CONTACT: Stanford University News Service (650) 723-2558
A global shortage of computer programmers and software professionals is about to trigger massive adjustments for industry and government alike.
That's the challenging news from Stanford University's Computer Industry Project software study team, Avron Barr, Shirley Tessler and William Miller. "Business thinking about software will undergo radical change in the next few years," the researchers say, as managers are forced to make hard choices about where to spend their software dollars. The companies likely to be hurt the most by the labor shortage are the ones least likely to realize it now.
Companies whose main product is software are already reacting to the labor crunch. Like major league baseball team owners, they are offering seven-figure signing bonuses, stock options and higher salaries to lure the most talented programmers, project managers and quality engineers. "We predict that salaries for good software people will ramp up rapidly in coming months, causing people to change jobs more frequently, which in turn will cause project delays as replacements take longer to find," Tessler and Barr wrote in a recent research brief.
Those likely to be hardest hit are government agencies, information systems departments of low-tech industries and non-profit organizations, which are least able to attract and maintain top talent.
Software, the complex sequences of instructions that enable computers to do everything from run microwave ovens to track Fed Ex packages, has become ubiquitous in products and services and is growing more complex every year. There are, for example, 300,000 lines of computer code imbedded in a cellular telephone, Barr said. New cars can contain millions of lines of code and airplanes contain tens of millions. A shortage of software workers may lead to a slowdown in the introduction of new products and services or quality upgrades, the researchers say. Such innovation historically has been one of the competitive advantages many U.S. companies have enjoyed in global markets.
Surplus labor not long ago
The demand for software has risen steadily for years, but the labor shortage comes as a surprise to many because it was masked by other trends in the late 1980s and early '90s, said Barr, who, like Tessler, works part time for Stanford's Computer Industry Project and part-time as a consultant to software companies on strategic planning. The third member of the team, William Miller, is a professor at the Graduate School of Business and directs the project, which includes research on other aspects of the industry.
The end of the Cold War prompted defense industries to downsize, throwing many software professionals out of work, Tessler said. The spread of the personal computer also prompted corporations to lay off people in their information systems departments. As a result, fewer college students enrolled in college computer science courses so that degrees granted in that field peaked in 1986.
"Senior management at many companies has been slow to realize their company is software enabled or software intensive," Tessler said, "which means that they haven't seen software development and maintenance as one of their core competencies."
On the surface, Barr said, "companies like Citibank and General Motors don't appear to be in the business of making software, yet the products and services they sell are increasingly computer reliant."
Many of those who were laid-off "are now employed at Andersen Consulting or EDS, selling their services back to the very corporations that laid them off, and they are going to become very expensive," he said.
Tessler and Barr divide the demand for software workers into four categories. The most visible segment software publishers like Microsoft and Oracle represent the smallest part of the demand, with an estimated $92 billion in business worldwide annually. Companies that write software for other companies are almost twice as large, with an estimated annual business of $170 billion. In-house information systems people who maintain and develop software to handle company functions such as payroll, assembly line automation and supply logistics are estimated to cost about $700 billion annually. The market value of the fourth category software written for computers embedded in products is probably the biggest. "We have no way to measure it because even corporations don't know how much they spend on embedded software," Barr said.
Not everyone believes the labor shortage is a long-term systematic problem. Many of the business people with whom Tessler has met see their current hiring problems as cyclical, fed by unusual and temporary demands. (See http://www.stanford.edu/news/release/970430shrtgside.html".) Others believe that "offshore programmers from countries like India and Russia will be a significant resource, perhaps even a threat to U.S. jobs," she said. The Stanford researchers estimate that there are 2 million software professionals in the United States and perhaps 6 million worldwide. Although the size of the demand is unknown, Tessler and Barr say that they believe "the shortage of software talent is global and that it will get much worse in coming years."
"The Indian software export industry grew to $700 million over the last seven years, but the net impact was a tiny fraction of worldwide demand for software," said Tessler. "Furthermore, we believe that it will be at least 10 years before any other nation becomes a major source of software products and services. China's economy is developing so fast, for example, that it will consume almost all of its software labor resources to develop its own government, communications, finance and industrial systems."
