By PAUL CONSTANCE

A tourist stopping for cash in Caracas will find that automatic teller machines are not responding. A security guard in Managua will get stuck in an elevator between two floors. Airports everywhere will fall strangely silent.

It sounds like low-brow science fiction, but variations of this scenario are being contemplated seriously in corporate boardrooms and government offices around the world. In fact, most people have heard of the "Year 2000 problem" ("Y2K" for short), which stems from the fact that most computers process dates based on two-digit codes that represent Dec. 31, 1999, for example, as "12/31/99." When confronted with a "00" year code on Jan. 1, 2000, many computers will either assume the year is 1900 or stop functioning altogether.

Media reports on Y2K tend to bounce between the extremes of alarmism and skepticism. In Latin America and the Caribbean, Y2K is sometimes dismissed as a sort of technological curse that will mainly afflict the United States and a few other rich countries. The result, according to Ricardo Miranda, an IDB information systems specialist who serves as the Bank's Year 2000 project manager, is that "many people know about it, but most doubt that it is a serious problem."

Domino effect. While debates over the severity of Y2K will rage at least until December 31, 1999, there is broad consensus among information technology experts that it will, in fact, mean trouble.

Here's why. Practically all the software used by governments and corporations--including those in virtually every Latin American and Caribbean country--relies on date codes. Such codes are particularly crucial in financial and banking software that controls things such as the electronic funds transfers, records of deposits, and withdrawals, and pension payments. They also affect transportation, energy, manufacturing, health, defense, telecommunications and international trade.

Individual software systems can be "fixed" or altered so that they will correctly process dates ending in "00." But the problem is complicated exponentially by the fact that most computers exchange information with others, via networks. Even if one bank has fixed its software, for example, it might still suffer damage if it processes funds transferred by a foreign bank that has not.

So much for the realm of visible computers. The Y2K problem also affects the vast world of "embedded processors," or computer chips that are buried in the guts of all sorts of appliances and machinery. Videocassette players, building elevators, alarm systems, air traffic control terminals, cellular phones and medical diagnostic equipment--these are but a few of the kinds of machines that could also malfunction because of date-code errors.

In effect, software and computers are now threaded through so many dimensions of the world's infrastructure that no one can effectively predict how many systems are vulnerable to Y2K. Even if a small percentage of them fail on January 1, 2000, however, they could have a domino effect on related systems, causing massive economic and social disruptions. Some observers believe the cost of these disruptions and after-the-fact repairs will be high enough to cause a global recession in the year 2000. Edward Yardeni, chief economist for Deutsche Bank Securities in New York, made headlines earlier this year by claiming that there is a 70 percent chance that such a recession will occur, and that it will be "at least as severe as the 1973 74 global recession ... caused by the OPEC oil crisis." Though he is quick to admit that there is very little reliable information on the extent of the problem, Yardeni has used detailed reports on the vulnerability of U.S. government computer systems to extrapolate on how private industry and other governments might be affected. His conclusion: a majority of governments and companies will not fix their systems on time.

What to do? As simple as it may sound, the Y2K problem is fiendishly difficult to fix. Many of the software applications on the mainframe computers that run critical business and government systems are one-of-a-kind creations written in the 1960s and 1970s. The programmers who created these systems have now retired or moved on to other jobs, and few of them left detailed documentation explaining how the systems work. In order to make a system safe for the new millennium, programmers today have to check each line of computer code. Most mainframe software programs have several hundred thousand lines of code, and most corporations and government agencies use dozens of programs.

Although most organizations are trying to tackle the Y2K with their existing information technology staff, many are finding that they simply lack the resources and expertise to do the job. A small industry of Y2K consulting firms has sprouted up in the last few years to help vulnerable companies and governments get the job done, resulting in fierce competition for the limited number of qualified software programmers. Even if an organization can afford to hire these consulting firms, they are not guaranteed success. It can take months --even years--o get the job done, and the best software programmers are constantly being "hired away" by competing firms that offer higher salaries.

Perhaps the most sobering indicator of the extent of the problem is the amount that is being spent to fix it. Citicorp, the largest U.S. bank, has budgeted $600 million, according to a recent issue of The Economist. Eight other leading U.S. banks are spending between $100 million and $400 million. The U.S. government, which had originally estimated the cost of preparing all public-sector systems at less than $2 billion, now plans to spend nearly $6 billion. The Gartner Group, a U.S. consulting firm that specializes in Y2K issues, has estimated that worldwide expenditures on preventing the problem will run between $300 billion and $600 billion.

Even the IDB has seen the cost of Y2K exceed its early internal estimates. Miranda says the Bank originally calculated that it would cost around $10 million to repair all its critical corporate systems. "Then we started looking at other areas, such as departmental systems, embedded systems in the headquarters building (security, telecommunications, elevators and even air conditioning) legal, contingency planning and so forth, and we realized the task will probably cost twice as much as we expected," says Miranda.

Budget buster. Unlike other kinds of spending on information technology, expenditures on Y2K repairs cannot be justified for the extra value or efficiency they might bring to an organization: their only purpose is to maintain the status quo. As a result, governments and corporations that are already facing spending constraints have found it extremely difficult to get money for Y2K efforts.

Over the last year, many governments in Latin America and the Caribbean have launched Y2K awareness programs and have begun to assess the vulnerability of their critical systems. James Castle, a Gartner Group consultant who has met with senior political and corporate officials in Argentina, Brazil, Mexico and Venezuela, says that as in much of the world, government computer personnel in the region are being told to "go fix the problem, but don't spend any money on it because we can't afford to." Although he believes most of the large national and multinational corporations in the region are setting aside sufficient resources to deal with the problem, Castle said the public sector "has started tragically late in terms of serious remediation."

There is one significant exception. Castle says Mexico has made impressive progress over the last year, since President Ernesto Zedillo took a personal interest in the problem and appointed a National Commission for the Year 2000 Conversion that is composed of cabinet ministers, key legislators and influential representatives of the corporate and academic communities. The commission has required Mexico's 263 government agencies and its 14 major banks to adopt a rigorous schedule for converting and testing all critical systems, and it has hired independent auditors to make quarterly progress checks at each institution. The October edition of Year 2000 Outlook, a newsletter published by the Information Technology Association of America, reports that 80 of those 263 agencies have already completed the conversion and tested their systems, and that the remainder will have done so by December 1998.

The Gartner Group's Castle cautioned that although several other Latin and Caribbean governments have established similar Y2K commissions, few have either the authority or the financial means to make a significant difference in the 14 months that remain before the date change. "Most of the region's countries have to assume that they won't get it done on time," he says. "So instead they are coming up with contingency plans for the critical systems at each government agency." Such plans usually involve devising "manual work-arounds," or procedures for providing vital services in the event a computer system fails.

In the meantime, the IDB's Board of Executive Directors is considering an emergency line of credit that will make loans for Y2K preparations and repairs available on an expedited basis to its borrowing member countries. The Bank also has been working with each of its project executing agencies to increase awareness and planning. "It is going to be a bigger problem than people realize, and it is going to cost more than they think it will," says Miranda. "The goal is to prepare for the worst and hope for the best."

For more information, visit www.iadb.org. Detailed resources in Spanish are available at www.Y2K.gob.mx.