Date: Wed, 1 Apr 2009 23:04:59 -0700 From: Norm Matloff To: Norm Matloff Subject: WSJ asks researchers for opinions of "1 H-1B creates 5 jobs" To: H-1B/L-1/offshoring e-newsletter There was a column in the Wall Street Journal today by "Numbers Guy," Carl Bialik, concerning the claim by NFAP/Stuart Anderson that each H-1B worker creates five new jobs, oft-repeated by the industry lobbyists. It's amazing that an assertion as obviously flawed as this one, with textbook statistical errors, has gotten so much attention. But then that's why the lobbyists are paid the big bucks. Bialik has done a generally excellent job here, but I do have a few comments. (I have excerpted the article here, and included the full blog posting.) I'll start by stating a point Bialik missed: A very serious problem with the NFAP study is that it ignores the fact that if there is a job-creation effect, it comes mainly from the JOB, not the WORKER. We'd get the same effect by filling such jobs with a qualified American as with H-1Bs, and yes, as has been shown many times, qualified Americans do exist. Now, I wish the Numbers Guy had gotten numbers on Stuart Anderson--where does his funding come from? Anderson has been making a living promoting H-1B for 15 years, and he literally wrote the H-1B legislation that expanded the program in 1998 and 2000, when working as a Senate staffer. Given that he is cited by the immigration lawyers so much, I've long suspected that his NFAP organization receives funding from the AILF, which is the research arm of the American Immigration Lawyers Association. The AILA sells a tape of featuring Anderson speaking on behalf of the AILF (www.csctapes.com/tapes/aila2003cd.htm), together with two other speakers, including Dan Siciliano, who is an AILF official (heather.cs.ucdavis.edu/Archive/NotBestAndBrightest3.txt). Now, concerning Bialik's own comments, let's start with this one: # With the filing period for employers beginning April 1, immigration # advocates argue that granting more H-1B visas to foreign nationals # creates jobs for Americans. Critics dispute that notion and say that # financial firms receiving bailout funds have been hiring foreign # workers while laying off tens of thousands of Americans. Both sides are # playing fast and loose with the numbers. That last sentence is very unfortunate. The only "critics" that Bialik finds wanting in terms of quality of analysis are the AP and Sen. Bernie Sanders. Putting aside the question of whether these two sources are accurate in their criticism, it must be pointed out--and Bialik should have pointed out--that neither the AP nor Sen. Sanders are professional researchers, and that Bialik's article does not take issue with the analysis done by the three professional researchers he cites in his blog on the H-1B-critic side, namely Ron Hira, John Miano and me. See http://blogs.wsj.com/numbersguy/the-debate-over-h-1b-visa-number s-652 I've found all the ruckus about the AP study to be an unfortunate distraction to the H-1B debate. The real issue is the banks' past (and by the way current, as seen in the last few days with Wachovia) poor track record regarding foreign workers. For details, see http://heather.cs.ucdavis.edu/Archive/BankOfAmWadhwaHira.txt Anyway, Bialik gets right to the heart of the matter, both in his blog, # I sent the study -- which has been cited by Microsoft chairman Bill # Gates, New York mayor Michael Bloomberg and Republican Senator Judd # Gregg of New Hampshire -- to several statisticians, and most questioned # the findings. and in the article # Microsoft Chairman Bill Gates, New York mayor Michael Bloomberg and # Republican Senator Judd Gregg of New Hampshire all have cited a study # they claim shows that each one of these visas awarded to technology # companies creates five jobs. Mr. Gates interpreted the study in # testimony to Congress last year as finding that "for every H-1B holder # that technology companies hire, five additional jobs are created around # that person." # # But the study shows nothing of the kind. Instead, it finds a positive # correlation between these visas and job growth. These visas could be an # indicator of broader hiring at the company, rather than the cause. # # Some think the claim that the study, conducted by the pro-immigration # National Foundation for American Policy, demonstrates awarding visas # boosts hiring is way off base. It "has all the scientific sense of cold # fusion," says Harvard University economist Richard B. Freeman, "though # of course it could be we have discovered the perpetual employment # expansion elixir." The middle paragraph is the key. It's a very famous statistical error, often called Simpson's Paradox, in which the relation between two variables is very misleading, due to their mutual relation to a third variable. (For details, consult a statistics textbook or see the Wikipedia entry on this topic.) Now this one really intrigued me: # Some researchers find the general premise of the study persuasive, even # if the study didn't prove it. Duke University statistician David Banks # said correlation can't prove causation, but he did think the study # "corroborates the idea that H-1B visas support job creation." It does # so, he says, by contradicting the theory that companies seek foreign # workers to replace domestic ones. In reading this, I wondered, "Who is this guy Banks, and why was he interviewed by Bialik, out of the literally thousands of statistics professors in the nation? What in the world does he have to do with H-1B?" Well, it turns out that he works with outspoken immigration attorney Bruce Hake! In fact, Banks has coauthored chapters with Hake in AILA publications. So Banks is certainly not the impartial observer ("finds the study persuasive"!) that