Introduction and summary

H-1B is a temporary nonimmigrant work visa that allows U.S. employers to hire college-educated migrant workers as well as fashion models from abroad; nearly 500,000 migrant workers are employed in the United States in H-1B status.1 The H-1B is an important—but deeply flawed—vehicle for attracting skilled workers to the United States. The H-1B visa is in desperate need of reform for a number of reasons that we have explained in other writings,2 but the fundamental flaw of the H-1B program is that it permits U.S. employers to legally underpay H-1B workers relative to U.S. workers in similar occupations in the same region. This report explains how this occurs by describing the H-1B prevailing wage rule and analyzing the available data on the wage levels that employers promise to pay their H-1B employees.

Background

The U.S. Department of Labor (DOL) has broad discretion to set H-1B wage levels, that is, the minimum wage employers must pay their H-1B workers, which corresponds to the H-1B workers’ occupation and the region where they will be employed. By law, DOL must set four H-1B wage levels—which it does according to wage survey data from the Bureau of Labor Statistics’ Occupational Employment Statistics survey. DOL has set the two lowest levels (of the four) well below the local median wage.

We believe that the median wage for an occupation in a local area is reflective of the minimum market rate that should be paid to an H-1B worker in order to safeguard U.S. wage standards and ensure that migrant workers in H-1B status are compensated fairly. By setting two of the four wage levels below the median—and thereby not requiring that firms pay market wages to H-1B workers—DOL has in effect made wage arbitrage a feature of the H-1B program. Changing program rules to require and enforce above-median wages for H-1B workers would disincentivize the hiring of H-1B workers as a money-saving exercise, ensuring that companies will use the program as intended—to bring in workers who have special skills—instead of using the H-1B as a way to fill entry-level positions at a discount.

Wage-level data make clear that most H-1B employers—but especially the biggest users, by nature of the sheer volume of workers they employ—are taking advantage of a flawed H-1B prevailing wage rule to underpay their workers relative to market wage standards, resulting in major savings in labor costs for companies that use the H-1B. Further, our analysis of H-1B prevailing wage levels raises serious doubts about whether H-1B employers, including the top 30 H-1B employers and major U.S. technology firms, use the program solely, or even mostly, to hire workers with truly specialized skills.

Key findings

The two lowest permissible H-1B prevailing wage levels are significantly lower than the local median salaries surveyed for occupations. The two lowest H-1B wage levels set by DOL correspond to the 17th and 34th wage percentiles locally for an occupation. This translates into salaries that are significantly lower than local median salaries—17% to 34% lower on average for computer occupations (which are among the most common H-1B occupations). H-1B employers can reap significant savings by selecting one of the two lowest wage levels instead of the Level 3 wage (the median, or 50th-percentile, wage) or the Level 4 wage (above the median, at the 67th percentile).

Not surprisingly, three-fifths of all H-1B jobs were certified at the two lowest prevailing wage levels in 2019. In fiscal 2019, a total of 60% of H-1B positions certified by DOL had been assigned wage levels well below the local median wage for the occupation: 14% were at H-1B Level 1 (the 17th percentile) and 46% were at H-1B Level 2 (34th percentile).

Likewise, three-fifths of H-1B jobs certified for the top 30 H-1B employers were at the two lowest prevailing wage levels. Twelve percent of all certified positions for the top 30 H-1B employers were set at the Level 1 wage, and nearly half (48%) were certified at Level 2, meaning that 60% (three in five) of all H-1B jobs for the top 30 employers were certified at wages lower than the local median wages for the occupations.

The top 30 H-1B employers play an outsized role in the program. In 2019, 53,377 employers had at least one petition approved for an H-1B worker. However, the top 30 H-1B employers accounted for more than a quarter, or one in four, of all H-1B petitions approved by U.S. Citizenship and Immigration Services for initial and continuing H-1B employment (105,660 of the 389,323 total). Looking at the DOL data on Labor Condition Applications, the top 30 H-1B employers received approval for 371,461 H-1B positions on LCAs, accounting for 38% of the 968,538 H-1B positions certified by DOL in fiscal 2019.

The vast majority of employers that use the program employ very few H-1B workers. In fiscal 2019, 86% of the 53,377 H-1B employers were approved by USCIS for five or fewer H-1B workers, including both new and continuing H-1B workers, while the top 30 H-1B employers were approved for an average of 3,522 H-1B workers each.

Half of the top 30 H-1B employers use an outsourcing business model. Fifteen of the companies listed in the top 30 H-1B employers have a business model based on outsourcing jobs; these companies place their H-1B hires at third-party client sites. These companies rely on the H-1B program to build and expand their business, which sometimes includes sending U.S. jobs overseas.

Major U.S. firms—not just outsourcing companies—pay low wages to their H-1B employees. Major U.S.-based technology firms that hire H-1B workers directly, rather than contract them out to third-party employers, had significant shares of their certified H-1B positions assigned as Level 1 or Level 2, the two lowest wage levels in fiscal 2019, both of which are below the local median wage:

• Amazon and Microsoft each had three-fourths or more of their H-1B positions assigned as Level 1 or Level 2.
• Walmart and Uber had roughly half of their H-1B positions assigned as Level 1 or Level 2.
• IBM had three-fifths of its H-1B positions assigned as Level 1 or Level 2.
• Qualcomm and Salesforce had two-fifths of their H-1B positions assigned as Level 1 or Level 2.
• Google had over one-half assigned as Level 2.
• Apple had one-third of its H-1B positions assigned as Level 2.

