won't. Two serious research teams using different data reach different answers on employment, and the honest summary is that the job effect was somewhere between zero and a few percent. That is not the catastrophe opponents predicted in 2023.

But the record supports three points that matter more than the headline number.

First, the cost showed up where theory said it would. The September 2026 establishment-level study finds the losses concentrated in small chains, which saw fewer openings and more closures. A chain with 60 restaurants competes against one with 40. A mandate that applies to one and not the other works as a barrier to growth, and it protects the largest incumbents. That is a familiar story about price floors, and it is a long way from the image of a law that only hurts big corporations.

Second, nobody disputes that consumers paid. Even the most favorable estimate has prices rising and half or more of the cost passed on. A minimum wage is a tax on a particular kind of transaction, collected from the people who buy cheap meals and spent on the people who make them. There are more direct ways to raise low earners' incomes. The Earned Income Tax Credit spreads the cost across all taxpayers, doesn't raise the cost of hiring, and doesn't depend on whether your employer has 59 locations or 61.

Third, the costs are the hard ones to see. As in Seattle, the gains are visible: a worker's paycheck goes up. The costs are counterfactual: the small chain that didn't open its next store, the teenager not hired, the shift that wasn't scheduled. Studies built on headcounts will always struggle to find them. That doesn't mean they are large. It means "no job loss found" is a weaker statement than it sounds.

The most defensible reading of California's experiment is that a $20 floor for big chains raised pay, raised prices, cost somewhere between nothing and a modest number of jobs, and shifted ground away from smaller competitors. Reasonable people can weigh that trade differently. What the evidence doesn't support is the claim on either side that the question has been settled.