Wages and High-Skilled Immigration

High-skilled foreign-born professionals are usually hired to come and work in the United States in two ways: through a temporary nonimmigrant work visa classification or an employment-based immigrant visa classification.

1. H-1B temporary work visa classification

The H-1B classification allows a foreign professional to work in the United States in a “specialty occupation” that requires at least a bachelor’s degree or the equivalent. In Fiscal Year (FY) 2016, roughly 69 percent of all H-1B visas went to workers in computer-related occupations; 8 percent went to workers in architecture, engineering, and surveying; and 6 percent to those with an administrative specialization. The H-1B classification is valid initially for a maximum of three years and generally can be extended once for up to an additional three years. An H-1B worker must work only for the employer who filed the petition and cannot freely change employers unless the new employer also files an H-1B petition on the worker’s behalf. Those employers who are recruiting on U.S. college campuses and hiring newly-minted undergraduate or graduate degree-holders may seek an H-1B visa classification for foreign students in order to continue to employ them once the employment authorization associated with their foreign student status expires.

When the H-1 visa category was created in 1952, there were no associated wage or labor certification requirements. In 1990, however, Congress revised the laws for temporary workers. The principal aim of the revisions was to ensure that H-1B workers were not being taken advantage of or used by employers to undercut U.S. workers. Thus, the H-1B classification was amended to include, for the first time, “labor market protections,” including wage requirements and a cap on the number of H-1B visas that could be issued each year. Congress mandated that H-1B workers must be offered the same wages and terms and conditions of employment as similarly-situated U.S. workers. Specifically, workers must be paid “wages that are not less than those prevailing in the occupation in the recruitment area, or the employer’s actual wage level, whichever is higher.”

2. Employment-based “green card”

Lawful Permanent Resident (LPR) status allows a foreign worker to work in the United States indefinitely and does not tie the worker to a particular employer. In 1965, Congress mandated that for most workers seeking to become LPRs based on performing skilled or unskilled labor, a labor certification would be required to establish that the worker’s employment would not have a negative impact on U.S. workers. In implementing this new requirement, the Department of Labor (DOL) relied on state employment agencies (which were responsible for unemployment programs and other job services) to determine if the wage level offered to the foreign workers was regarded as “prevailing” in a particular industry and location, and if U.S. workers were present in sufficient numbers to fill available jobs. Congress assumed that DOL’s determination would be straightforward since its Employment Security Agencies and State Workforce Agencies were already responsible for assessing the availability of U.S. workers and their working conditions.

When a U.S. employer seeks to sponsor a foreign worker for an employment-based immigrant visa classification that requires a test of the labor market, the employer must file the application for labor certification through the Program Electronic Review Management (PERM) system in order to start the process. As a prerequisite to filing the application, an employer must obtain a prevailing wage determination. This is to confirm that selecting a foreign national to fill the job will not negatively impact the wages of U.S. workers in similar jobs in the geographic area of employment. 

In order to institute a prevailing wage requirement, DOL needed data on what wages were actually prevailing. The House of Representatives at first had great confidence that the Bureau of Labor Statistics (BLS) would be able to provide that information. The House Judiciary Committee explained in 1990 that:

The bill requires the Bureau of Labor Statistics to make determinations on prevailing wages and this information is to be made readily available to employers and workers… BLS has the capability to analyze these labor markets and provide assessments as to prevailing wages in particular occupations and areas… The Committee intends and expects the BLS to have prevailing wage information readily available.

However, Congress also realized that it was not necessary to rely solely on BLS to satisfy the new prevailing wage requirements. Existing wage surveys could be used, as the law did not require employers to use any specific methodology for complying with the wage requirements. Indeed, DOL clarified that the prevailing wage determination could be based on a wage survey conducted by a state employment agency or by any published survey that was specific enough with regard to occupation and locale of employment. Initially, DOL believed the state agencies could determine the prevailing wages. However, in the 1990s DOL found that these state agencies were not interpreting prevailing wage guidance in a consistent manner and concluded that “the most efficient and cost effective way to develop consistently accurate prevailing wage rates is to use the wage component of the BLS Occupational Employment Statistics (OES) program.” The OES survey reaches about 200,000 establishments twice a year—with a 73 percent response rate in the May 2016 survey—and while the survey intends to provide a broad-range of occupational employment information across many locales and regions, it does not collect occupational wage level estimates. Thus, relying on BLS data became a primary way to establish prevailing wages.

At first, two wage levels for each occupation were identified based on OES survey results:

• Level 1: beginning level employees who performed routine or moderately complex tasks.
• Level 2: fully competent employees who used advanced skills.

However, publicly available, independent compensation surveys used to determine prevailing wages typically identified several wage levels reflecting different amounts of education, experience, and supervision. Some believed that more than two wage levels were necessary to capture the full range of skills. And so, in 2004, Congress amended the INA to require that the number of wage levels be expanded from two to at least four. The new system still utilizes the original two wage levels, but Congress also devised a new calculation to construct four levels. Thus, two of the four levels are artificially manufactured by a formula Congress contrived—not one made by BLS economists and statisticians. The same four levels identified by Congress in 2004 are still used today and discussed in more detail below.