Recommendations

DOL should act. The H-1B prevailing wage should reflect realistic market wage levels and help prevent downward pressure on U.S. wage rates in H-1B occupations. To accomplish this, we recommend that DOL use its existing authority to set the lowest (Level 1) wage to the 75th percentile for the occupation and local area and also require that wage offers to H-1B workers never be lower than the national median wage for the occupation.

Congress should act. Further, to ensure that future administrations do not reduce wage levels, Congress should enact a statute setting reasonable minimums for H-1B wage levels and providing DOL with new legal authority and funding to conduct random audits of H-1B employers to verify that they are not manipulating job titles and wage levels in order to underpay H-1B workers. The H-1B and L-1 Visa Reform Act of 2017, introduced by Senators Chuck Grassley (R-Iowa) and Richard Durbin (D-Ill.), would vastly improve the H-1B program along these lines.3 The Act would eliminate the two lowest wage levels, so that H-1B workers could not be paid at a wage that is lower than the local median (50th-percentile) wage, and would grant DOL new authority to increase audits and hire additional staff.

H-1B visas, Labor Condition Applications, and prevailing wage levels

The statute creating the H-1B visa—which allows U.S. employers to hire college-educated workers as well as fashion models from abroad—contains language establishing a “prevailing wage.”4 This prevailing wage requirement is intended to protect the wages of U.S. workers in occupations requiring a college degree from adverse impacts and to prevent college-educated migrant workers from being underpaid and exploited. Corporate lobbyists and other H-1B proponents often cite this prevailing wage requirement in the H-1B law as evidence that H-1B workers cannot be paid less than U.S. workers. However, the reality is that the H-1B statute, regulations, and administrative guidance allow employers wide latitude in setting wage levels.

Hiring an H-1B worker is an action that occurs outside of the normal operation of the labor market, with the government setting key hiring and employment rules. As such, setting an appropriate wage level is critical to ensure the program operates in a way that is fair to both U.S. workers in major H-1B occupations and the migrant workers who are hired through the H-1B program. The migrant workers hired through the H-1B program should possess specialized skills and fill genuine shortages in the U.S. labor pool. The shortages should be significant enough that they cannot easily be filled by standard market mechanisms such as: increasing offered wages to the existing U.S. labor pool, training and developing the skills of U.S. workers, or expanding recruitment to find new employees from the U.S. labor pool. To ensure that migrant workers possess specialized skills and are filling genuine shortages, H-1B policy should set the prevailing wage for an H-1B worker at a wage that is higher than the market wage.

Conceptually, the market wage is the wage a U.S. worker would command for a position in a specific occupation and region. We believe that the most reasonable and closest proxy for a market wage is the median wage for an occupation in a local area. However, employers seeking to hire workers through the H-1B program may select from among four permissible “prevailing” wage levels—the two lowest of which the U.S. Department of Labor (DOL) sets significantly below the local median wage. The 2005 statutory language from Congress requires there be four H-1B prevailing wage levels,5 but does not prescribe what these wage levels should be relative to the local wage distribution.6 DOL has yet to explain its reasoning and justification for setting the two lowest levels below the local median wage.

The process of assigning prevailing wage levels to H-1B positions is done through what is known as a Labor Condition Application (LCA)—the first stage of the H-1B process. Employers must submit LCAs to DOL, and those LCAs must be certified by DOL before employers can submit petitions to United States Citizenship and Immigration Services (USCIS) to hire H-1B workers. In an LCA, an employer specifies one or more positions it wants to hire an H-1B worker for, including the occupations being hired for, the geographic locations where the workers will be placed, and, for each position, the wage level chosen, the prevailing wage (the government-required minimum wage at that wage level), and the salary the employer intends to pay for that position (which must be at least as high as the specified prevailing wage).

The LCA is the H-1B program’s primary mechanism to ensure employer accountability, by requiring employers to promise they will comply with H-1B visa rules and pay at least the prevailing wage that corresponds to a specific occupation in a geographic area. The LCA is intended to preserve the integrity of the labor market by safeguarding the wages and working conditions of U.S. workers and of migrant workers employed with H-1B visas.

DOL reviews individual LCAs and then either certifies or denies the requested positions. The review process is perfunctory, with “minimal human intervention,” according to a report from DOL’s own inspector general.7 Electronically submitted LCAs may be approved in a matter of minutes. To be approved, applications must only be “complete and free of obvious errors.” The inspector general’s report labeled the DOL’s review of LCAs as simply a “rubber stamp” that “adds nothing substantial to the process.”

In fiscal 2019, 968,538 H-1B positions were certified through LCAs. However, it is important to note that not all certified positions in LCAs go on to become approved USCIS petitions that ultimately allow employers to hire H-1B workers; every year there are many more positions certified by DOL than the ultimate number of H-1B petitions and work visas that are approved, because employers may decide not to use the certified positions from an LCA in a petition (application) for a worker, or the subsequent petition might be denied by USCIS, or the employer may not be allowed to petition because of the H-1B visa’s annual numerical limit.8 The total number of H-1B petitions approved for H-1B workers by USCIS in fiscal 2019 was 389,323, which includes approved petitions for both initial (new) and continuing employment.

