|
||||||||
|
Outsourcing brainpower
By PETER D. ZIMMERMAN
© St. Petersburg Times, Forty years ago, the United States had a pretty good shoe industry. You could walk into any department store or shoe store and buy a pair of dress shoes or sneakers, and almost every one would say "Made in USA" somewhere. A few high fashion products came from Italy, Switzerland or England. Forty years ago, the United States also had a flourishing consumer electronics business. McIntosh, Sherwood, Fisher, and Harman-Kardan products, just to name a few brands, were all made in American factories. Some Swiss, German, and English turntables and tape recorders were competitive in the market, and for high style, Bang & Olufsen of Denmark had already found a niche. There wasn't a top-notch Japanese tuner or amplifier on the market. Twenty years ago, most home computers and their peripherals were made in the United States. Even such things as disk drives, motherboards and memory chips came from American factories, made by American workers. Today it's hard to find an affordable shoe made in the United States, although U.S. brand names continue. It's even harder to find an affordable piece of stereo equipment made here, although some American names continue to show up in the marketplace, and some of the highest-end equipment still emerges from on-shore factories. Most consumer electronics sold in the United States are fabricated in Japan, China, Taiwan, Malaysia, or Thailand, with production moving from higher- to lower-cost countries as time goes on. The economic point is that Americans are best at pushing technology's limits, not producing commodities that are the same essentially no matter who makes them. For that reason, the United States makes top-line microprocessors, but buys memory chips, floppy disks and the like from the low bidder. For that reason, the United States has also been the premier innovator in computer software, providing not only the computers that power American society but also the codes that make them work. It is a distinction that is significant, especially as it applies to our national intellectual capital. It is a distinction, however, that is about to change. The U.S. computer software industry is headed the way of the shoe manufacturer. But in this case, it will be "outsourcing" something entirely different: brainpower. With American programmers in short supply, software companies from Adobe, Oracle and industry giant Microsoft Corp. have begun farming out their work. Initially, the industry sought to bring in top-notch software engineers from countries with excellent universities but limited opportunities for their graduates. More than 100,000 of these "H1-B workers," so called after the special visa program that grants them entry, have come here, worked for firms large and small, and have frequently found ways to stay. They have contributed to the American economy, are productive, and earn American-size salaries. Though successful, the H1-B program is too small to end the shortage of programmers, and it is deemed too expensive because H1-B workers earn approximately the same salaries as American workers in similar jobs. Instead, much of the software industry is turning to direct procurement from low-wage countries. An Indian programmer earns $5,000 to $8,000 a year, according to Forbes Magazine. An American doing a similar job is likely to earn at least $50,000 to $80,000. The Los Angeles Times reports that 10 percent of the software firms in Southern California already outsource their work to low-wage countries, up from barely 1 percent just two years ago. That number can only rise as firms struggle to contain costs. Labor, after all, is the major cost in software development. And India turns out spectacularly good computer scientists, as the presence in America of so many enormously successful immigrants from the Subcontinent attests. India is not, of course, our only low-wage competitor. Pakistani programmers earn just 75 percent of an Indian's wages. Chinese programmers work for even less money. Russian computer scientists earn about the same as Indians. Russian strengths in logic, mathematics and physics, together with the large number of underemployed nuclear weapons makers experienced on supercomputers, permit their outsourcing shops to go head-to-head with the best in the world, including Americans. Outsourcing is big business. Last year, the Indian computer industry earned about $6-billion from software exports. China, with only 35,000 high-level professionals, did $1-billion dollars worth of business. If the information technology industry follows the pattern of other manufacturing sectors, low-wage labor will skim off a large fraction of that American industry. But information is the technology at the core of American prosperity. The Indian Economic Times reports that Dell Computers has signed up for Indian outsourcing. Microsoft and Adobe already have Indian subsidiaries providing code for those companies' mainstream products. India is now becoming a comparatively high-cost country, and Indian software houses are in turn outsourcing less complex projects to China. As a result, Indian programmers have begun to ask their government to take steps to keep "Indian jobs in India." Businesses that outsource argue that as long as the American educational system cannot produce enough computer scientists and engineers to do the available work, using foreign labor is essential and does no harm to Americans. However, exporting the work damages the American industry in the long term if foreign programmers work for a pittance, and using cheap labor becomes a way to build profits at home. It is