The proposed merger between wireless giants Sprint and T-Mobile will depress wages in the retail sector of the industry by as much as $3,200 per year for some workers, according to a new report released by the Economic Policy Institute and the Roosevelt Institute, two left-leaning think tanks.

The report deals with the question of monopsony power, which occurs when employers have sufficient leverage over workers to pay them less than they would get in a truly competitive labor market. By reducing the number of large firms active in the wireless market from four to three, the report argues, the merger would make it more difficult for retail employees in the remaining three firms to receive competing offers of employment that would boost their wages. A recent thread of economics research has identified monopsony power as a driver of the wage stagnation observed in many sectors of the economy.

“Concentration of employers confers monopsony power because workers lack the job opportunities that would ensure pay would track their productivity,” the report’s authors, Adil Abdela and Marshall Steinbaum, explained. They called for antitrust authorities currently reviewing the merger at the Federal Communications Commission and elsewhere to take these concerns into account. “Enforcers with a mandate to preserve competition must take labor markets as well as product markets into account when assessing competitive effects of any merger or conduct they might review,” they wrote.

To arrive at their estimates, Abdela and Steinbaum relied on previous research looking at the effects of employer concentration on earnings. They worked to translate those estimates to the retail telecommunications market at the level of commuting zones — small geographic areas meant to “closely reflect the local economy where people live and work,” according to the Agriculture Department.

The models used in the previous research yielded a range of possible estimates for the effects of a merger of Sprint and T-Mobile. In the 50 labor markets most likely to be affected, the estimates of wage stagnation ranged from an average decrease of about $10 a week, or $500 per year, at the low end to a decrease of $63 dollars a week, or $3,200 a year, at the high end.

Via email, Steinbaum explained that retail workers in the wireless industry wouldn’t necessarily see a pay cut if the merger goes through. Rather, their wages will simply fail to rise over time they way they would if the merger does not happen. In the merger scenario, “outside job offers come less frequently, in part thanks to concentration,” he said.

The Federal Communications Commission declined to comment on the study, citing the ongoing review of the merger. A representative from T-Mobile referred a request for comment to Jeffrey A. Eisenach, an economist with NERA Economic Consulting who has been advising T-Mobile on issues related to the deal. Eisenach sharply criticized the way the Roosevelt researchers defined labor markets in the paper. “How can you define a labor market which is only workers in wireless retail stores?” he asked in an interview. He said the paper assumes “that people who work in wireless retail stores are incapable of working anywhere else … it’s an obviously false assumption.”

Eisenach said he expects that in “real towns, where there are hundreds or thousands of employers,” the effects of the merger on wages “would be de minimis.” He added that an analysis of the merger he prepared at the request of T-Mobile showed it would create 24,000 jobs over three years.

In response, Steinbaum pointed out that the prior research the study drew on “var[ies] the market definition substantially, both more widely and more narrowly” than he and Abdela did in the Roosevelt paper. Those studies show the estimated wage effects “tend not to vary very much as a function of market definition.” He added that their paper did not explore the creation or destruction of jobs, only the impact of the merger on wages.

Jay Shambaugh, an economist and director of the Hamilton Project at the Brookings Institution, said the Roosevelt paper “is done in a clear and straightforward way, and the magnitudes [of the wage effects] seem plausible for the exercise being done.” Shambaugh, who is not involved in the merger, added that the questions about labor market definition were valid, but said the authors “have reasonable grounds that they have defined the market appropriately.”

He went on to echo one major contention of the paper: Antitrust regulators at the FCC and elsewhere typically don’t pay enough attention to how big mergers affect workers. He said the Roosevelt paper is “the type of analysis regulators would need to do if they were going to take the labor market implications seriously,” as a number of economists have urged.

“If mergers like this go through,” Shambaugh said, “concentration of employers will rise and that will raise the stakes to deal with other ways firms exercise (or expand) their power in labor markets (e.g. using franchise no poach agreements or non-compete contracts).”

Even with increased attention to labor markets, Abdela and Steinbaum wrote, antitrust enforcement alone “will never be a solution to the crisis of worker power in this country. It must be considered alongside such policies as increasing the minimum wage, ensuring macroeconomic full employment, increasing progressive taxation, improving labor standards and their enforcement, and mitigating shareholder power over companies that comes at the expense of other stakeholders.”