The Wage Went Up

On April 1, 2024, California raised the minimum wage for fast food workers to $20 an hour. The governor held a press conference. The headlines wrote themselves. Four dollars more per hour for some of the lowest-paid workers in the state. A 25% raise, overnight.¹

Within six months, the National Bureau of Economic Research estimated that 18,000 fast food jobs had disappeared relative to what employment would have been without the increase.² Two Pizza Hut franchisees had eliminated every delivery driver position across their locations before the law even took effect. More than 1,200 workers, gone by February.³ Rubio's Coastal Grill closed 48 California restaurants and filed for bankruptcy five days later.⁴ One McDonald's franchisee with two dozen locations cut 170,000 labor hours in the first year.⁵

The wage went up. The jobs did not follow.

This is a story about what happens after the announcement. After the press conference, after the headline, after the number on the sign changes. The belief is that raising the minimum wage helps workers. And it does help workers. The ones who keep their jobs, keep their hours, and keep their benefits. The rest disappear from the count.

The franchise

The economics of a fast food franchise are thin by design. A typical location does about $1.5 million in annual revenue. Food costs take 28 to 35 percent. Labor takes 25 to 35 percent. What remains after rent, franchise fees, insurance, and equipment is a net margin of 3 to 9 percent. On a good year, the owner takes home somewhere around $90,000.⁶

The increase from $16 to $20 is a 25 percent raise in the wage floor. For a McDonald's franchisee, that translated to roughly $250,000 in additional annual labor costs per location.⁵ Against a $90,000 margin, the arithmetic is self-explanatory.

Three responses were available. Raise prices. Cut hours. Cut heads. Most franchisees chose all three.

Chipotle raised its chicken burrito price 8.3 percent in California. McDonald's franchisees lifted prices 5 to 7 percent before the law took effect. Fast food prices across the state rose 56 percent faster than sit-down restaurant prices over the following year.⁷

The Pizza Hut franchisees did not cut hours. They eliminated an entire job category. PacPizza LLC and Southern California Pizza Co. outsourced all delivery to Uber Eats, GrubHub, and DoorDash. The 2,000 in-house delivery drivers were not reassigned. They were replaced by a platform.³

Rubio's Coastal Grill, in its bankruptcy filing, cited "rising food and utility costs that, combined with significant increases to the minimum wage in California, put pressure on a number of its locations." Thirteen closures in San Diego. Twenty-four in Los Angeles. Eleven in Northern California.⁴

But the layoffs and the closures were the visible cost. The invisible cost was quieter.

In 2021, economists Jeffrey Clemens, Jeremy Kahn, and Jonathan Meer published a study in the Journal of Labor Economics examining what happens to the composition of low-wage workers after minimum wage increases. They used American Community Survey data and job vacancy records. Their finding: after increases, workers in low-wage occupations are older and more likely to have a high school diploma. Job postings add education requirements that did not exist before. The workforce shifts upward. The least experienced, least credentialed workers are not fired. They are never hired.⁸

The mechanism is invisible because it does not produce a layoff notice. It produces an application that is never submitted, a job posting that now requires a diploma, a position that goes to someone with three years of experience instead of someone with none. The people priced out do not file a complaint. They simply do not appear in the data.

The research on California's increase is contested. UC Berkeley's Institute for Research on Labor and Employment published analyses finding no significant employment losses, with wages rising 18 percent for 90 percent of non-managerial workers and prices increasing roughly 3.7 percent.⁹ The NBER paper, using Bureau of Labor Statistics payroll data covering 95 percent of businesses, found 18,000 jobs lost. CalMatters, reviewing both camps in September 2025, wrote: "Two competing studies from less-than-objective sources leave us still wondering what the true impact might have been." Berkeley's researchers have been characterized as uniformly supporting labor positions. The Employment Policies Institute, which published the most aggressive job loss figures, was created by a restaurant industry lobbyist.¹⁰

The most telling data point may sit outside the research entirely. AB 1228 created the Fast Food Council with the authority to raise wages beyond $20 starting in 2025. The Council never did. It went dormant. No meeting was held for the rest of the year.¹¹

The paycheck

In 2017, a team of economists at the University of Washington published one of the most granular minimum wage studies ever conducted. They used administrative payroll records from Washington State's Employment Security Department to track what happened when Seattle raised its minimum from $11 to $13 per hour.¹²

Hourly wages rose 3 percent. Hours worked in low-wage jobs fell 6 to 7 percent. The net result: total payroll for low-wage workers decreased. The average low-wage worker lost $74 per month.¹² The wage went up. The paycheck went down.

