Date: Mon, 19 Jun 2006 20:37:53 -0700 From: Norm Matloff To: Norm Matloff Subject: EE Times article on H-1B wages To: H-1B/L-1/offshoring e-newsletter Enclosed below is an interesting article from the Electronic Engineering Times, different from most. I have a few comments. First and foremost, the author here correctly notes that Stuart Anderson, a pro-industry writer (more on him later), challenged John Miano's published study on the LCA data, on the grounds that the "wage rate" column in the data is not necessarily the wage paid. The author also correctly quotes me as saying that statistically speaking, the LCA data is in fact sound. But the "So there!" tone in the final quote of me is taken a bit out of context, and does not convey the point I was making, which was that the percentiles one sees in the LCA data are reasonably close to those for the BCIS (INS) and census data, both of which are based on actual wages. The other important point is that the LCAs allow you can see what each firm considers to be prevailing wage, often quite low. For instance, one often sees firms in the LCA claiming that a programmer's prevailing wage is $40,000, in spite of the fact that the average wage for even new graduates is over $50,000, and much higher for experienced workers. KEEP IN MIND THAT IN MOST CASES THIS IS PERFECTLY LEGAL, DUE TO THE HUGE LOOPHOLES IN THE LAW. To me, this is one of the major values of the LCA data, in that it shows how huge the loopholes are. That's why the following passage is incorrect: # LCA "wage rate" and "prevailing wage" data are not actual salaries. But # the average salaries calculated by EE Times based on that data indicate # that employers are paying H-1B workers less than Bureau of Labor # Statistics wage estimates. That's illegal, according to the Department # of Labor, which administers the H-1B program. The department requires an The key point is that the definition of prevailing wage in the law and the regulations is quite different from what's in the BLS data that the author refers to here, for various reasons. First, the BLS data referred to here give the median wage for all people in the given occupation, whereas the prevailing law and regulations allow the employer to define prevailing wage according to experience level. Since most H-1Bs have just a few years of experience, the employer can get a prevailing wage level much lower than the BLS value. This is one of the major ways in which employers save money by hiring H-1Bs; recall that I call this Type II savings. In addition, the employer can use a much finer occupational system than the BLS uses. This gives rise to odd occupational titles, such as Associate Software Engineer, which can be exploited by the employer. There are many other things an employer can do. For example, prevailing wage is defined in terms of the job, not the worker. So, if the employer has a job which requires only a Bachelor's degree but hires an H-1B with a Master's, the employer gets a Master's worker for the price of a Bachelor's. The author (or likely his editor) left out an important point in the following passage: # Lower salaries undermine employers' contention that they need H-1B # workers to fill jobs for which Americans can't be found, Matloff # said. "Otherwise, salaries would be rising." What I had pointed out was that starting salaries (adjusted for inflation) for new graduates in the field have been flat since 1999, counterindicating a shortage. So, again, the underpayment of the H-1Bs is usually done in full compliance with the law, as Intel points out: # Intel Corp. compensates its H-1B workers "under the same structure as # our domestic employees," a company spokeswoman said. "It's actually # illegal to pay less than the prevailing wage for any employee sponsored # under the H-1B visa." Intel employs approximately 3,000 H-1Bs, a # majority of whom work in VLSI design, device physics or optics, the # spokeswoman said. Hopefully you, the reader, immediately detected the deceit in Intel's technically correct but highly misleading statement here. If not, please reread what I wrote above. I made an analysis of Intel's wages in http://heather.cs.ucdavis.edu/Archive/IntelH1BWages.txt Basically, Intel was $20-40K below the medians. # IBM complies with all Department of Labor wage requirements, the # spokesman added. Again, hopefully the reader sees the deceit in IBM's statement as well. # Current limits on H-1B visas essentially promote a nonimmigration policy # for the United States and endanger the competitiveness of U.S. industry, # said Stuart Anderson, executive director of the National Foundation for # American Policy, a think tank that "pursue[s] and promote[s] debate # consistent with an American entrepreneurial spirit that is welcoming to Virtually Anderson's entire career has been devoted to promoting H-1B and offshoring. He wrote pro-H-1B reports for a group called Empower America and for the ITAA, one of the main industry groups lobbying for expansion of the H-1B program since 1997. He went to work for then-Senate Spencer Abraham and was the author