When reporting wage levels for H-1B positions on an LCA, the employer follows DOL guidelines for determining the appropriate prevailing wage that corresponds to each H-1B position. Since wages for workers in an occupation can vary widely, DOL relies on data from one of the U.S. Bureau of Labor Statistics’ major surveys—the Occupational Employment Statistics (OES) survey—to construct a distribution of wages for each occupation in a specific geographic location. DOL then sets four prevailing wage levels, with each level set at a specific percentile in the distribution. Employers must use either the OES survey or a private wage survey (more on this later) to determine the wage levels that correspond to the occupation and geographic location for each position, so they do have some constraints in identifying the prevailing wages they are asking DOL to certify. However, employers have significant latitude to decide which of the four wage levels get assigned to particular jobs.

As noted above, employers have four wage levels to choose from: They may pay the Level 1 “entry-level” prevailing wage, which DOL sets at the 17th percentile of wages surveyed for the occupation in the local area. This is clearly the bottom of the distribution, with 83% of workers in that occupation being paid more than the Level 1 H-1B worker. Employers may also opt to pay the Level 2 wage, which is at the 34th percentile. The Level 3 wage is at the 50th percentile—the median wage—and Level 4 is at the 67th percentile, the only wage level that is higher than the median.9 While the wage level is intended to correspond to the H-1B worker’s education and experience, in practice the employer gets to choose the wage level and the government doesn’t verify that a prevailing wage is appropriate unless a lawsuit or a complaint is filed by a worker. Such complaints are unlikely since it would require an H-1B worker to blow the whistle on their own employer, the same employer that controls the H-1B worker’s immigration status and ability to remain in the United States. We know of no cases in which DOL has investigated an LCA-stage misclassification of an H-1B wage level.

Three-fifths of all H-1B jobs were certified at the two lowest prevailing wage levels in 2019

Although salary information that corresponds to requested positions on LCAs has been made available by DOL for a number of years through the Office of Foreign Labor Certification’s LCA disclosure data, until recently the prevailing wage levels selected by employers were not readily available. In 2011, the Government Accountability Office (GAO) for the first time reported what some had suspected and speculated about but to that point were not able to officially confirm: The vast majority of H-1B jobs were being certified by DOL at the two lowest wage levels.10

GAO reported that between June 1, 2009, and July 30, 2010, 83% of H-1B jobs were certified at Level 1 or Level 2. Only 11% were certified at the median wage and a mere 6% (one in 17 workers) at a wage above the median. DOL has since released wage-level data for fiscal years 2015, 2017, 2018, and 2019.11 Our analysis of the data, shown in Table 1, reveals that in every year for which the raw data on prevailing wage levels has been made available, at least three-fifths of all H-1B positions certified were assigned by employers as Level 1 or 2: 80% in fiscal 2015, 62% in fiscal 2017, 63% in fiscal 2018, and 60% in fiscal 2019. The fiscal 2019 wage-level data from DOL show that Level 1 accounts for 14% of all certified positions, Level 2 accounts for 46%, Level 3 accounts for 19%, and Level 4 accounts for 12%; alternative wage surveys were used for 9% of certified positions.

The two lowest permissible H-1B wage levels are significantly lower than the median of salaries surveyed for the occupation in the local area

In order to understand the differences among salary amounts that correspond to prevailing wage levels, we provide an example in Table 2 that comes from the Foreign Labor Certification (FLC) Data Center’s Online Wage Library. The Online Wage Library lists prevailing wage levels for every available occupation and geographic area, based on DOL’s OES survey data, and employers visit this site to find the appropriate wage levels for the vast majority of H-1B positions they list on LCAs. (Employers used alternative wage surveys to set the prevailing wage for 9% of positions certified in 2019; see Table 1.)

For our example, we selected the occupation of Software Developers, Applications—nationally the most common certified H-1B occupation in 2019—and selected the Washington, D.C., metropolitan area as the region. Table 2 shows the DOL minimum annual salary that employers must pay H-1B workers for this occupation and region at each of the four corresponding prevailing wage levels. Employers hiring at Level 1 receive a discount of 36%, or $41,746, versus paying the median wage for the job in the region—represented by Level 3—and those hiring at Level 2 receive a discount of 18%, or $20,863.

This example is typical of the wage differentials between the levels and shows how, because of where DOL has set the percentiles for the wage levels, employers can reap significant savings by selecting the two lowest wage levels instead of the Level 3 wage (the median) or Level 4 wage (above the median, at the 67th percentile). The Level 1 wage for computer occupations nationwide in the FLC’s data set is, on average, 34% lower than the median and Level 2 is 17% lower than the median for the occupation in the local area.12

The top 30 H-1B employers play an outsized role in the program and half use an outsourcing business model

We now take a closer look at the 30 H-1B employers with the largest number of approved petitions at USCIS—which we refer to as the “top 30” H-1B employers. Approved USCIS petitions for H-1B workers are the best way to identify the number of actual H-1B workers employed in a given fiscal year, as opposed to LCA positions certified by DOL.13 We identify these top 30 employers based on the number of H-1B petition approvals (including petitions for both initial and continuing H-1B workers) reported for fiscal 2019 in the USCIS H-1B Employer Data Hub.14 The top 30 are listed in Table 3 along with the total number of H-1B petitions that were approved for each company in fiscal 2019.