The study was disputed. UC Berkeley's competing analysis found no significant negative effects in the restaurant sector specifically. The University of Washington team's data excluded multi-location chains. But the finding survived peer review and was published in the American Economic Journal: Economic Policy in 2022.¹²

Seven years later, Seattle extended the same logic to gig workers. The PayUp legislation, effective January 2024, guaranteed delivery drivers $17.27 per hour plus per-mile fees. The compensation included time spent in traffic, waiting for food, or idling on the app. The law also restricted the platforms' ability to deactivate drivers.¹³

The incentive structure inverted immediately. A driver who sat in traffic for an hour earned more than a driver who delivered quickly. One driver reported earning $58 for a single delivery that would have paid $17 before the law. He spent close to an hour in traffic. The longer the delay, the higher the pay.¹⁴

The owner of Spice Waala, a Seattle restaurant, told a local news outlet that delivery orders had dropped 30 to 40 percent. "A meal might be $12 or $15 in our restaurant," he said, "but by the time the customer gets it through the app, it becomes $35 or $40. I wouldn't buy our own food at that price."¹⁵

Within months, drivers who had been choosing between too many orders reported waiting hours without a single one. DoorDash pushed for full repeal of the ordinance. The city began revising the legislation less than three months after implementation.¹⁴

In 2021, the Congressional Budget Office projected what a $15 federal minimum wage would produce. The numbers were published in the same report, on the same page. 900,000 people lifted out of poverty. 1.4 million jobs eliminated.¹⁶

The 900,000 became the headline. The 1.4 million became page two.

The receipt

Before New York City set a minimum pay rate for app-based food delivery workers, the Department of Consumer and Worker Protection studied what they earned. The figure was $5.39 per hour before tips.¹⁷

The city established a minimum of $17.96 per hour for trip time, phasing up to $19.96 by April 2025, adjusted for inflation to $21.44. DoorDash, Uber, and Grubhub sued to block the rule. They lost in State Supreme Court. They lost on appeal. Enforcement began December 4, 2023.¹⁸

The apps responded in three moves.

First, they added fees. DoorDash listed a $1.99 "NYC Regulatory Response Fee" on every checkout screen. Uber Eats added a $2 "City of New York Courier Fee." Consumer delivery fees across all platforms rose from $12.7 million per week to $20.1 million per week. A 58 percent increase. Restaurant fees to the apps, over the same period, rose 13 percent. The cost went overwhelmingly in one direction.¹⁹

Second, they moved the tip. DoorDash and Uber Eats relocated the tip prompt from checkout to after delivery. The change happened the same week enforcement began. Average tips dropped from $3.66 per delivery to $0.93 within seven days. By early 2026, tips on those platforms had settled at $0.76 per delivery. On apps that kept the tip prompt at checkout, tips held at $2.17. The city estimated the total loss at $550 million in tip income over eighteen months. Roughly $5,800 per worker per year, clawed back through a design change that took one week to implement.²⁰

Third, and this is where the story turns, they named the cost.

By labeling the surcharge a "Regulatory Response Fee," the apps made the cost transfer visible. Not hidden in a general price increase. Not folded into a higher delivery charge. Itemized. Named. Attributed. The consumer could see, line by line, who had caused the increase.

When Instacart later added its own $5.99 version of the fee for grocery delivery, Councilwoman Sandy Nurse, who had championed the worker protection laws, wrote an op-ed calling it a "junk fee." She accused Instacart of "reaching into consumers' wallets to pad its own profits." The city passed a tipping transparency law requiring apps to show the tip prompt before checkout. DoorDash and Uber sued. A federal judge denied the injunction two days before the law took effect.²¹

The anger was not about the cost. The anger was about the label.