of Abraham's H-1B expansion bill which was enacted in 2000. He then went to work for the BCIS (INS), and now has again formed his own think tank to promulgate the industry's views. # Anderson said employers don't use H-1B visas to lower wages and decried # the "myth" that each H-1B worker replaces a U.S. worker. It's hardly a myth when major employers such as the Bank of America and Siemens admit to doing it. # Rather than preserving U.S. jobs, H-1B visa caps drive companies to # expand overseas Actually, H-1B is used to FACILITATE offshoring. The typical ratio in an offshored project is one H-1B onshore for every two offshore workers. In addition, here is a great rejoinder (Searching for Skills, Lorraine Ash, Gannett News Service, in the Asbury Park Press, August 15, 2005): % But Eileen Appelbaum, an economist and member of a National Research % Council committee that studied the impact of H-1Bs on the U.S. % economy, does not accept the way the H-1B option is typically framed: % One can have an H-1B worker in an American job, or lose that job to % exportation. % "Industry said in 2001, "Let us have the H-1B visas and we'll do the % work here, or you can say no and we'll just move the work offshore,' % " she said. "Well, they got all the H-1Bs they wanted, and they % still moved work offshore. In 2005, that's an argument industry can't % make with a straight face." Another industry line: # Less than 3 percent of Intel's employees hold H-1B visas, and more # than 50 percent of its 99,000 workers worldwide are in the United # States. That 3 percent figure includes all their nontechnical people, i.e. the clerks, the bookkeepers, the sales and marketing people, etc. If you look only at engineers, the figure is considerably higher. # IBM's total of 2,500 H-1Bs pales next to the company's 43,000 # employees in India. Again, those are not 43,000 engineers in India. The number of engineers in India is probably comparable to the number of H-1Bs in the U.S. # The number of H-1B visas issued for high-tech occupations is too few to # affect the salaries of the larger U.S. labor force, according to Jeremy # Leonard, chief economist at American Sentinel University. By 2004, a # total of 139,000 H-1B visas were issued for information technology # professionals, a broad classification that includes computer occupations # and engineers. The word "by" here apparently means that by 2004 the number of visas had declined to 139,000. But the H-1B visa is good for six years (or more, under exceptions), so large numbers of workers who had been issued visas earlier were still in the workforce in 2004. For example, just in fiscal year 2001 alone, visas were issued to 191,397 workers in the Computer-Related Applications category, and to 40,388 workers in the Architecture, Engineering and Surveying category. That's just 2001. One must then add the figures for the years 2000 and 2002-2005 in order to compare to Leonard's jobs figure for 2005 (see below). Of course, there are other issues here, e.g. the possible exit of some of the H-1Bs and the fact that the Architecture, Engineering and Surveying is not just engineering and a lot of engineering should not be considered IT. But it is certainly clear that Leonard's 139,000 is way, way too low. And keep in mind that then there are lots of L-1 visa holders on top of that. # "In comparison, the U.S. IT labor force, using a relatively narrow # definition, numbers about 3 million," Leonard said. In That's not narrow at all; it's too broad. The 2005 OES data that he uses, for instance, gives 79K computer hardware engineers, 275K EEs, 776K software engineers, 99K database administrators, 492K system administrators, 26K computer scientists and 389K programmers, for a total of 2.1 million. He's including all the technicians, call center workers and so on, which are jobs which are ineligible for the H-1B program because they don't normally require a Bachelor's degree. # While U.S. electronics industry employment levels and pay increases both # trail boom-year levels (see "Jobs data spurs debate" in By the Numbers, # June 12, page 30), H-1B visas are not to blame, Leonard said. BLS data # shows a 23.3 percent increase in hardware engineering jobs from 2000 to # 2005, and a 5.1 percent increase in electronics engineering jobs over # that span. That 23.3% figure is very misleading, as it is for a small category. In 2000, there were 64K computer hardware engineer jobs but 286K EEs, while in 2005 those numbers were 79K and 275K. The totals were 350K and 354K, essentially the same. # "So employment in these occupations certainly hasn't declined # due to H-1B visas," he said. Leonard appears not to realize that these jobs numbers he's quoting INCLUDE jobs held by H-1Bs and L-1s. One can't tell anything at all about the impact of H-1B by looking at numbers of jobs. The fact is that we had about the same numbers of jobs in 2005 as in 2000 (see above), while the numbers of H-1Bs in the country has increased by an estimated 26%, with an even greater increase in L-1. (See my recent CIS article, at http://www.cis.org/articles/2006/back506.html) The