Date: Mon, 2 Jul 2007 15:17:30 -0700 From: Norm Matloff To: Norm Matloff Subject: Re: Dean Takahashi expanded column on H-1B To: H-1B/L-1/offshoring e-newsletter On Thu, Jun 28, 2007 at 03:55:54PM -0700, Norm Matloff wrote: Here I will post an update to my analysis of Dean Takahashi's column last week in the San Jose Mercury News. (For newcomes to this e-newsletter, that posting is at http://heather.cs.ucdavis.edu/DeanTakahashi.txt) Here is the relevant exceprt from my posting: > Dean says: > > # In addition, Matloff says, "If there really is a shortage of tech > # workers, why have wages in computer science and electrical engineering > # been flat since 1999." > > # Robert Hoffman, vice president of government and public affairs at > # Oracle, says wages in the electrical engineering category rose 34 > # percent from 1997 to 2005. And he says Department of Labor data shows > ... > I gave Dean a citation for my claim, a BusinessWeek study. I also > emphasized that in that study the salaries were adjusted for inflation. > I'm willing to bet Dean a dozen doughnuts that (a) Hoffman offered no > evidence for his claim, (b) Dean didn't ask him for any evidence, and > (c) Hoffman's data did not adjust for inflation and were otherwise > misleading. I'll address Hoffman's claim presently, but first let me give the data that I referred to. Michael Mandel, BusinessWeek's chief economist, studied the starting salaries for new Bachelor's degree graduates in various fields, publishing his findings on Sept. 15, 2005 (http:// www.businessweek.com/the_thread/economicsunbound/archives/2005/09/ good_time_to_le.html). He got his data from the National Association of Colleges and Employers (NACE), which is the main organization to track salaries of new graduates. As any good ecnomist would, HE ADJUSTED FOR INFLATION, and found the following: CHANGE IN REAL STARTING SALARIES, 2001-2005 Computer Engineering -12.0% Electrical Engineering -10.2% Psychology -9.3% Sociology -4.1% Computer Science -12.7% Marketing -6.5% Business Administration -5.7% Accounting -2.3% Later I performed my own analysis of the NACE data, going back to 1999, and also doing a similar analysis at the Master's level. The latter is important, as the industry lobbyists often claim that the industry needs H-1Bs because not enough Americans pursue advanced degrees. (There was not enough data at the PhD level, but it is irrelevant anyway, since almost no jobs in industry require a PhD.) In both cases--extension of the Bachelor's data back to 1999, and then performing a separate analysis at the Master's level--I find the same trends as BusinessWeek did (www.cis.org/articles/2006/back506.html), naamely flat or falling salaries. OK, now to the data cited by Hoffman. Note carefully that he is talking about overall salaries, not salaries for new graduates. Though you will see below that as I speculated in my posting, Hoffman DID NOT ADJUST FOR INFLATION and in fact his data CONFIRMS THAT WE DO NOT HAVE A TECH LABOR SHORTAGE, I do wish to state that one must be careful when working with overall salaries. They do not control for factors such as time trends in education level, and most importantly these days, do not control for the effects of offshoring, as follows: When lower-level, and thus lower-salary, jobs go offshore, the jobs remaining will be higher-level and thus have higher salaries than those of the jobs shipped out. The result is that the average salary in the field goes up EVEN IF NO ONE'S INDIVIDUAL SALARY GOES UP A NICKEL. This point will come up again below, so I'll give it a name, the "salary shift phenomenon." The main government source of salaries is OES, compiled by the Bureau of Labor Statistics (BLS). You can access these at www.bls.gov/oes (click on "Get detailed statistics"). However, for convenience I will use the BLS-sourced data in a UC Irvine research paper on the Web at pcic.merage.uci.edu/papers/2007/ImpactsOfGlobalization.pdf Again, this is the same BLS source, just more conveniently packaged. Here is their Table 3: TABLE 3. Mean Annual Wage of Selected Engineering Occupations in the Computer Industry, 1999-2005 Year 1999 2000 2001 2002 2003 2004 2005 Computer software engineers, applications $70,630 $74,350 $78,240 $81,270 $85,570 $95,180 $94,760 Computer software engineers, systems software $70,150 $76,130 $81,180 - - $91,430 $92,030 Computer hardware engineers $74,880 $78,760 $83,940 $82,820 $96,540 $96,980 $94,690 Electrical engineers $67,030 $71,870 $73,210 $75,490 $80,180 $82,810 $84,820 Electronics engineers, except computer $68,920 $70,940 $75,580 $76,930 $81,320 $85,270 $86,330 (If the columns of the table do not line up properly in your e-mail utility, pleass see the original document on the Web.) Here are the percent increases between 1999 and 2005, NOT ADJUSTED FOR INFLATION: Computer software engineers, applications 34% Computer software engineers, systems software 31% Computer hardware engineers 29% Electrical engineers 27% Electronics engineers, except computer 25% Between 1999 and 2005, the Consumer Price Index rose 17%, so in each case above we must