Is the Price Right?

No, you're right, the worldwide year 2000 mess really won't cost $600 billion to fix.

That gargantuan estimate works out to roughly $2,000 per man, woman and child living in the United States. At that price, the nation would go bankrupt -- dragging down the infotech industry and a fast-breeding horde of year 2000 consultants with it.

Think about it; $600 billion is almost 10 percent of the nation's gross national product. It is enough cash to buy 300 billion Big Macs, priced at $1.98. If piled atop one another, the Big Macs would reach 30 million kilometers into space.

The $600 billion estimate was prepared by the Gartner Group, a market analysis firm based in Stamford, Conn. Gartner's estimate was built on the assumption that companies and governments worldwide will hire programmers, consultants, analysts, cooks and bottle washers to check and rewrite every line of code they own. At a price of $1.10 per line of the 575 billion lines of code in service, the estimate quickly reached $600 billion. Industry's share of the bill could reach $275 billion.

But "that won't come true," said Charles Billingsley, an analyst with the Tysons Corner, Va., office of Input, a market analysis firm based in Mountain View, Calif. For example, roughly 80 percent of date routines built into software need not be changed because they aren't used by the software when performing mathematical calculations, he said. If that's true, the year 2000 bill immediately drops to $120 billion, which is only enough to buy a pile of hamburgers 6 million kilometers high.

But that's too optimistic, said Bruce Hall, a year 2000 research director at Gartner. Companies and governments will need to spend a lot of money to check their software just to separate the date routines that do and do not need to be fixed, he said.

Moreover, the 80 percent figure is not justified, he said. One of Gartner's client companies checked approximately 55 million lines of its software code and discovered roughly 1 million date routines. Of those, 75 percent may need to be fixed, he said.

There are many ways to reduce the potential bill, said industry officials. For example:

• Segregate software and data with year 2000 problems from other computer networks.

• Use newly developed software tools to automatically find and repair faulty date routines.

• Encapsulate faulty date routines with a standard software shell that includes a four-digit data routine.

• Change program logic to compensate for incorrect two-digit dates, an option that may cut repair costs by 35 percent to 60 percent, according to Louis Marcoccia, the data administration director of New York City's transportation division.

• Replace faulty mainframe software with new, off-the-shelf, client/server software packages developed for businesses by companies such as SAP America Inc., Wayne, Pa.; Baan Company USA Inc., Menlo Park, Calif.; or Oracle Corp., Redwood Shores, Calif.

• Persuade the government-backed Financial Accounting Standards Board to permit companies to amortize year 2000 repair costs over several years, instead of one year as is currently required.

A less breathtaking estimate is Input's $56 billion prediction, which assumes that the top 1,000 corporations will incur 50 percent of the year 2000 cost, said Billingsley. Their cost will amount to $28 billion -- $20 billion in programming changes and $8 billion to implement the changes, yielding a total year 2000 bill of $56 billion, he said.

But even this $56 billion bill will be reduced by companies' purchase of alternative client/server technology, he said. Moreover, the bill will be spread through a period of eight to 10 years, cutting the annual repair cost to several billion dollars, he said.

And with the software industry taking in almost $90 billion in 1996, the year 2000 problem will be a nice boost for the software business, rather than an apocalyptic crash of computer code.

But Hall argued that the cost cannot be stretched past 2000. Instead, the vast majority of repair dollars must be spent by January 1, 2000, he said.

Other beneficiaries of the year 2000 software problem include the media, which is earning readers and advertising revenue from the worries, as well as companies specializing in fixes, said Billingsley. "To a certain extent, [the year 2000 issue] is a sky-is-falling scare."

Regardless of the final tally, the year 2000 problem will cost companies many billions of dollars to fix, Hall and Billingsley warned.

The California state government had problems issuing driver's licenses scheduled to expire after 2000, and the Social Security Administration has already spent at least 300 man-years since 1989 to fix its software, Billingsley said. One insurance company in the United States has already earmarked $100 million to fix its year 2000 problem, said Hall.

"Many, many organizations have not even pulled the cover back to look at [their year 2000] problem. That's the scariest part," said Hall.