FRAMINGHAM (07/03/2000) - After last month's ruling that ordered the breakup of Microsoft Corp., Bill Gates made a number of public statements at a press conference that warrant closer examination of their credibility. "Microsoft has brought widespread benefits to the economy ... making the vision of low-cost computing a reality ... [with] low prices that really arise out of the PC industry."
One of the best ways to judge this claim is to refer to Gartner Group Inc.'s "TCO Manager for Distributed Computing" methodology. It was acquired from a corporation on whose board I served, so I understand its validity. The total cost of ownership (TCO) of a desktop computer at a typical firm runs about $10,000 per year. Hardware costs account for only 8 percent. Microsoft receives only a part of the total software costs (about 3.5 percent) of TCO. What, then, accounts for the remaining TCO of the estimated 95 percent of desktops that depend on Microsoft software?
By far, the largest share of these expenses appears as administrative and operating support costs, as well as the unproductive employee time that can be attributed to the difficulties of keeping Microsoft software running. So Gates' claim that Microsoft has delivered "low cost computing" has little merit.
"[The] judge's decision represents an unwarranted intrusion into the engine of economic growth for America."
Microsoft has consistently called itself a major source of current U.S. prosperity. It has bought newspaper ads stating: "The government is spending millions of taxpayer dollars in a court case that would stifle competition and interfere with an industry that is responsible for 25 percent of the nation's economic growth." But such claims are misleading and based on statistical distortions. Microsoft, as the fourth most profitable U.S. company, has been exceptionally adept at extracting profits for itself while leaving most of the desktop TCO to its customers to absorb as additional overhead.
To judge the impact of desktop computing costs, it's worth examining how corporate profits compare with TCO. My database includes financial data for 7,794 corporations that represent this country's key economic producers of profits. They reap US$286.2 billion in pretax profits and employ nearly 21.7 million people. Using these two figures, I calculate pretax profit per employee to be $13,202 during 1998 and 1999 - years of unprecedented prosperity.
My database also tracks the spending patterns of corporations with a combined 3.6 million employees who use 1.5 million PCs - a sample that suggests that U.S. companies place computers on the desks of 42 percent of their employees.
Thus, the TCO of PCs amounts to about one-third of total pretax profits.
What does this mean? The "engine of economic growth" is diminished by any excess costs of desktop computer ownership. Whether profits would be less without PCs is irrelevant. If everyone uses a PC, then the net competitive effect is not decisive.
Another way of looking at Microsoft's dubious claim of being a beacon of economic prosperity is to compare corporate taxes with the TCO of PCs. After all, tax revenues support the viability of the U.S. as a global economic power.
Corporate tax payments for the 7,794 firms in my database average $5,607 per employee, or about 56 percent of the average TCO.
The Microsoft antitrust case has diverted attention from issues that really matter, such as Microsoft's threat to national security and its pursuit of maximizing profits at the expense of its customers' exorbitant TCO. Instead of speculating about the consequences of splitting Microsoft into two potential monopolies as a replacement for one real monopoly, paying greater attention to software security risks and to excessive ownership costs of personal computing would be of much greater benefit to everyone.
Strassmann (firstname.lastname@example.org) has found that the number of PCs in a firm explains a large share of its IT budget. Legal remedies should demonstrate whether they reduce the costs and risks of desktop computing on a Microsoft platform.