It is notable that one-half of all companies listed in the top 30 H-1B employers are companies that have an outsourcing business model. Companies with an outsourcing business model rely on the H-1B program to build and expand a business model based on outsourcing jobs.15 In this arrangement, rather than being employed directly by the company that hired them, the H-1B workers ultimately work for third-party clients, either on- or off-site. In some cases the work is later moved abroad to the H-1B worker’s country of origin once the worker has become proficient enough in the job to perform it remotely from abroad.16 The last column in Table 3 indicates which of the top 30 H-1B employers have an outsourcing business model. The implications of the outsourcing business model for the H-1B program are discussed in a later section.

The USCIS H-1B Employer Data Hub data show that in 2019, 53,377 employers participated in the H-1B program, meaning they had at least one approved petition for an H-1B worker.17 Of those 53,377 employers, the top 30—which represent 0.06% of all employers participating in the H-1B program—were issued more than one in four of all approved H-1B petitions (105,660 of the 389,323 total) (see Table 3). That amounts to an average of 3,522 H-1B workers per company in the top 30 in 2019. In contrast, the vast majority of H-1B employers have very few approved petitions. Petition data in the USCIS H-1B Employer Data Hub show that for fiscal 2019, 45,651 employers—86% of the 53,377 total employers participating in the program—had five or fewer H-1B petitions approved (see Table 4).

The top 30 employers accounted for an even larger share of the H-1B positions certified by the U.S. Department of Labor (through LCA approvals) in fiscal 2019 than they did for approved petitions to USCIS. (As described above, getting an H-1B position certified by DOL via an LCA is a required initial step for an employer applying to hire, renew, or move an H-1B worker.) DOL disclosure data show that in fiscal 2019, employers submitted LCA requests for certification totaling 1,051,707 H-1B positions to DOL.18 Of those submitted, nearly all (92%, or 968,538) were certified by DOL.19 The top 30 employers received approval for 371,461 H-1B positions—accounting for 38% of all H-1B positions certified in fiscal 2019.

Three-fifths of all H-1B jobs certified for the top 30 were at the two lowest prevailing wage levels

The data in Table 5 show the total number of certified H-1B positions for each of the top 30 employers, as well as how those certified positions break down by prevailing wage level. Table 6 is calculated from the data in Table 5; for each employer, it shows the shares of H-1B positions by wage level. The wage-level data from DOL are useful measures for understanding the wages that employers pay their H-1B workers, the sophistication of the positions an employer is seeking to fill, and the possible impact on U.S. wage standards. (Because H-1B prevailing wage levels are specific to an occupation and local area, they allow comparisons to be made with the salaries that are being paid to workers who are currently employed in an occupation and local area.)

Table 6 shows that 12% of all certified positions for the top 30 H-1B employers were at Level 1, and nearly half (48%) were certified at Level 2. A total of 223,509 H-1B positions certified for top 30 employers were at either Level 1 or 2, meaning that 60% (three in five) of all H-1B positions for the top 30 employers were certified at a wage lower than the local median wage for the occupation.

Just over one-fifth of all H-1B positions (21%) were certified at the Level 3 wage, which is the local median wage, and only 11% (one in nine) were certified at the highest prevailing wage level—Level 4—the 67th percentile wage.

A total of 26,877 positions, accounting for 7% of all H-1B positions certified for the top 30 companies, had prevailing wages established by an “independent authoritative source” or “another legitimate source” that was not DOL, which means a non-DOL wage survey was used to determine the H-1B worker’s salary. Because of data limitations, we cannot make any definitive claims about why employers would opt for an independent wage survey when the DOL OES wage surveys are free and easily accessible; however, based on evidence from other visa programs, it seems likely that they are doing so in order to justify paying even lower wages to H-1B workers. In the case of the H-2B, a temporary work visa for jobs that do not require a college degree, employers have long used private wage surveys to undercut the OES-determined prevailing wage rates.20 Further investigation is needed to identify the reasons employers use private wage surveys when seeking H-1B workers. Currently, more information about private wage surveys in H-1B is not readily available; DOL does not disclose the corresponding prevailing wage levels when firms use private wage surveys, so it is impossible to make comparisons with the OES wage levels.

Major U.S. firms—not just outsourcing companies—pay low wages to their H-1B employees

Much of the policy discussion around the H-1B program in recent years has focused on the problematic practices of H-1B employers that use an outsourcing business model (Table 3 identifies outsourcing companies in the top 30). Previous data analyses have revealed that H-1B outsourcing companies pay their H-1B employees relatively lower wages in absolute terms,21 and these companies’ practices have been well documented by media reports and congressional hearings: Outsourcing and staffing firms like Infosys, Cognizant, and Tata have replaced U.S. workers with H-1B workers earning tens of thousands of dollars less per year; the laid-off U.S. workers were required to train their H-1B replacements to do their former jobs—and in some cases sign nondisclosure agreements saying they would not speak publicly about their experiences—as a condition of receiving severance pay.22

The data in Tables 5 and 6 show that all companies with an outsourcing business model in the top 30—Cognizant, Deloitte, Tata, Infosys, Capgemini, Larsen and Toubro, Wipro, Accenture, IBM, Ernst & Young, Tech Mahindra, HCL America, Pricewaterhouse Coopers, MPhasis, and Syntel—had roughly half or more of their H-1B positions certified at the two lowest wage levels, and six had over 90% certified at the two lowest wage levels. These lower wage levels are consistent with the previous findings showing that outsourcing firms pay relatively lower wages to H-1B workers.23

