The Economic Policy Institute (EPI) released a report last week that purports to show that “H‑1B employers undercut local wages.” Employers use the H‑1B program to hire temporary foreign workers in specialty occupations. EPI writes, “By setting two of the four wage levels below the median—and thereby not requiring that firms pay market wages to H‑1B workers—the DOL [Department of Labor] has in effect made wage arbitrage a feature of the H‑1B program.” This post explains why this is mistaken.

The median wage for an entire occupation is not the “market” wage for a specific worker. EPI confuses the median wage—the statistical midpoint in the entire range of market wages—for the singular “market wage.” A market wage is just whatever an employer would pay a worker in an open market—which depends entirely on the characteristics of the job and the productivity of the worker. Some employees receive offers above the median because they have more skills, more experience, or more significant job duties and so are more productive and valuable to the company than most other workers, while others receive wage offers below the median for the opposite reason.

As the Bureau of Labor Statistics (BLS) has detailed, wages in skilled occupations can vary dramatically within an occupational category—a fact that EPI acknowledges but fails to explain. BLS explains that these pay differences are a result of differing credentials, experiences and skills, industries or employers, job tasks, and performance.

By EPI’s definition of the market wage, half of American workers also receive “below market” wages since half are by definition paid below the median wage, yet it states—correctly—that “conceptually, the market wage is the wage a U.S. worker would command for a position in a specific occupation and region.” Since U.S. workers command wages both above and below the median, that means all those wages are market wages, not just the median. Yet EPI wants to stop H‑1B employers from paying foreign workers in accordance with their skills (strangely, it doesn’t object to skill-based pay for U.S. workers).

Nearly all H‑1B employers offer above the median wage for H‑1B workers of a specific skill level. The DOL estimates the market wage for H‑1B jobs by taking the average wage for workers in the same geographic area and same occupation who have similar levels of experience and skills, using wage surveys from the Bureau of Labor Statistics. DOL calls this average market rate the “prevailing wage” and provides for four levels based on skills, experience, and responsibilities (1. Entry; 2. Qualified; 3. Experienced; 4. Fully competent or supervisory).

Because the prevailing wage is the average wage at a given skill level, and averages were above the medians for 97 percent of occupations in 2019, nearly all H‑1B employers were already offering for a specific skill level wages above the median, which EPI insists is the best statistical proxy for the market wage. EPI wants to mandate the median wage, just not the median most relevant to the worker being hired. If DOL did use the median wage rather than the average at a given skill level, it would reduce the required wage for nearly all occupations. For computer and math occupations—the largest H‑1B occupation—for example, the national average was $5,190 more than the national median.

100 percent of H‑1B employers offer H‑1Bs at least the average prevailing market wage for similar U.S. workers. By law, H‑1B employers must offer their foreign workers the average prevailing wage in the occupation for the relevant skill level. DOL provides the prevailing wage rate through BLS prior to approving the hire. This means that no H‑1B employer can offer below the market wage for the occupation, and the DOL data bear this out (see Table 1 below).

Some H‑1B employers might sometimes fail to pay out what they offer, which is why DOL audits employers to verify that they meet their obligations. Some H‑1B workers would also make more money if they had a status that allows them to easily change occupations or positions to ones where they would be most productive and so earn higher wages. Allowing H‑1B workers to change jobs as easily as legal permanent residents would fix this problem. But these issues are very different from claiming, as EPI does, that H‑1B employers are offering below-market wages for the actual job being performed. They are not.

78 percent of H‑1B employers offer wages, on average, above average market wages—20 percent above. Each year, DOL publishes the prevailing wage determination for each H‑1B job offer and the actual wage offer for each employee in those jobs. The average offered wage for all 61,420 H‑1B requesting employers in FY 2019 was $100,461, while the average prevailing wage determination was $83,619, meaning H‑1B employers were offering an average of $16,842 more than the average market wage that the law requires—20 percent above. In fact, 78 percent of H‑1B employers had average wage offers above their average prevailing wage determination. The vast majority of H‑1B employers pay at least some employees more than they are required to pay.

Table 1 shows the average offered wages and average prevailing wage determinations for the top 500 H‑1B requesting employers. The takeaway from the data is that H‑1B employers are, on average, offering a premium for many of their foreign workers. The average wage was higher than the average prevailing wage determination for every company in the top 45 H‑1B requesting companies. For major companies like those EPI calls out in its report—such as Microsoft, Amazon, and Google—the average increases over the prevailing wage were substantial: 17 percent for Microsoft, 19 percent for Ama​zon​.Com Services, and 64 percent for Google.

71 percent of H‑1B employers have average wage offers above average market wages at every skill level. EPI repeatedly emphasizes that employers are underpaying H‑1B workers who receive prevailing wage determinations at Level 1 entry or Level 2 qualified wage levels. Yet DOL data show that in fact, not only were 100 percent of Level 1 and Level 2 H‑1B employers offering the prevailing market wage for U.S. workers at the workers’ skill levels in 2019, 71 percent of Level 1 and 76 percent of Level 2 H‑1B employers were offering wages that were an average of 19 percent and 18 percent higher, respectively, than the prevailing wage for U.S. workers at their skill level.

