Citicorp is one of America's two largest banks. Chase Manhattan Bank is the other. Citicorp has a staggering total of 400 million lines of code. The computer division estimates that between 5% and 10% of this code must be repaired. That is as much or more code to repair than Social Security has code to check. It took Social Security's 400 in- house programmers from 1991 to mid-1996 to go through 6 million lines and correct the bad code. That left 24 million. But Citicorp only began its correction project in late 1995. That was when its president was informed about the problem. (Investor's Business Daily, Feb. 12, 1997.)
Chase Manhattan has 200 million lines of code to go through, in 1500 applications on 60,000 desktops, 400 midrange computers, and scores of mainframes. Like Citicorp, Chase began its Enterprise Year 2000 project in 1995.
So far, there are no available reports on the Japanese banks, which dwarf Citicorp and Chase.
There is one brief report on the Federal Reserve System, which appeared in the Federal Reserve Bank of Atlanta's magazine, FINANCIAL UPDATE (Oct.-Dec 1996, p. 3). A task force has been formed to study the problem. As of late 1996, the FED hoped to develop a project plan. No actual repairs yet, let alone testing. I am not calmed by the report.
This is from SOFTWARE MAGAZINE ON-LINE (March, 1997).
* * * * * * * *
Of all the industries affected by the Year 2000 problem, none has a more urgent need to fix its date fields than banking and financial services. A two-digit date field in a core lending or savings application is a time bomb that has already caused a few unwary banks and credit unions to calculate inaccurate account balances.
Most have been dealing with the Year 2000 on a daily basis for years, executing manual "workarounds" whenever they issued a long-term mortgage or dealt in bonds whose maturities carried into the next century. As the millennium nears, however, more and more banking applications run the risk of malfunctioning.
Moreover, manual workarounds are impossible in transaction systems where the daily dollar volume can mount into the hundreds of billions of dollars. Indeed, Chase Manhattan Bank in New York City -- the nation's largest bank with more than $336 billion in assets -- launched its formal Enterprise Year 2000 effort more than two years ago precisely because bank officials realized that a technological fix was the only way to ensure that its trillion-dollar-a-day transaction processing systems would continue to function without error. . . .
Sheinheit and his deputy, Brian Robbins, vice president, enterprise architecture, standards and Year 2000 program, have direct and overall responsibility for guiding Chase and its 68,000 employees through their enterprise-wide conversion process. It's a daunting task. Sheinheit, Robbins and their colleagues must come to grips with more than 200 million lines of code running in 1,500 applications on more than 60,000 desktops, 400 midrange computers and scores of mainframes. The bank runs a laundry list of mission-critical applications, from check processing to commercial lending and deposit systems, consumer credit systems, insurance, payroll, cash management, securities lending and a host of trading systems. According to Sheinheit, Chase -- which invests $1.8 billion a year in information technology -- expects to spend from $200 million to $250 million over a three-year period on its Year 2000 work. "That's 4% to 5% of our overall spending per year or 12% to 15% of our applications, development and maintenance costs," he says.
Managers at Chase were aware of Year 2000 issues long before the bank launched its Enterprise Year 2000 Program in 1995. "A number of systems were updated already for the Year 2000 because they were already booking past the millennium," says Robbins. "The loan system and the trading systems dealing in long-term bonds all had to deal with this issue years ago." About two years ago the industry hype around the Year 2000 convinced a group of senior managers in the bank that such ad hoc solutions would no longer carry the day. . . .
Sheinheit is unclear on how many additional staff and consultants he will hire to complete the date-field work, but he estimates he will add several hundred people -- most of them during this year and next. The plan calls for many mission-critical applications to be converted this year, with the rest finished in 1998 -- leaving an entire year for testing.