Until now, much of the public discourse and proposals for reforming H-1B have focused on rules that would constrain the practices of these outsourcing companies. But Tables 5 and 6 also reveal a fact that has not been previously been part of the H-1B policy discussion: Many firms that employ H-1B workers directly (i.e., they do not use an outsourcing model)—including some of the biggest names in the technology industry such as Amazon, Google, Microsoft, Apple, Qualcomm, Salesforce, and Uber—pay a large share of their H-1B workers at one of the two lowest wage levels, Level 1 or Level 2. (In addition, these direct-hire firms also hire many H-1B workers on a contract basis through outsourcing firms.24)

As Table 6 shows, Microsoft, the seventh-largest H-1B employer in 2019, assigned one-third (35%) of its positions on LCAs as Level 1 and two-fifths (42%) as Level 2. In total, Microsoft assigned more than three-quarters (77%) of its H-1B positions as Level 1 or Level 2, a wage level below the local median wage. Microsoft assigned only 18% of its positions as Level 3 (the median) wage, and a mere 3% as Level 4, the only above-median wage level.

Amazon—which appears twice in the H-1B top 30, as both “Amazon.com Services” (no. 4 among the biggest H-1B employers) and “Amazon Web Services” (no. 27)—also assigned the vast majority of its H-1B positions at one of the two lowest wage levels. Amazon.com Services assigned 34% of its H-1B positions as Level 1 and 51% as Level 2, for a total of 86% of all positions certified. Amazon Web Services assigned 47% of its H-1B workers as Level 1 and 36% as Level 2. Combined, Amazon.com Services and Amazon Web Services had 12,428 positions certified at Level 1 or 2, for a total of 85% certified at a wage level below the median. Only one in eight (1,684) were certified at or above the 50th percentile (Level 3 or Level 4).

Apple, eleventh on the list, assigned 558 of its H-1B positions (2%) as Level 1 and one-third (32%) as Level 2, for a combined total of 34% at Levels 1 and 2. Apple assigned 32% as Level 3 and 34% as Level 4.

Google, ranked the fifth-largest H-1B employer, had 9,085 H-1B positions certified by DOL in fiscal 2019. Google assigned less than one-half of one percent of its certified H-1B jobs as Level 1, and 54% as Level 2. Only 37% of Google’s jobs were certified at or above the median wage.

Facebook assigned only one position as Level 1 and 10% of its 6,118 total H-1B positions as Level 2. Twenty-five percent were certified at Level 3 and 16% at Level 4. Nearly half (49%) of Facebook’s H-1B positions were certified at a wage established by an alternative wage survey making it difficult to assess its H-1B wage distribution.

Uber, the 29th-ranked H-1B employer in 2019, had 5,708 H-1B positions certified by DOL. Less than 1% were assigned as Level 1 and just over half (53%) as Level 2. Just over one-third were assigned as Level 3 and 13% as Level 4. While Uber had 5,708 H-1B positions certified by DOL and hired 1,160 H-1B workers in 2019 (see Table 3), in the same year Uber made headlines by laying off 400 employees, including 125 software engineers, nearly half of whom were “senior” software engineers. The firm was hiring H-1B workers for the same types of positions it was conducting mass layoffs. According to analysis by Ron Hira reported in The Mercury News, 1,800 of the certified H-1B positions were for “new software engineer jobs and about 1,500 for new senior software engineer jobs.” Uber’s wage-level classification for positions the firm identified as senior is questionable. The Mercury News article reported that “Uber’s applications put nearly half the senior software engineer positions at the Labor Department’s ‘Level 2’ wages, the same level it listed for more than half of the non-senior jobs.”25 The DOL’s prevailing wage guidance clearly states that, “Frequently, key words in the job title can be used as indicators that an employer’s job offer is for an experienced worker. Words such as ‘lead’ (lead analyst), ‘senior’ (senior programmer)…would be indicators that a Level [3] wage should be considered.”26 This illustrates the major weaknesses in the LCA. The employer has discretion over picking the wage level and DOL does not ensure compliance.

Qualcomm, ranked number 23, had 32,309 H-1B positions approved, with 4% assigned as Level 1 and 38% as Level 2, for a total of 42% assigned below the median wage. Salesforce, ranked number 25, is a relative newcomer to the top 30 H-1B employer list, and had wage-level shares similar to Qualcomm’s. Only 2% of Salesforce’s certified H-1B positions were assigned as Level 1 and 37% were assigned as Level 2, for a total of 40% assigned below the median wage.27

Intel Corporation, ranked number 14, had 7,409 certified H-1B positions. It did not assign any certified positions as Level 1, but it assigned one-third (33%) of the positions as Level 2, 29% as Level 3, and a mere 1% as Level 4. Intel, like Facebook, frequently set its prevailing wage through an alternative survey, accounting for more than one-third (36%) of its certified positions.

Nearly half (49%) of Walmart’s 2,056 certified H-1B positions were assigned below the median wage: 15% as Level 1 and 34% as Level 2. Thirty-nine percent were assigned as Level 3 and 11% as Level 4.