Determining the prevailing wage is not a straightforward process. In fact, DOL’s instructions on assigning one of the four wage levels run more than 30 pages. Even the first step of this process is not always simple; the first step is to place the offered H-1B or PERM job into a particular detailed occupational code (one of the 840 detailed occupations identified by BLS’s Standard Occupational Classification—or SOC—system) by comparing the employer’s job requirements to the occupational requirements described in the DOL’s Occupational Information Network (O*Net) to determine the minimum requirements generally required for acceptable performance in the job being filled by the sponsoring employer. When an employer submits the Application for Prevailing Wage Determination, Form ETA-9141, DOL’s National Prevailing Wage Center uses the information contained in it to determine which wage level to assign to the position (provided that the employer did not request that DOL use an alternative wage survey). DOL will select the standard occupational classification it concludes most closely resembles the employer’s submission—which may differ from the classification suggested by the employer—and then selects the wage level based on its interpretation of the experience, education, special skills, and other requirements specified by the employer.

For each occupation, there are four possible wage levels:

• Level 1: entry level workers who perform routine tasks and receive close supervision.
• Level 2: qualified workers who perform moderately complex tasks and exercise limited independent judgement.
• Level 3: experienced workers who exercise judgement and have coordination and supervisory authority.
• Level 4: fully competent workers who perform independent evaluation, selection, modification, application of standard procedures and techniques, and generally have management or supervisory responsibilities.

The four wage levels are created by taking the difference between the two levels available through OES results, dividing by three, taking that dividend and then adding it to the first (lowest) level to create level two, and taking that same dividend and then subtracting it from the fourth (highest) level to create level three. In other words:

• Level 1: the lowest approximately 20% of all wages for that occupation.
• Level 2: the difference between Level 1 and Level 4 divided by three plus Level 1.
• Level 3: the difference between Level 1 and Level 4 divided by three and subtracted from Level 4; or the mean wage for the occupation, based on all of the wage information collected from wage surveys.
• Level 4: the highest approximately 20% of all wages for that occupation.

For example, the four prevailing wage levels for Accountants and Auditors in Washington, D.C., for employers that do not qualify to use the separate database for institutions of higher education and certain other entities (the “All Industries Database”) for July 1, 2017, through June 30, 2018, are:

• Level 1: $56,971 per year
• Level 2: $74,048 per year
• Level 3: $91,104 per year
• Level 4: $108,181 per year

Calculating the prevailing wage in an occupation is critical to ensuring that foreign-born workers do not get paid more than native-born workers (or vice versa) for performing comparable tasks. Pay inequity could give an unfair advantage to one group of workers over another. It is for that reason that the prevailing wage determination is such an important aspect of the employment-based immigration system.

However, there are three sets of problems surrounding the way in which a current Prevailing Wage Determination is issued in our high-skilled immigration system: (1) the underlying data is based on very broad pay ranges; (2) there are intrinsic weaknesses in issuing prevailing wage determinations for specific positions offered by an individual employer based on generalized occupational employment statistics; and (3) the current system does not reflect education, experience, and supervision.

1. Underlying wage data is based on overly broad pay ranges

Ideally, prevailing wage data for the H-1B classification or PERM certification would be narrowly tailored to H-1B and PERM-relevant occupations. However, the OES collects information for over 840 detailed occupations in 380 metropolitan areas and 170 non-metropolitan areas. Data for over 174,000 combinations of detailed occupations and labor market areas is collected.

Employers who respond to the OES survey do not provide data about individual employees. Instead, participating employers provide grouped data responses, categorizing employees into 12 wage groups (the pay bands were last updated in 2013). The same wage groups are used for all occupations in all geographic areas. Each establishment indicates how many employees in a detailed occupation are included in each of 12 wage groups, ranging from under $9.25 per hour to over $100 per hour. For example, if an establishment employs both software developers and administrative assistants, it will only report the number of software developers and administrative assistants in each of the 12 wage groups. The employer does not provide any information about how much each employee earns by occupation. The wide wage intervals make the OES survey less precise and less likely to accurately identify the average wages in any particular occupation or any geographic area.

In order to report estimates of average hourly wage rates for each detailed occupation, OES must supplement its survey with data from the National Compensation Survey (NCS). The OES uses NCS to convert reported information about groups of individuals in pay ranges to an average hourly wage rate computed to the nearest cent. The NCS canvasses about 18,000 establishments annually with the intent of providing reliable national indices, specifically on national compensation cost trends and national trends in employer-provided health care and retirement benefits. The NCS is not focused on particular occupations or geographic areas and, by definition, it has a limited number of observations of individual hourly wage rates in the key occupations. Those limited observations are distributed nationally, providing very little insight into wages in a particular geographic area of employment.

2. Prevailing wage determinations for specific positions are based on generalized occupational employment statistics

The method by which the prevailing wage is calculated suffers from several shortcomings, primarily related to (a) how few workers are sometimes represented in the available occupational data, (b) how wage data change over the course of time, and (c) how different industries pay differing wages to workers in the same occupation.