subtract 17%. For Hoffman, who claimed a 34% wage growth for electrical engineers since 1997, two further years than our presentation here, he should have subtracted 22%, the inflation amount from 1997 to 2005. So we have confirmed that HOFFMAN DID NOT ADJUST FOR INFLATION. So, let's subtract the 17% inflation figure from the salary growth figures above: Computer software engineers, applications 17% Computer software engineers, systems software 14% Computer hardware engineers 12% Electrical engineers 10% Electronics engineers, except computer 8% Now, as I said above, these increases may not be real, in the sense of the offshoring effect ("salary shift phenomenon") I described above. As the entry-level software development work gets shifted overseas, the remaining jobs are mainly at the higher level, causing the computed average to rise EVEN IF SALARIES AT ANY GIVEN LEVEL DON'T RISE. But even if one takes those increases as real, the largest figure there, 17%, works out to an increase of only 2.6% per year, compounded. So, even under an interpretation generous to the industry lobbyists, i.e. ignoring the "salary shift phenomenon," YEARLY SALARY INCREASES HAVE BEEN TINY, AND THUS AGAIN CONFIRM THAT WE DO NOT HAVE A TECH LABOR SHORTAGE. As I have mentioned before, the lack of a tech labor shortage was also confirmed in a different way by Vivek Wadhwa, a former tech CEO who has recently served as an adjunct professor at Duke University. In order to investigate the industry's shortage claim, he went right to the source--he conducted an employer survey (which you can read at www.memp.pratt.duke.edu/downloads/Duke_Industry_Trends_in_Engineering_ Offshoring_10_24_06.pdf). Wadhwa found that we do NOT have a tech labor shortage, and has said so in numerous venues since then, including his BusinessWeek column (www.businessweek.com/smallbiz/content/jul2006/ sb20060710_949835.htm), his NPR interview ("Engineer Shortage? Duke Study Says No," April 30, 2007, http://www.npr.org/templates/story/ story.php?storyId=9910492) and most interestingly in his testimony to the House Committee on Education and the Workforce, May 16, 2006. I highlight the latter because I'll bet there isn't a single staffer on the Hill who is aware of Wadhwa's findings, in spite of all the hoopla on H-1B. Of course, congressional staffers are also not aware of the two studies Congress itself sponsored in which many surveyed employers were actually admitting that they pay H-1Bs less than Americans, and hire them for that reason. All the staffers "know" is what the ever-present and increasingly strident industry lobbyists tell them. One final point: The industry lobbyists' strategy this year has been to make frequent use of the buzzword "innovation," with the claim that we need the H-1Bs to stay innovative. Needless to say, it's a red herring; the vast majority of H-1Bs are not hired for the purpose of innovation, certainly not at salaries of $60,000 per year, even accounting for the exploitation factor. The real innovators in the field make well over $100,000. But much more importantly, the point is that it has long been recognized that innovation is America's forte'. Given that, why would we want to bring in so many people from the putatively less-innovative cultures? It is interesting that in both of the academic studies I've cited above, the employers themselves characterized Americans as being more innovative. The Duke study wrote # On the question "what capabilities do your US engineers have that # make it advantageous to keep their jobs in the US", the response was # that U.S. engineers...[are] more creative, excelled in problem # solving, risk taking, networking and [have] strong analytical # skills... The UC Irvine study summed up the comments of employers they surveyed by quoting have four-year degrees. According to one interviewee as saying that China's engineers "work perfectly at doing what they have been told, but cannot think about what needs to be done; they lack both creativity and motivation. They are good at legacy systems, but not new things; they can't handle `what if' situations." In this regard, it is worth noting an article written by an engineering professor in China ("China's New Engineering Obstacle," by Chen Lixin, Prism, pub. by the American Society for Engineering Education, September 1999). Chen warns his nation that the engineers being produced by Chinese universities are not good enough for China to compete in the global high-tech market. Professor Chen says the educational system in China produces students who cannot think independently or creatively, and cannot solve practical problems. He writes that the system "results in the phenomenon of high scores and low ability." This turn of phrase captures the problem quite succinctly. To be sure, SOME of the Indian and Chinese foreign engineers in the U.S. are truly outstanding, including as innovators, and I have always strongly supported facilitating their immigration. But they are the exceptions. Contrary to the industry lobbyists' calculated buzzword, the hiring of H-1Bs is not about innovation Norm