Conclusion and recommendations: Federal agencies and Congress should step in and fix the H-1B wage rule

The highest priority for H-1B reform is fixing the prevailing wage rule. The new wage-level data presented in this report make clear that most companies that use the H-1B program—but especially the biggest users, by nature of the sheer volume of workers they employ—are exploiting a flawed H-1B prevailing wage rule to underpay their H-1B workers relative to market wage standards. These largest H-1B employers include not only outsourcing companies—whose abuses of the program have been well documented—but also major U.S. firms such as Microsoft, Amazon, and Google.

The purpose of the H-1B program is to allow employers to hire workers with specialized skills that are not available in the existing local workforce.28 Specialized skills should command high wages; such skills are typically a function of inherent capability, education level, and experience. It would be reasonable to expect that these workers should receive wages higher than the median wage. One would therefore expect most H-1B positions to be assigned as Level 4 (the only wage level above the median), but as the data presented in this report show, H-1B employers as a whole assign only a very small minority of H-1B positions as Level 4—just 12%—and the top 30 H-1B employers assign even fewer H-1B positions as Level 4, just one in nine (11%).

The data in this report show the top 30 H-1B employers are in fact hiring H-1B workers to fill a very large number of routine (Levels 1 and 2) positions that require relatively little experience and ordinary skills. H-1B proponents might argue that the H-1B workers they are hiring for these routine positions are recent graduates with little experience, and therefore it is appropriate to pay them prevailing wages set far below the median. There are two problems with this proposal.

First, there is a large existing U.S. labor pool for Level 1 and 2 types of positions that could be expanded even further through private investments in training. U.S. citizens and lawful permanent residents have been graduating in record numbers with bachelor’s degrees in computer science and engineering over the past five years;29 these recent graduates can and should be filling most positions that H-1B employers have assigned as Levels 1 and 2, and they should be prioritized for those positions. But since most H-1B employers are not required to advertise H-1B positions to U.S. workers before hiring H-1B workers,30 it is unclear whether very many U.S. workers are ever afforded an opportunity to apply for these positions. Further, employers could develop the workers they need to fill these positions through training, since the positions are routine and require only modest skill levels. Instead, employers have all but disinvested in workforce training, in part because of the disincentives created by ready access to lower-paid H-1B workers.31

The second reason we should be skeptical of claims that Level 1 and 2 wages should be set low since most H-1B workers are recent graduates with little experience, is that such claims are not consistent with the available data on the characteristics of H-1B workers. In fact, USCIS data show that most H-1B workers do not fit that description: In fiscal 2018, 70% of approved H-1B petitions were for workers 30 years of age and older—a significant indicator that those workers already possess at least six to eight years of experience. Further, H-1B workers’ educational levels, which are an important determinant of skills, indicate they should be filling higher-skilled positions. In fact, 63% of all H-1B workers held an advanced degree (master’s, professional, or doctorate degree),32 meaning one could reasonably conclude that a majority of H-1B workers have the educational attainment and/or years of experience to fill positions at wage levels 3 and 4. These data suggest it is likely that H-1B employers are underpaying workers relative to their skill levels. The case of Uber assigning Level 2 wages to positions it described as “senior software engineer” may illustrate such misclassification.

The data presented in this report indicate that all H-1B employers, but especially the largest employers, use the H-1B program either to hire relatively lower-wage workers (relative to the wages paid to other workers in their occupation) who possess ordinary skills or to hire skilled workers and pay them less than the true market value of their work. Either possibility raises important policy questions about the use and allocation of H-1B visas.

By setting two of the H-1B prevailing wage levels so low relative to the median and not requiring that firms pay at least market wages to H-1B workers, DOL incentivizes firms to earn extraordinary profits by legally hiring much-lower-paid H-1B workers instead of workers earning the local median wage. The fact that firms earn those profits through poorly crafted wage rules and by underpaying H-1B workers—instead of by offering a better or more innovative product or service—means DOL has in effect made wage arbitrage a feature of the H-1B program. And as the wage-level data in this report show, nearly all H-1B employers are exploiting these H-1B wage rules in order to pay below-median wages. The top 30 employers capture a large and disproportionate share of the visas. These firms are not using the H-1B program sparingly to hire truly specialized workers and they are not using it only when U.S. workers are unavailable. Some are using the program as a substitute for workforce development.

As noted above, the existing statutory language that sets out the H-1B prevailing wage requires that there be four H-1B wage levels, but it does not prescribe specific percentiles, and no law requires DOL to set any of these prevailing wage levels below the local median wage. To ensure that H-1B workers possess specialized skills and are fairly paid, and to protect local wage standards and eliminate wage arbitrage as a feature of the H-1B program, DOL should promulgate regulations and/or issue administrative guidance that sets the lowest (Level 1) wage to the 75th percentile for the occupation and local area, and requires that wage offers to H-1B workers never be lower than the national median wage for the occupation. Requiring and enforcing above-median wages for H-1B workers would disincentivize the hiring of H-1B workers as a money-saving exercise, ensuring that companies will use the program as intended—to bring in workers who have special skills—instead of using H-1B as a way to cheaply fill entry-level positions.

Further, to ensure that future administrations do not reduce wage levels, Congress should enact a statute setting reasonable minimums for H-1B wage levels and providing DOL with new legal authority and funding to conduct random audits of H-1B employers to verify that they are not manipulating job titles and wage levels in order to underpay H-1B workers.