author and EE Times are to be commended for doing their own analysis of the H-1B data. Norm H-1B Pay Drags Down All Salaries By David Roman, EE Times Jun 19, 2006 http://www.ddj.com/dept/global/189500374 Immigrant engineers with H-1B visas may be earning up to 23 percent less on average than American engineers with similar jobs, according to documents filed with the U.S. Department of Labor. Salary data from Labor Condition Applications (LCAs) lends credence to arguments that lower compensation paid to H-1B workers suppresses the wages of other electronics professionals. The lower LCA-based pay rates, considered unreliable by some, raise questions about the value of H-1B workers--and of U.S. engineers in general--to their employers, and add fuel to the debate that has long swirled around the H-1B program. LCA "wage rate" and "prevailing wage" data are not actual salaries. But the average salaries calculated by EE Times based on that data indicate that employers are paying H-1B workers less than Bureau of Labor Statistics wage estimates. That's illegal, according to the Department of Labor, which administers the H-1B program. The department requires an employer to pay H-1B workers the same as other workers with similar skills and qualifications, or the prevailing wage, whichever is higher. "There are plenty of studies, including my own, that show this disparity in wages," said Norm Matloff, a professor of computer science at the University of California, Davis, who writes frequently on immigration employment and H-1B visa issues. Lower salaries undermine employers' contention that they need H-1B workers to fill jobs for which Americans can't be found, Matloff said. "Otherwise, salaries would be rising." The average H-1B salaries calculated by EE Times are based on data from 459 of 65,536 LCA petitions filed by employers seeking permission to hire immigrant professionals in federal fiscal year 2005. Specifically, the data comes from LCAs naming one of three positions commonly held by engineers: electronics engineers, electrical engineers and computer hardware engineers. The average salary cited in the LCAs for each of the three positions was below the mean annual salaries for those jobs in 2004 as determined by the U.S. Bureau of Labor Statistics' Occupational Employment Statistics survey of employers. The average annual wage or salary for electronics engineers was $69,851 in the LCAs, or 9.8 percent less than the $77,450 mean annual-wage estimate determined by the BLS OES survey. The LCA average for electrical engineers was $63,268, or 14.7 percent less than the OES survey's $74,220 mean. And the LCA average for computer hardware engineers was $64,426, or 23.3 percent less than the $84,010 average found by the OES survey. (A detailed comparison of LCA-based and OES salaries can be found in this week's By the Numbers, page 26.) Underpaid H-1B workers displace American information technology workers and put undue pressure on salaries, said Kim Berry, a software developer who serves as president of The Programmers Guild, an activist group for IT industry professionals. "This is happening under the radar across the country," he said. Prevailing wage Intel Corp. compensates its H-1B workers "under the same structure as our domestic employees," a company spokeswoman said. "It's actually illegal to pay less than the prevailing wage for any employee sponsored under the H-1B visa." Intel employs approximately 3,000 H-1Bs, a majority of whom work in VLSI design, device physics or optics, the spokeswoman said. IBM Corp. takes guidance on salaries for its roughly 2,500 H-1B visa workers from BLS OES data. "We get our numbers from the OES page," an IBM spokesman said. "We do that every time." An overwhelming majority of those workers are engineers, and their benefits package is similar to what IBMers receive. IBM complies with all Department of Labor wage requirements, the spokesman added. H-1B visas allow IBM to "tap into global sources of information," the spokesman said, echoing industry supporters of the H-1B visa program. Started in 1990, the program has polarized factions within the electronics industry. Employers say the visas allow them to hire needed talent; detractors say it puts U.S. citizens out of work, engenders fraud and promotes exploitation of immigrants. "I work with those H-1Bs, and as far as I know they are getting half of what we get," said Shahid Sheikh, a senior software developer with TAC Worldwide in Jacksonville, Fla. "I get a normal salary. I get $80,000 a year. They get a maximum $40,000 a year." Sheikh, who worked under an H-1B visa when he emigrated from Bangladesh 12 years ago, said the program is "filled with fraud and cheating." He was naturalized about two years ago. President Bush weighed in on H-1B visas in February, when he called the current annual limit of 65,000 visas a "problem" and urged Congress to "raise that cap." The U.S. Senate voted in March to increase the annual limit to 115,000 for fiscal 2007. The House hasn't taken up the issue. Employer applications for H-1Bs reached