time to recognize that the basic programs powering our computers are manufactured products, no more and no less than steel or stereos. No matter where technology firms are located, they can compete with American businesses if they compete fairly. If software is "dumped" to U.S. customers at a cost below the fair value in its originating country, it should be subject to American anti-dumping laws. When foreign engineers work for extremely low salaries, which is the high-tech equivalent of making running shoes in sweatshops, the competition is unfair to American business and American labor. Not so long ago, one way to level the playing field between American software engineers and foreign ones in low-wage countries would have been to charge a tariff on imported software. The 1996 Singapore Ministerial Declaration on Trade in Information Technology Products obligated the developed countries to remove all duties on IT products by 2000, while developing countries were allowed until 2005, depending on the country. Thus, the historic course of action is not one the United States can take. Non-tariff barriers, such as quotas, irrelevant specifications, and explicit government subsidies are also prohibited by the World Trade Organization, and they should be. Other nations erected such barriers against U.S. products, and the American government has opposed them fiercely because they hurt our own companies. Only a sophisticated and progressive approach to the problem of a technology drain to low-wage countries is likely to succeed. Because there are too few American graduates in software engineering and programming each year to satisfy the demand for educated brainpower, American industry currently has little choice but to import foreign programmers under the expanded H1-B visa program. That initiative should be continued, and new rules adopted to allow H1-B programmers the opportunity to find new jobs in the United States if they are laid off. This will provide a transition solution until more U.S. students fill the pipeline. The long-term problem doesn't have a neat solution. The federal government must establish educational programs -- undergraduate scholarships and graduate fellowships -- comparable to those instituted after the United States was surprised by Sputnik. A larger pool of trained American graduates will reduce the pressure on U.S. companies to recruit abroad and to send work overseas. Unfortunately, an American labor pool does nothing to reduce the incentive to export work to maximize corporate profits. But one useful arrow does remain in the American quiver. The "Buy American Act," embodied in the Federal Acquisition Regulations since 1954, requires the federal government to restrict the purchase of manufactured goods to domestic products except in special cases. For supplies, and most software falls in that category, an article must be manufactured in the United States, and U.S. components must make up at least 50 percent of the total component cost. If the U.S. government restricted its software purchases in accordance with the Buy American Act, it would discourage the export of work and the importation of software, because a major market for information products would disappear. If an "office suite" computer program could not be sold to the government, those who needed to exchange files with government offices would probably choose another package that Uncle Sam could buy. The usual software package is a colorful box containing a CD-ROM with a manual. Superficially it would seem that software products meet the Buy American requirements, but most of the value in the box lies not with the media or the paper. It is in the ones and zeroes that make up the program itself. Mainframe software comes in bigger boxes but the principle remains. The value is in the code. At least half the value of the code in any software bought by the government should be added in the United States by domestic programmers. That seems fair to American workers, to American taxpayers, and ought to be no more burdensome to U.S. companies than complying with the other Buy American provisions of the Federal Acquisition Regulations. There are clear problems with this approach, the biggest of which is determining what is the cost of the code and of the foreign-written components. If Indian or Russian wages are low enough, they will always be less than half the cost of writing any program, no matter what their contribution to the product. Potentially the restriction could instead be imposed on the actual lines of code. There is no single method that will deal appropriately with the software outsourcing. Establishing the actual foreign contributions to a product is too complex a problem to solve so quickly. The appropriateness of invoking the Buy American Act as opposed to setting countervailing customs duties must be studied at length, and the kind and amount of funding to sustain a full pipeline of American programmers is certainly a matter for discussion and bargaining. But the need to ensure a vigorous American software industry at the absolute cutting edge of information technology and innovation is not open to discussion. Information has been the bedrock of American prosperity for more than a decade and will remain so for the indefinite future. We cannot swap our dominant industry for cheap software. Software isn't tennis shoes. -- Peter Zimmerman was the science adviser for Arms Control in the U.S. State Department during the Clinton Administration. © 2006 • All Rights Reserved • Tampa Bay Times
490 First Avenue South St. Petersburg, FL 33701 727-893-8111
|
From the Times Opinion page Editorials Letters Columns |
![]()