OES breadth. Because the OES survey encompasses so many occupations, not all of the resulting 174,000 average wage rates in the OES are reported with the same accuracy. The accuracy of the estimated average wage depends on the sample size (the number of establishments in the labor market area that employ workers in the occupation and participate in the survey). There are often too few workers in the OES sample to get accurate information for many occupations and geographic areas, even where private industry or compensation surveys collect such data. 

Change over time. Each reported OES average wage estimate is based on the six most recent “panels” of the survey spanning 2.5 years. For example, wage data from survey panels in May 2016, November 2015, May 2015, November 2014, May 2014, and November 2013, were used to calculate the May 2016 OES average wage rate estimates. Using six “panels” of data means that information from up to 1.2 million establishments (200,000 x 6) is reflected. But this also means that wage data from different survey periods are not equivalent in real-dollar terms due to inflation and changing compensation costs. Consequently, DOL uses data from the Employment Cost Index (ECI) to adjust older OES data to the current survey.

Variation among employers. When reporting average wages by occupation and labor market area, the OES assumes that all types of employers pay the same wage on average for the same occupation. This assumption is clearly false; in many occupations, different types of employers pay higher or lower wages. For example, it is well understood that there are differences in most locales between universities, not-for-profits, government, and for-profit private sector entities, along with differences between entity size and industry, among other factors, within these sectors.

OES also includes as wages certain forms of compensation above the base salary, such as commissions or incentive payments (meaning the amount relates to actual individual or group performance). This means that the wages being averaged do not have a uniform source. OES also excludes nonwage benefits such as health or disability insurance or employer payments to retirement plans, even though they have significant value to employees.

3. Current data does not reflect education, experience, and supervision

DOL attempts to satisfy the statutory mandate that reported wages be commensurate with education, experience, and supervision by making inferences about the wage rates of different workers in an occupation based on the pay ranges observed in the OES. Similarly, because wage rates are related to skills, economists can make inferences about the wage rates of more skilled and less skilled workers in an occupation based on the pay ranges in the OES. Yet these inferences made by professional economists are not as good as the data that would be included in a more precise survey. 

In order to overcome the deficiencies in the data now being used to make prevailing wage determinations, a more detailed survey is needed. Specifically, workers must be classified not only according to occupation, but also education, years of experience, and supervisory responsibilities. For example, in the category of Software Developers, Systems Software (SOC Code 15-1133.00), what is the wage for a person with a bachelor’s degree and 4-6 years of experience, compared to someone with a Master’s degree and 0-2 years of experience? Will that person be supervising other workers? If so, how many? Between one and five? Over 25?

In addition, all sources of wage data should be geographically-based and take into account the size of the employer. At present, OES must supplement its survey with data from the NCS, which does not take geographical location into account. For instance, the prevailing wage of a database administrator in a large corporation in New York City is going to be different than the wage of the same database administrator at a smaller company located in Oklahoma City.

The current system is furthered flawed in that the existing SOC codes do not cover all possible occupations. Sometimes, there simply is not a good match between an employer’s job description and any of the existing SOC codes. Because the federal government takes several years to issue updates, DOL has not been able to keep pace with evolving occupational classifications—particularly in the fast-changing tech sector.

Another open question is whether a new survey should be designed by the federal government (the Department of Labor, specifically) or a private-sector company. In either case, however, care must be taken not to repeat the mistakes of the past. For example, some legislators have proposed changing the number of wage levels by collapsing the current four-level system to a three-level system. Unfortunately, this approach would likely reduce the accuracy of wage determinations for immigration purposes, given that there would be fewer levels to cover all education and experience cohorts in professional occupations.

Other legislative proposals are similarly problematic, such as efforts to change how H-1B visa numbers are issued in years when the number of petitions exceeds the annual numerical cap. Rather than the current lottery system that randomly selects from among the petitions received, petitions would be prioritized for workers with the most education or those who will command the highest wage. Yet this approach simply presumes that jobs which require the most education and experience are more valuable to our national interest than entry- or mid-level jobs. It also fails to take into account the large variation in salaries across occupations and geographic regions. Moreover, this approach ignores the importance of employers engaging in legitimate on-campus recruitment at U.S. universities to hire newly-minted graduates, where some selected as the ideal candidate are foreign-born students who have education-related visas. Such new hires would almost universally be entry-level professionals earning level 1 wages, and such employers would be at a severe disadvantage in ever employing such a foreign student on a long-term basis under the proposed H-B cap lottery framework. 

Given what we know about the lack of precision in the way prevailing wage levels are calculated, Congress needs to seriously consider updating the current system. An approach far better than the status quo would address the underlying deficiencies in the data that are being collected, ensuring that the data accurately reflect differences between workers in education, experience, and level of supervisory experience. Arbitrarily changing the number of wage levels on the basis of the same flawed statistics, as Congress has done in the past, is not an effective solution. If prevailing wage determinations are to live up to their promise of maintaining a fair and level playing field for workers, they must be as accurate as possible.