Bialik apparently thought him to be. Bialik then cites Vivek Wadhwa, but though Vivek has done some excellent work on H-1B, he later found that his patent study had data problems, and he has explicitly stated that he never found that immigrants are more innovative than Americans. The point is that we cannot conclude that the demonstrated displacement of U.S. citizens and permanent residents by foreign workers has led to a net increase in patenting, a point made by Anthony Hayter in the blog. Norm http://online.wsj.com/article/SB123854027728075537.html#mod=todays_us_page_one THE NUMBERS GUY Wall Street Journal, MARCH 31, 2009, 11:18 P.M. ET Work-Visa Numbers Get Squishy -- and Get Played Both Sides in the Debate Over Employee Immigration Policies Misuse Data to Advance Their Positions on the Issue By CARL BIALIK The recipients of U.S. visas for specialized job functions are a small blip in the overall employment market. But they've taken on great symbolic importance in the past year. With the filing period for employers beginning April 1, immigration advocates argue that granting more H-1B visas to foreign nationals creates jobs for Americans. Critics dispute that notion and say that financial firms receiving bailout funds have been hiring foreign workers while laying off tens of thousands of Americans. Both sides are playing fast and loose with the numbers. "It's another form of March Madness," says Gordon Day, president of IEEE-USA, a professional association for engineers. Every year, "partisans on both sides of the immigration debate try to interpret every bit of data available to support their case." Microsoft Chairman Bill Gates, New York mayor Michael Bloomberg and Republican Senator Judd Gregg of New Hampshire all have cited a study they claim shows that each one of these visas awarded to technology companies creates five jobs. Mr. Gates interpreted the study in testimony to Congress last year as finding that "for every H-1B holder that technology companies hire, five additional jobs are created around that person." But the study shows nothing of the kind. Instead, it finds a positive correlation between these visas and job growth. These visas could be an indicator of broader hiring at the company, rather than the cause. Some think the claim that the study, conducted by the pro-immigration National Foundation for American Policy, demonstrates awarding visas boosts hiring is way off base. It "has all the scientific sense of cold fusion," says Harvard University economist Richard B. Freeman, "though of course it could be we have discovered the perpetual employment expansion elixir." There are other problems with the study: It's confined to S&P 500 technology companies. That excludes the leading users of these visas, mainly Indian technology companies such as Infosys Technologies, Wipro Technologies and Satyam Computer Services. The analysis is based on changes in global employment, rather than just domestic employment. And the study also uses an indirect measure: letters seeking approval for visas, rather than the visas themselves. Stuart Anderson, executive director of the Arlington, Va., foundation that conducted the study, said these limitations arose from the data available. He said the proxy used for visas rendered the findings conservative, since some firms file multiple letters for the same worker. Representatives for Mr. Gates declined to comment. Representatives for Mr. Bloomberg and Mr. Gregg didn't respond to inquiries. Some researchers find the general premise of the study persuasive, even if the study didn't prove it. Duke University statistician David Banks said correlation can't prove causation, but he did think the study "corroborates the idea that H-1B visas support job creation." It does so, he says, by contradicting the theory that companies seek foreign workers to replace domestic ones. Vivek Wadhwa, a senior research associate at Harvard, pointed to other studies that have shown positive relationships between immigration and innovation. But these generally lacked a hook as catchy as five-for-one. An Associated Press article provided a hook for opponents of the visa program in today's lean economic climate. The AP reported in February that the dozen U.S. banks receiving the most bailout funds had requested visas for more than 21,800 workers over the last six years, but maybe a fraction were hired. Days later, Bernie Sanders, an independent senator from Vermont, was citing the report -- and modifying its conclusions. "What has been the response of Wall Street to the loss of 100,000 of their own workers?" Sen. Sanders said, citing layoff figures from the last three months of 2008. "What these banks have announced is they are requesting 21,000 foreign workers over the next six years through the H-1B program to fill those jobs." Yet the six-year period in AP's study ended just as that period of layoffs began. Confusion over the AP stats reflects the somewhat murky nature of H-1B visas. The visas last for three years and can be extended to six years. Employers begin by applying to the Labor Department for each potential hire, sending multiple applications if the worker may be based in various locations. Once those are approved, employers petition U.S Citizenship and Immigration Services for the visa, which is generally granted unless the cap of 65,000 is reached. The AP wasn't counting petitions but the larger number of applications since these data were more readily available. Mr. Anderson, however, counted fewer than 1,000 petitions granted to 12 of the largest bailout recipients in 2008. He also compared the number of workers receiving visas last year to the total work force, finding it was fewer than 1% for each bank. Sen. Sanders's co-sponsor on an amendment