Some experts doubt that a crisis is really at hand. At a recent Stanford seminar to discuss software labor issues, one Silicon Valley human resource recruiter recalled that in the 1980s, computer companies filled software positions with "high school kids on skateboards." Others said employers weren't taking full advantage of community college graduates, and another recalled that when physicists were in high demand, so many science-minded young people entered physics training that surplus physicists now drive taxis.
Yet students have begun to enter college computer science programs in greater numbers. Nationally, enrollments rose 5 percent in 1995 and 40 percent in 1996, according to the Computing Research Association [http://www.cra.org]. Still, Tessler and Barr don't think the supply will grow as fast as the demand.
With physicists, Tessler said, science- and math-proficient students were "recruited for government-funded projects in defense and research that the government doesn't fund any more. The people we need in computer science will go into every kind of software-enabled industry and organization. We are short so many people that we could easily triple the number coming out of computer science programs and still have a shortage."
Barr also thinks there is a limit to the number of people who have the ability and aptitude to be software professionals as long as the field remains more of an art than a science. Studies have shown that the best programmers are 10 times more productive than the average programmers, he said.
Unlike civil engineering or medicine, Tessler said, computer programming is still too new to become systematized so that people of different skill levels perform different functions according to well established rules. "The fact is there is no real set of methodologies behind developed software no standards on what to name variables, for example," said Tessler. "It is all idiosyncratically built. That means that it is harder to maintain and debug."
There will be cyclical ups and downs in the labor market, Barr said, "but this is the beginning of a period where people are going to have more ideas about what software they need than there will be people to build it. . . . As soon as one airline figures out that a 'frequent flyer' program is just a piece of software that lets you offer your frequent customers a little bit extra, every airline has to do it. It is those creative uses of technology by businesses that will drive the demand."
Corporate strategies, public policies
Many industries will be forced to change their perception of software talent and to scale back their software projects to only the most advantageous, the researchers say. "Some companies may go out of business simply because their competitors adjusted better to the labor conditions," Barr predicted.
Programmers could be earning more than their managers within a few years, according to Tessler and already are at some companies. "Employers will need to change their incentive packages and working conditions to keep their best people."
Government agencies may have the worst time maintaining their talent because of their more rigid pay scales and work environments, she said. Firms at the high-tech end of software development will probably lure some of the best talent from other companies because they provide the working conditions software professionals like, such as the most talented colleagues and best computers.
Added Barr: "All you can do as senior management is to say 'OK, these particular projects are really important for us. We are going to spend the money we have to spend to attract, retain and retrain the talent."
For many companies, however, the first step will be auditing to find out what they are spending now on software. Such audits are rare outside of software companies themselves.
Companies also will need to learn from their mistakes. Only about one-third of the software projects undertaken are successfully completed as planned, according to one study by the Standish Group, a consulting research firm in Dennis, Mass. [http://www.standishgroup.com/chaos.html.]
In the long run, Tessler, Barr and Miller expect software programmers to become much more productive. Professional standards will eventually develop, and perhaps certification for professionals of different levels, Tessler said. Curricula will probably gradually change to make college graduates better meet industry's needs.
At a Stanford roundtable discussion of labor issues, some employers complained that colleges should train better. "There is very heavy emphasis on theory and math and a basic understanding of the operating system and the architecture of the computer" in most college programs, Tessler said. "It is really quite valuable, but there are other things that industry needs that are not normally taught." Those include what she calls software engineering students learning to develop and test software for commercial situations that is easy to maintain and that can be easily integrated with existing software.
New technologies to enhance the productivity of programmers will not have a major impact in the next decade, the Stanford researchers say, but ultimately they believe that such tools will be invented. "Universities and companies are already doing research in this area," Tessler said, "but they should accelerate their efforts and government policies should encourage it." [For more information, see http://www-scip.stanford.edu/scip/.]
By Kathleen O'Toole