After a lot more time and money than expected, and a rocky launch greeted with months of customer complaints, CalPERS has succeeded where two other state agencies recently failed. Its big new computer system is up and running.
This week CalPERS plans to ask for bids from consultants for a review of a project that began in 1995, had an estimated cost of $279 million that more than doubled to $586 million, and finally enabled the unplugging of all old computer systems last June.
The review will look at how well the project achieved its goals, the benefits and any lessons learned. Another big pension system, CalSTRS, aiming to learn from history, plans to ask for bids this fall on a major computer system to be completed in six years.
The new MyCalPERS system, which began operating two years ago this month, once seemed in danger of adding another chapter to a long and costly history of state computer systems abandoned in mid-construction as unworkable.
The CalPERS board president, Rob Feckner, gave the main contractor, Accenture, an angry public tongue lashing in December 2009 after a long delay was belatedly reported. He also complained to the firm’s top executive and board chairman.
A “recovery plan” was launched for what was already called the “Pension System Resumption” project. The original attempt to modernize and consolidate old computer systems, begun in 1995, failed to perform some major functions.
When the new system launched on Sept. 19, 2011, call wait times ballooned. So did backlogs in death benefit processing, retirement allowance adjustments, service credit purchases and retirement estimates. Most problems were solved or eased.
Last October, CalPERS announced the new system received an award from Government Computer News, and a computer team working with 3,125 local government agencies got an award from the National Association of State Chief Information officers.
A CalPERS news release said the project streamlined “109 separate and outdated IT systems that were supporting 3 billion records . . . retiring 200 interfaces, 90 database and 109 silos, utilizing 550 staff working on 50 sub-projects.”
Among other system facts added by a California Public Employees Retirement System spokesman last week: “more than 2.3 million lines of code” and “correction of over 3 million employer and partner historical data anomalies to ensure valid data.”
CalPERS board member J.J. Jelincic, a project skeptic, said being four years late with a quarter-billion dollar overrun, which is probably much higher if all costs are counted, is not a “shining” success.
“The system works,” Jelincic said last week. “But the day it rolled out we learned that it was only doing about 60 percent of the calculations it was supposed to. The board did not know that at the time. If we had known that, I think we would have delayed it.”
As chairman of the risk committee, Jelincic said he has unsucessfully tried to get a report on what went right and wrong. He said the independent reviewer should talk to former CalPERS employees who worked on the system and are likely to be candid.
At a committee meeting last month, CalPERS board member Bill Slaton said he has seen successes and “spectacular” failures during three decades of working on the financing of state and local government information technology projects.
“I think we fall toward the successful category,” he said of MyCalPERS.
Slaton drew agreement from two other board members when he told staff, which estimates the independent review will cost $800,000 to $1.2 million, to spend more if needed to get a complete review that may aid other projects in the future.
He said the review will “not just have us learn, but the state learns, and I think that’s money well spent.”
The CalPERS staff plans to go beyond the standard Post-Implementation Evaluation Report required by the state Department of Technology, doing a “PIER-Plus” review.
MyCalPERS cost more than doubled from original estimate
In 1973, three years before Steve Jobs and Steve Wozniak released the first Apple computer, the nonpartisan Legislative Analyst’s Office issued a report on failed state computer systems.
“Many of these efforts have been characterized by cost and schedule overruns, user dissatisfaction and operational problems which have resulted in either major modification or abandonment,” the report said 40 years ago.
Two recent costly examples of troubled state computer projects halted before completion: a half-billion dollar court case management system last year, and a $373 million state payroll system this year.
A staff report to the CalSTRS board this week said steps to reduce risk in the development of a new pension computer system have “only increased with the numerous highly publicized project failures within state and local governments.”
The California State Teachers Retirement System has hired a consultant, L.R. Wechsler, with international experience in developing 59 public pension systems. An independent oversight consultant, Grant Thornton, will monitor technological progress.
In addition, CalSTRS staff has reviewed “lessons learned” from a variety of large pension computer projects.
“The most common themes of communication, staffing, planning and data cleansing have been incorporated into the project plan,” said the staff report, and the goal is “to help ensure that CalSTRS does not repeat the mistakes of other large projects and reproduces those activities that were key to project success.”
As a result of the research, the staff expects to request additional staff to reduce risk during the development of the new pension computer system: the equivalent of 75 full-time employees between fiscal 2014-15 and 2019-20 at a cost of $42.5 million.
“It needs to be emphasized that these estimated expenditures should not be viewed as project overhead costs but rather an investment in enterprise risk mitigation,” said the staff report.
CalSTRS also is getting first-hand experience in the problem of developing a new computer system in this era. A smaller accounting system that began operating in January is expected to cost $20 million.
The launch of the accounting system was delayed six months, and there have been backlogs and other startup problems. Another report to the CalSTRS board this week said staff needs more training in the new accounting system.
In a lessons-learned study of the accounting system, CalSTRS and the contractor, Ciber, interviewed individuals and 13 focus groups. The results are still being analyzed, but the report said some themes are likely to emerge.
“What could be improved” includes change management, training and more realistic resource planning and scheduling. “What went well” includes commitment, expertise and collaboration.