restricting banks' use of the visas, Republican Sen. Charles Grassley of Iowa, also confused the number of applications with the number of workers. Sen. Grassley's spokeswoman didn't respond to a request for comment. While the AP may have double- or triple-counted some prospective workers, Mr. Anderson, in turn, may have overplayed his hand. As an AP spokesman noted, Mr. Anderson initially counted zero visas for Capital One rather than 104, because he misspelled the bank's name as "Capitol One." Also, it would have been fairer to compare the visa numbers to new hires, instead of comparing them to the total work force, as he did. Several banks corroborated the notion that this isn't a major source of employees, though no bank would fully provide all relevant numbers. A Citigroup spokeswoman said holders of H-1B visas account for fewer than 1% of all employees; the equivalent figure was fewer than 0.04% at Wells Fargo. A Bank of America spokeswoman said such workers accounted for fewer than 1% of new hires last year. No bank would say how many such workers its contractors use. "The major abuse occurs not with the banks' direct hires of H-1Bs but instead their shadow work forces," says Ron Hira, an assistant professor of public policy at Rochester Institute of Technology. Ted Bridis, head of the AP's Washington Investigative Team, notes that the AP sought information from banks, including on third-party workers, but was rebuffed. AP spokesman Paul Colford says, "The AP stands by its scrupulous, rigorous reporting." Others have backed off -- a bit. Warren Gunnels, senior policy advisor to Sen. Sanders, acknowledged that the AP numbers covered the last six years, and may overstate actual hiring of foreign workers. Then he suggested that the numbers, though they were used to back the amendment requiring banks to prove they can't find qualified Americans, aren't critical. Even if alternative, lower numbers such as Mr. Anderson's are more accurate, "why would the supporters of increasing H-1B visas be so concerned about our amendment?" Mr. Gunnels asks. "At a time when the unemployment rate is soaring, do you really think there aren't enough intelligent Americans to fill these jobs?" Write to Carl Bialik at numbersguy@wsj.com About Carl Bialik Carl is a former technology reporter for the Online Journal who lives in Brooklyn, N.Y. In addition to the Numbers Guy, he co-writes The Daily Fix, a sports column that appears each weekday morning on WSJ.com. Carl has a degree in mathematics and physics from Yale University. Email:numbersguy@wsj.com http://blogs.wsj.com/numbersguy/the-debate-over-h-1b-visa-numbers-652/ The Debate Over H-1B Visa Numbers By Carl Bialik My print column this week examines statistics used on both sides of a debate about a program providing U.S. visas for workers from overseas. Proponents of the H-1B visa program point to a study they claim demonstrates that hiring such workers boosts domestic employment. Critics of banks' use of the program at a time they are receiving bailout funds and laying off U.S. workers point to a report about the number of such hires. Both sides appear to be taking some liberties with the numbers. globe I sent the study -- which has been cited by Microsoft chairman Bill Gates, New York mayor Michael Bloomberg and Republican Senator Judd Gregg of New Hampshire -- to several statisticians, and most questioned the findings. "Can we be sure that the extra U.S. nationals wouldn't have been hired anyway, regardless of the H-1B visa hires?" asked Anthony Hayter, a statistician at the University of Denver's Daniels College of Business. The study itself said, "most changes in a company's employment are not driven solely by their temporary-visa high-skilled hires. It is likely that increased hiring of both H-1Bs and other workers are both influenced primarily by business opportunities specific to the firm." "It is hard to prove that one thing led to another," said Vivek Wadhwa, a senior research associate at Harvard, who added, "There are other studies which make lesser claims but are much more defensible." He cited these studies as examples. (Wadhwa co-wrote a recent paper on a similar theme.) Meanwhile, Sens. Bernie Sanders, an Independent from Vermont, and Charles Grassley, an Iowa Republican, cited an AP report in claiming that major banks receiving bailout funds sought to hire 21,800 workers over the last six years. But the AP was counting applications to the Department of Labor, which can be filed multiple times for a single worker who may be stationed in more than one location. The applications "are more a wished-for number that the HR types file to cover any eventuality, while the petitions reflect ultimate hiring," said Lindsay Lowell, director of policy studies at Georgetown University's Institute for the Study of International Migration. The actual numbers for individual banks were lower in 2007 than the AP report suggested. "Our members really account for a very, very small percentage of the H-1B visas," said Diane Casey-Landry, chief operating officer of the American Bankers Association. What do you think? How significant are these visas to overall employment? Should the program be limited or expanded in tough economic times? Please let me know in the comments. Further reading: John Miano and Norm Ratloff offer detailed criticisms of the job-creation study. Grassley and Sanders cited 100,000 layoffs by banks in the fourth quarter of last year. The Wall Street Journal's Real Time Economics blog tracked those grim numbers and has a tally for the first quarter of this year. The Journal also reported on other visa restrictions last week. Ron Hira argued in 2007 that guest workers promote outsourcing, an argument bolstered by this report from Charlotte this week.