the fiscal 2006 cap of 65,000 last August, two months before the current fiscal year began on Oct. 1. Applications for fiscal 2007 have already maxed out the 65,000-visa allotment. Current limits on H-1B visas essentially promote a nonimmigration policy for the United States and endanger the competitiveness of U.S. industry, said Stuart Anderson, executive director of the National Foundation for American Policy, a think tank that "pursue[s] and promote[s] debate consistent with an American entrepreneurial spirit that is welcoming to new people, ideas and innovation," according to the group's Web site. "Except for a short window of time, no one can hire a new H-1B because the cap keeps getting hit before the experiment's even started," Anderson said. Anderson said employers don't use H-1B visas to lower wages and decried the "myth" that each H-1B worker replaces a U.S. worker. Immigrants help create jobs and innovation, he said. Rather than preserving U.S. jobs, H-1B visa caps drive companies to expand overseas to preserve flexibility, in Anderson's view. "The more we continue current policy, the more this will happen," he said. H-1B visas account for a fraction of the work forces of multinational employers like Intel and IBM. Less than 3 percent of Intel's employees hold H-1B visas, and more than 50 percent of its 99,000 workers worldwide are in the United States. IBM's total of 2,500 H-1Bs pales next to the company's 43,000 employees in India. The number of H-1B visas issued for high-tech occupations is too few to affect the salaries of the larger U.S. labor force, according to Jeremy Leonard, chief economist at American Sentinel University. By 2004, a total of 139,000 H-1B visas were issued for information technology professionals, a broad classification that includes computer occupations and engineers. "In comparison, the U.S. IT labor force, using a relatively narrow definition, numbers about 3 million," Leonard said. In 2003, 12 percent of H-1B holders were in engineering occupations and 28 percent were in computer jobs, Leonard said. While U.S. electronics industry employment levels and pay increases both trail boom-year levels (see "Jobs data spurs debate" in By the Numbers, June 12, page 30), H-1B visas are not to blame, Leonard said. BLS data shows a 23.3 percent increase in hardware engineering jobs from 2000 to 2005, and a 5.1 percent increase in electronics engineering jobs over that span. "So employment in these occupations certainly hasn't declined due to H-1B visas," he said. Nigel Brent, president of Nigel B. Design Inc., an amplifier manufacturer based in California, said he's fed up with engineers complaining about immigrants taking American jobs. "This is so bogus," he said. "The truth is that people are lured to come here from many different countries with the promise of higher salaries, better lifestyle and standard of living than their home countries can provide." The United States has grown strong with the constant influx of immigrants, said Brent, who emigrated from Britain almost 30 years ago and became a U.S. citizen in the 1980s. "The upside with the present H-1B system is that jobs stay in America. Every H-1B who comes here supports the economy with the food they buy, the cars they buy, the house they buy. I could go on with the many benefits that the trickle effects of being here bring to the economy." Writing in the June 12 issue of the Financial Times, IBM chairman and chief executive officer Sam Palmisano said that a modern company must resist anti-globalization fervor and become a "globally integrated enterprise." The alternative is grim, he wrote. "Left unaddressed, the issues surrounding globalization will only grow. People may ultimately choose to elect governments that impose strict regulations on trade or labor, perhaps of a highly protectionist sort," said Palmisano. NFAP's Anderson criticized what he called basic flaws in using LCA wage data as a stand-in for H-1B salaries. "One is that you're comparing the prevailing wage, basically the minimum an employer would have to pay others similarly employed at the firm," he said. Second, LCA minimums wouldn't approach average salaries of all U.S. professionals, some with decades of job experience, he said. The tactic of presenting LCA wages as salaries was used in "The Bottom of the Pay Scale," a report on the salaries of computer programmers published last December by the Center for Immigration Studies, a nonprofit research organization "animated by a . . . vision which seeks fewer immigrants but a warmer welcome for those admitted," according to the CIS Web site. Professor Matloff rebuffed Anderson's criticism and raised questions about his objectivity. "He's a lobbyist," Matloff said. Using LCA data to reflect salaries is sound methodology, Matloff said, because the data "tracks very well" with H-1B compensation data in an annual report published by U.S. Citizenship and Immigration Services, a bureau of the U.S. Department of Homeland Security. "I have confidence the LCA data is reliable," Matloff said. "And I'm a former statistics professor."