In this April 1, 2015, photo, students and other supporters protest on the University of Washington campus in Seattle, in support of raising the minimum wage for campus workers to $15 an hour.
AP

When Seattle officials voted three years ago to incrementally boost the city’s minimum wage up to $15 an hour, they’d hoped to improve the lives of low-income workers. Yet according to a major new study that could force economists to reassess past research on the issue, the hike has had the opposite effect.

The city is gradually increasing the hourly minimum to $15 over several years. Already, though, some employers have not been able to afford the increased minimums. They’ve cut their payrolls, putting off new hiring, reducing hours or letting their workers go, the study found.

The costs to low-wage workers in Seattle outweighed the benefits by a ratio of three to one, according to the study, conducted by a group of economists at the University of Washington who were commissioned by the city. The study, published as a working paper Monday by the National Bureau of Economic Research, has not yet been peer reviewed.

On the whole, the study estimates, the average low-wage worker in the city lost $125 a month because of the hike in the minimum.

The paper’s conclusions contradict years of research on the minimum wage. Many past studies, by contrast, have found that the benefits of increases for low-wage workers exceed the costs in terms of reduced employment — often by a factor of four or five to one.

“This strikes me as a study that is likely to influence people,” said David Autor, an economist at the Massachusetts Institute of Technology who was not involved in the research. He called the work “very credible” and “sufficiently compelling in its design and statistical power that it can change minds.”

Yet the study will not put an end to the dispute. Experts cautioned that the effects of the minimum wage may vary according to the industries dominant in the cities where they are implemented along with overall economic conditions in the country as a whole.

And critics of the research pointed out what they saw as serious shortcomings. In particular, to avoid confusing establishments that were subject to the minimum with those that were not, the authors did not include large employers with locations both inside and outside of Seattle in their calculations. Skeptics argued that omission could explain the unusual results.

“Like, whoa, what? Where did you get this?” asked Ben Zipperer, an economist at the left-leaning Economic Policy Institute (EPI) in Washington.

“My view of the research is that it seems to work,” he said. “The minimum wage in general seems to do exactly what it’s intended to do, and that’s to raise wages for low-wage workers, with little negative consequence in terms of job loss.”

Economists might not readily dismiss the new study as an outlier, however. The paper published Monday makes use of more detailed data than have been available in past research, drawing on state records of wages and hours for individual employees.

As a result, the paper is likely to upend a debate that has continued among economists, politicians, businesses and labor organizers for decades. In particular, the results could exacerbate divisions among Democrats, who are seeking an economic agenda to counter President Trump’s pitches for protectionism, reduced taxes and restrictions on immigration.

Meanwhile, states and cities around the country are continuing to implement increases in the minimum wage. In November, voters in Washington approved an increase in the statewide minimum to $13.50 an hour by 2020. The idea is popular in conservative states as well. In Arizona, for instance, the minimum wage will be $12 an hour in 2020 after voters there cast ballots in favor of a hike.

“If I were a Seattle lawmaker, I would be thinking hard about the $15 an hour phase-in,” Autor said.

Economists have long argued that increasing the minimum wage will force some employers to let workers go. In 1994, however, economists David Card and Alan Krueger published research on minimum wages in Pennsylvania and New Jersey that contradicted this theory, motivating dozens of studies into the issue over the coming years.

Card and Krueger conducted a survey of fast-food restaurants in the two states while New Jersey was implementing an increase in the minimum wage. They found that restaurants in New Jersey had, in fact, added more workers to their payrolls more than restaurants in neighboring Pennsylvania, where the minimum wage remained constant.

Since then, economists have brought better data and more sophisticated statistical methods to bear on the question of the minimum wage, but without resolving the debate.

Their studies examined the overall numbers of workers or their annual incomes, but lacked precise information on how much workers were being paid by the hour. As a result, past research might be less reliable because the results might reflect many workers who are not paid low wages, said Jacob Vigdor, an economist at the University of Washington and one of the authors of the new study.

Their research, using detailed records from the state of Washington, addresses that problem.

“That’s really a step beyond what essentially any past studies of the minimum wage have been able to use,” said Jeffrey Clemens, an economist at the University of California, San Diego who was not involved in the research.

When the authors of the study took the same approach as Card and Krueger, measuring overall employment in the restaurant industry, they found similar results. The minimum wage did not substantially affect how many people were working in the industry or how many hours they were working.