Level 1 or Level 2 wage offers may be below the median for their entire occupation, but they are higher than those for similarly skilled Americans. At the major companies that EPI calls out specifically, this is also true. The average H‑1B offers at Microsoft, Amazon, and Google for Level 1 or Level 2 jobs were as much as 96 percent higher than the average prevailing wage determinations for those companies. Overall, 71 percent of H‑1B employers had average wage offers above their average prevailing wage determination at every skill level at which they request H‑1B workers.

The lowest skilled H‑1Bs were the most likely to receive above market wage offers. Looking at the share of jobs rather than the share of employers shows that 100 percent of all certified job offers were at least at the prevailing wage in FY 2019 and about half (47 percent) were above it. In other words, the actual market wage was actually above the estimated average market wage for nearly half of H‑1B hires. This includes 52 percent of Level I jobs, 46 percent of Level II, 50 percent of Level III, and 44 percent of Level IV. Just because a worker had a lower wage offer did not mean that they were more likely to be underpaid relative to U.S. workers. In fact, the opposite was true (Table 3). Microsoft, Ama​zon​.Com Services, and Google exceeded the prevailing wage for U.S. workers on 57 percent, 67 percent, and 98 percent of their job offers, respectively. Yet EPI states that these companies are “are exploiting a flawed H‑1B prevailing wage rule to underpay their H‑1B workers relative to market wage standards.” This is incorrect.

H‑1B employer requests at higher wage levels have doubled since 2010. EPI might assert that H‑1B employers can write whatever they want on the H‑1B application, so these numbers are meaningless. Of course, they are the same numbers on which EPI bases its report that concludes H‑1Bs are underpaid, but more importantly, this theory would fail to explain why employers are increasingly requesting employees at higher wage levels. From 2010 to the second quarter of 2020, the share of H‑1B jobs requested at Level I entry level wages fell from 54 to 13 percent, while wages for Level 2 workers almost doubled 29 to 48 percent. Level 3 grew from 11 to 25 percent, and Level 4 from 6 to 15 percent (Figure 1). EPI reports similar numbers, but fails to explain why employers would choose to do this, if it were not reflective of the market.

H‑1B workers’ median wage was double the U.S. median wage and growing twice as fast as all U.S. wages. While the data unequivocally show that H‑1B workers are not being underpaid relative to similar U.S. workers, it also shows that H‑1B workers do not receive “low wages”—as EPI alleges—in an absolute sense either. The median H‑1B worker received a salary of about $98,000 in FY 2019. The median for all workers was just short of $40,000. It is absurd to describe workers in the top 10 percent of wage earners as “low” paid. Figure 2 is also inconsistent with the idea that H‑1B employers can pay whatever they want, regardless of the market. H‑1B wages grew twice as fast as wages in the labor market overall from 2004 to 2019 (88 percent v. 38 percent).

H‑1B law requires the government to allow wages based on skills and job duties. EPI’s entire report is based on the idea that employers shouldn’t offer wages based on a workers’ actual job responsibilities and skills and that the government should stop allowing this practice. Instead, they state that the Trump administration should bar wages below the 75th percentile for the entire occupation. That would effectively bar all current job offers at Level I‑III wages and even most Level IV offers as well. It would shrink the program by about 95 percent. The law mandates that the government provide “at least 4 levels of wages commensurate with experience, education, and the level of supervision.” EPI asserts—without actually quoting the law except in a footnote—that the government can interpret this requirement as requiring all wage levels be far above the median wage for the entire occupation.

EPI claims that “DOL has yet to explain its reasoning and justification for setting the two lowest levels below the local median wage.” Yet this is false. DOL has explained how it arrives at the four levels in its Prevailing Wage Determination Policy Guidance. Level 1 “entry” wages are for those who have a basic understanding of the job and who perform only routine duties. Level 2 “qualified” wages are for those who have a good understanding of the job and perform moderately complex tasks. Level 3 “experienced” wages are for experienced workers who have a sound understanding of the job and have special skills or knowledge. Level 4 “fully competent,” supervisory wages are for those workers who solve “complex problems” that require “judgment and independent evaluation” with minimal supervision.

These wage classifications are entirely legitimate categories for evaluating pay and correspond to real observed differences in the wages of U.S. workers, as U.S. Citizenship and Immigration Services’ Administrative Appeals Office has recognized. Private businesses use similar classifications for determining what wage to offer when seeking new hires anyway. EPI fails to explain how DOL could base wages on “experience, education, and level of supervision” as the law requires without permitting wages across the entire wage distribution. Instead, it simply asserts that the median wage for the entire occupation is the “market” wage and that H‑1B workers should always receive wages far higher than the median occupational wage, regardless of their experience, responsibilities, or skills.

If employers carried out EPI’s recommendation, it would result in H‑1Bs being paid far more than comparably skilled U.S. workers in the same occupation. Obviously, this is an untenable result, and instead, the policy would effectively ban hiring nearly all H‑1B workers. The H‑1B program is the main on-ramp that recent foreign college graduates have to the U.S. labor market. Excluding these new skilled foreign workers, the entire immigration system for skilled foreign labor would shrink by more than 50 percent, and America would lose talented workers that are essential to the post-COVID-19 recovery.