Senators Chuck Grassley (R-Iowa) and Richard Durbin (D-Ill.) have pursued such legislation for over a decade, jointly introducing and reintroducing their H-1B and L-1 Visa Reform Act in the Senate, most recently in 2017. The 2017 version of the Act would strengthen the statute governing the H-1B program by eliminating the two lowest wage levels, so that H-1B workers could not be paid at a wage that is lower than the local median (50th-percentile) wage. It would also grant DOL new authority to increase audits and hire additional staff.33 Passing Durbin and Grassley’s H-1B and L-1 Visa Reform Act is the easiest and simplest solution to ensure that migrant workers with H-1B visas are never paid below the market rate according to U.S. wage standards, and that the wages and working conditions of college-educated U.S. workers are not undermined. Future legislation can and should go further by permanently setting the lowest H-1B wage level at the 75th percentile of wages surveyed for an occupation in the local area.

Endnotes

1. Daniel Costa and Jennifer Rosenbaum, Temporary Foreign Workers by the Numbers: New Estimates by Visa Classification, Economic Policy Institute, March 2017.

2. See, for example, Daniel Costa, H-1B Visa Needs Reform to Make It Fairer to Migrant and American Workers (fact sheet), Economic Policy Institute, April 2017.

3. H-1B and L-1 Visa Reform Act of 2017, S. 180, 115th Cong. (2017).

4. 8 U.S.C. 1182 (n)(1)(A)(i)(II).

5. Division J, Sec. 423. H-1B Prevailing Wage Level, Consolidated Appropriations Act, 2005, U.S. Public Law 108-447 (Dec. 8, 2004).

6. 8 USC §1182(p), Computation of Prevailing Wage Level, at subsection (4), reads: “Where the Secretary of Labor uses, or makes available to employers, a governmental survey to determine the prevailing wage, such survey shall provide at least 4 levels of wages commensurate with experience, education, and the level of supervision. Where an existing government survey has only 2 levels, 2 intermediate levels may be created by dividing by 3, the difference between the 2 levels offered, adding the quotient thus obtained to the first level and subtracting that quotient from the second level.”

7. Office of Inspector General, U.S. Department of Labor, Overview and Assessment of Vulnerabilities in the Department of Labor’s Alien Labor Certification Programs, Report No. 06-03-007-03-321 (September 30, 2003).

8. A statutorily mandated annual limit of 65,000 H-1B visas may be issued and an additional 20,000 may be issued to persons who have obtained a master’s degree or higher from a U.S. university, for a total H-1B annual “cap” of 85,000. In addition, as specified by law, certain employers are exempt from the H-1B cap, including institutions of higher education or related nonprofit entities, nonprofit research organizations, and government research organizations.

9. See discussion about prevailing wage levels in Wage Methodology for the Temporary Non-Agricultural Employment H-2B Program, U.S. Department of Homeland Security and U.S. Department of Labor, 80 Fed. Reg. 24155 (April 29, 2015).

10. Government Accountability Office, H-1B Visa Program: Reforms Are Needed to Minimize the Risks and Costs of Current Program, GAO-11-26 (January 2011), Table 5.

11. While LCA disclosure data are available from DOL’s Office of Foreign Labor Certification for fiscal 2016, information about H-1B wage levels was not included in the data set.

12. To calculate these averages, we reviewed the wage differentials between wage levels 1 vs. 3 and 2 vs. 3 in the FLC’s prevailing wage data for July 2018 to June 2019, for all geographic areas and for all computer occupations except Computer User Support Specialists (SOC code 15-1151). Data are available at FLC Data Center – Downloadable Files, https://flcdatacenter.com/Download.aspx (retrieved February 5, 2020).

13. For more background, see the H-1B discussion in Daniel Costa and Jennifer Rosenbaum, Temporary Foreign Workers by the Numbers: New Estimates by Visa Classification, Economic Policy Institute, March 2017.

14. USCIS H-1B Employer Data Hub, https://www.uscis.gov/h-1b-data-hub.

15. For background on the H-1B outsourcing business model, see Ronil Hira, “The Impact of High-Skilled Immigration on U.S. Workers,” testimony before the U.S. Senate Committee on the Judiciary, Subcommittee on Immigration and the National Interest, Washington, D.C., February 25, 2016.

16. Senator Richard Durbin, “How American Jobs Are Outsourced,” YouTube video, 1:08, posted April 15, 2016.

17. Because the USCIS data do not contain a unique identifier for each company, some unknown but likely small percentage of the separately listed employers in the USCIS Employer Data Hub may be either the same company or affiliated through related entities.

18. The number of LCA positions requested and positions certified found in the raw microdata, which we discuss here and which are available from DOL’s Office of Foreign Labor Certification (OFLC), differ slightly from what is reported on the OFLC’s “Selected Statistics” fact sheet for fiscal 2019. We do not know what accounts for this discrepancy, but we rely on the microdata for our analyses of LCA data throughout this report. See U.S. Department of Labor, Office of Foreign Labor Certification, Labor Condition Applications Disclosure Data, Fiscal Year 2019, downloadable at https://www.foreignlaborcert.doleta.gov/pdf/PerformanceData/2019/H-1B_Disclosure_Data_FY2019.xlsx; and U.S. Department of Labor, Office of Foreign Labor Certification, “H-1B Temporary Specialty Occupations Labor Condition Program – Selected Statistics, FY 2019,” data as of September 30, 2019.