The city's own data told a more complicated story than either side wanted. Total deliveries across all platforms rose 8 percent year over year. Average hourly earnings for delivery workers jumped from $11.72 to $19.26, a 64 percent increase. By early 2025, the city reported $1.2 billion in cumulative additional wages since enforcement began.²²

But $550 million in tips had been engineered away. The base pay went up. The tips were clawed back. The consumer paid more. And when the receipt showed exactly why, the political response was not to revisit the policy. It was to fight the receipt.

The count

The economist Thomas Sowell spent a decade as a Marxist. At thirty, he took a position at the U.S. Department of Labor. His assignment was to study the effects of minimum wage laws. The data contradicted everything he believed. Increases in the minimum wage correlated with increased unemployment among the youngest and least experienced workers. The people the policy was designed to protect were the most likely to lose their jobs.²³

Sowell revised his worldview. The Department of Labor, he later wrote, showed no interest in requesting data that might prove its own program was destroying jobs. The man changed. The institution did not.²³

The observation that followed was simpler than the ideology it replaced: the first visible result of any decision is almost never the final result. The more emotionally satisfying the immediate outcome, the more likely the full consequence chain is being ignored. Costs do not disappear. They transfer. To someone with less visibility, less power, less voice.²⁴

The minimum wage has a version of this that is structural, not political.

When a fast food worker in California loses their job, they are no longer a minimum wage worker. They are no longer part of the dataset that measures whether the minimum wage increase worked. If they find another job at the new wage, they reappear. If they stop looking within four weeks, the Bureau of Labor Statistics no longer counts them as unemployed. They have left the labor force. They are not in the numerator. They are not in the denominator. They are nowhere in the data.²⁵

The statistic "average hourly wage of minimum wage workers" measures only the survivors.

When someone is priced out of the labor market by a wage floor that exceeds what an employer will pay for their labor, the reported average wage goes up. The number of workers goes down. The headline reports the average. Not the headcount. The policy is evaluated by the experience of the people it kept, not the people it lost.

In 2019, the Institute of Labor Economics found that for workers aged 20 to 24, a 10 percent increase in the minimum wage was associated with a measurable decrease in labor force participation. Not unemployment. Participation. These workers did not show up as jobless. They stopped being counted entirely.²⁶

The substitution effect compounds this. After minimum wage increases, the workers who fill low-wage jobs skew older, more educated, more credentialed.⁸ The positions still exist. The people who hold them changed. The eighteen-year-old without a diploma who would have worked the register last year is not in the unemployment line. She is not anywhere in the system. She did not lose a job. She never got one. And because she never got one, she is not a data point. She is not a story. She is not in the count.

This is not a critique of the minimum wage. In concentrated labor markets with few employers, moderate increases have produced minimal displacement. The UK's National Living Wage showed an employment elasticity of negative 0.17, barely distinguishable from zero.²⁷ Denmark has no statutory minimum wage. Its McDonald's workers earn roughly $20 per hour at base rate through collective bargaining, with evening and weekend premiums pushing effective pay higher, six weeks of paid vacation, a 10 percent pension contribution, and fully paid sick leave.²⁸

The question is not whether wages should be higher. Wages should reflect what a society decides labor is worth, and the mechanisms for arriving at that number range from legislation to collective bargaining to market dynamics. Reasonable people disagree on which mechanism works best.

The question is what gets counted afterward.

In California, 18,000 jobs disappeared from the data. In Seattle, workers earned higher wages and took home less money. In New York, the apps made the cost visible and the politicians fought the label. In each case, the wage went up. In each case, the cost went somewhere else. And in each case, the people who bore the cost were removed, by the structure of the measurement itself, from the statistic that evaluated whether the policy worked.

Sowell asked the question at the Department of Labor sixty years ago. The department did not want the answer then. The measurement system does not produce the answer now. Not because the data is hidden. Because the people are.

The wage went up. It always does. The question was never the wage. It was who was left to collect it.