The data, however, shows that about seven in 10 workers in Seattle restaurants make more than $13 an hour, suggesting that the overall level of employment in the industry might not be a reliable guide to how the minimum wage affects workers with low pay.

Indeed, while employment overall did not change, that was because employers replaced low-paying jobs with high-paying jobs. The number of workers making over $19 an hour increased abruptly, while the number making less than that amount declined, Vigdor and his colleagues found.

Vigdor said that restaurateurs in Seattle — along with other employers — responded to the minimum wage by hiring more skilled and experienced workers, who might be able to produce more revenue for their firms in the same amount of time.

That hypothesis has worrisome implications for less skilled workers. While there those with more ability might be paid more, junior workers might be losing an opportunity to work their way up. “Basically, what we’re doing is we’re removing the bottom rung of the ladder,” Vigdor said.

There could be another explanation for the results, however: the fact that large employers are not included. It could be that even if employers with only a single location cut payrolls, large firms expanded at the same time, giving low-wage workers other opportunities to earn money.

Other researchers have found that large employers are better able to raise wages in response to changes in the minimum. Liberal economists often argue workers have less bargaining power when negotiating their contracts at larger firms, and that as a result, employees at those companies are often underpaid in the absence of a wage floor.

“I think they underestimate hugely the wage gains, and they overestimate hugely the employment loss,” said Michael Reich, an economist at the University of California, Berkeley who was part of a group that published its own study of the minimum wage in Seattle last week.

Reich’s study uses more conventional methods in research on the minimum wage, relying on a publicly available federal survey. His group’s data did not allow the researchers to distinguish between high- and low-wage workers at a given firm, but they were able to separate large firms’ locations in Seattle from those outside the city.

Their results from the University of California accorded with past research. The minimum wage increased wages for workers in the restaurant industry, without reducing employment overall — in contrast to the findings from the University of Washington.

“Their results are so out of the range,” Reich said.

One way of explaining the disagreement could be that small businesses in Seattle have been forced to downsize in response to the increased minimum wage, while larger firms have expanded.

Yet when Vigdor and his colleagues examined the overall number of workers at small firms with a single location, they did not find that employment had decreased. That fact could could suggest that small businesses have responded to the increase not by downsizing but instead by hiring more experienced workers.

There’s another explanation for the growth in high-paid jobs and the decrease in lower-paid ones. The authors of the study argue that that’s occurring because employers are focusing on high-paid workers and leaving low-paid workers out, but it’s possible that something far more positive is happening.

Seattle’s economy is booming, and in a booming economy, more workers are likely to get raises or find jobs that pay better, and it may be that phenomenon — of workers getting raises, promotions or better paying jobs — that explains the shifts in the labor market the researchers see in Seattle.

Vigdor and his colleagues sought to address this problem, in essence, by constructing an index based on data from other parts of the state of Washington where local economies performed similarly to Seattle’s before the increases in the hourly minimum.

Low-wage employment declined in Seattle relative to this benchmark. Even compared to parts of the state with similar economies, there was less low-wage work in Seattle, suggesting that the minimum wage might have forced employers to cut some of those positions.

The method Vigdor’s group used to develop this index is on the cutting edge of economic research, but it is not perfect. It is possible that Seattle’s economy simply took a different direction at the same time as the minimum wage began to increase — even compared to economies in other places that seemed similar to Seattle’s before the vote.

EPI’s Zipperer argued that was the best explanation, given how pronounced the gains were for workers making more than $19 an hour.

“You’re just seeing an independent shift in the Seattle labor market toward higher wage employment,” he said, calling the figures for better-paid workers “a red flag.”

The broader national economy could have an effect on the results as well. In the past, noted San Diego’s Clemens, increases in the minimum wage have occurred when the economy was expanding rapidly and prices are going up. Employers could expect to ask consumers to pay more and to give their workers wages anyway. Increases in the minimum wage might just have been part of the cost of doing business.

Currently, though, inflation is at historically low levels, and the minimum wage in Seattle will be indexed to inflation after it reaches $15 an hour, forcing firms to plan for the long term.

Vigdor agreed that the effects of increasing the minimum wage could differ by time and place.

“The effect of the minimum wage depends on a lot of things. It depends on where you’re starting form. It depends on what kind of economy you’re raising it in,” Vigdor said. “There is no one ‘the effect of the minimum wage.’ “

That means that future research on the question could come to different conclusions. Vigdor said he looks forward to receiving criticisms of his group’s paper and suggestions for improving their approach.

“It’s really important to emphasize it’s a work in progress,” he said.