19. The ultimate approval rate may be higher than this. When an LCA is denied (e.g., because the application is incomplete or contains inaccuracies), the employer may resubmit the request for certification as a new application. If most or all resubmissions are approved (as they likely are if they are filled out correctly), then the ultimate approval rate is likely to be much closer to 100%, though that is not reflected in the raw data because DOL treats resubmitted LCAs as new applications.

20. For more discussion about employers using private wage surveys to establish prevailing wages in the H-2B context, see Daniel Costa, “H-2B Crabpickers Are So Important to the Maryland Seafood Industry That They Get Paid $3 Less per Hour Than the State or Local Average Wage,” Working Economics Blog (Economic Policy Institute), May 26, 2017; and Daniel Costa, The H-2B Temporary Foreign Worker Program: For Labor Shortages or Cheap, Temporary Labor? Economic Policy Institute, January 2016.

21. See, for example, Haeyoun Park, “How Outsourcing Companies Are Gaming the Visa System,” New York Times, November 10, 2015; U.S. Immigration and Customs Enforcement, “Indian Corporation Pays Record $34 Million Fine to Settle Allegations of Systemic Visa Fraud and Abuse of Immigration Processes” (press release), October 29, 2013; and Ronil Hira, “The Impact of High-Skilled Immigration on U.S. Workers,” testimony before the U.S. Senate Committee on the Judiciary, Subcommittee on Immigration and the National Interest, Washington, D.C., February 25, 2016.

22. See, for example, Stef Kight, “U.S. Companies Are Forcing Workers to Train Their Own Foreign Replacements,” Axios, December 29, 2019; Julia Preston, “Pink Slips at Disney. But First, Training Foreign Replacements,” New York Times, June 3, 2015; Michael Hiltzik, “A Loophole in Immigration Law Is Costing Thousands of American Jobs,” Los Angeles Times, February 20, 2015; Ron Hira, “New Data Show How Firms Like Infosys and Tata Abuse the H-1B Program,” Working Economics Blog (Economic Policy Institute), February 19, 2015; and Bill Whitaker, “Are U.S. Jobs Vulnerable to Workers with H-1B Visas?” 60 Minutes, August 13, 2017.

23. See, for example, Ronil Hira, “The Impact of High-Skilled Immigration on U.S. Workers,” testimony before the U.S. Senate Committee on the Judiciary, Subcommittee on Immigration and the National Interest, Washington, D.C., February 25, 2016.

24. Newly disclosed LCA data from DOL indicate that Microsoft has major contracts with Tata Consultancy Services, Wipro, and HCL, by which those firms place large numbers of lower-paid H-1B workers at Microsoft’s worksites. Similarly, Google contracts with outsourcers Accenture, Deloitte Consulting, HCL, and Wipro to provide lower-paid H-1B workers at Google’s facilities. Authors’ analysis of U.S. Department of Labor, Office of Foreign Labor Certification, Labor Condition Applications Disclosure Data, Fiscal Year 2019, downloadable at https://www.foreignlaborcert.doleta.gov/pdf/PerformanceData/2019/H-1B_Disclosure_Data_FY2019.xlsx.

25. Ethan Baron, “H-1B: Uber Snatches Up More Foreign-Worker Visas as It Lays Off Hundreds of Employees,” Mercury News, October 17, 2019.

26. U.S. Department of Labor, Employment and Training Administration, Prevailing Wage Determination Policy Guidance, Nonagricultural Immigration Programs, revised November 2009.

27. The 40% total is calculated using unrounded numbers.

28. The U.S. Department of Labor describes the purpose of the program this way, “The intent of the H-1B provisions is to help employers who cannot otherwise obtain needed business skills and abilities from the U.S. workforce by authorizing the temporary employment of qualified individuals who are not otherwise authorized to work in the United States.” See “H-1B Program” (web page on the DOL website, Wage and Hour Division, at https://www.dol.gov/agencies/whd/immigration/h1b.)

29. See National Science Board, Science & Engineering Indicators 2018, “Appendix Table 2-21. Earned Bachelor’s Degrees, by Sex and Field: 2000–15,” and “Appendix Table 2-22. Earned Bachelor’s Degrees, by Citizenship, Field, Race, and Ethnicity: 2000–15.”

30. U.S. Department of Labor, Wage and Hour Division, “Fact Sheet #62O: Must an H-1B Employer Recruit U.S. Workers Before Seeking H-1B Workers?” (July 2009).

31. For example, see the following work by Peter Cappelli: “Why Companies Aren’t Getting the Employees They Need,” Wall Street Journal, Oct 11, 2011; and “Skill Gaps, Skill Shortages, and Skill Mismatches: Evidence and Arguments for the United States,” ILR Review 68, no. 2 (March 2015): 251–290.

32. U.S. Citizenship and Immigration Services, Characteristics of H-1B Specialty Occupation Workers, Fiscal Year 2018 Annual Report to Congress, Tables 5 and 7.

33. H-1B and L-1 Visa Reform Act of 2017, S.180, 115th Congress (2017–2018); see also Senator Chuck Grassley, “Grassley, Durbin to Introduce Reforms to Skilled Non-Immigrant Visa Programs” (press release), January 19, 2017.