If the first of several tests had been a real payday, families of some state employees on military duty would not have received the right salary supplement from the state, according to a Tallahassee Democrat report of the trial run’s 37% error rate. In other cases, a state Department of Financial Services (DFS) official said, Convergys would have withheld federal income taxes on amounts allocated by employees for pre-tax benefits, such as child care and medical savings accounts.
“Of the approximate 7,100 employees we tested using People First input, 2,655 errors were identified,” Doug Darling, director of auditing and accounting for DFS, wrote in a report. “Convergys has addressed many of the errors, but solutions for some are not available at this time.”
The biweekly payroll sampling was chosen to reflect different agencies and parts of the state. That’s because shift differentials, pre-tax benefits, deferred compensation and management classifications mean many variables in the state’s system.
Darling said the department won’t run a second mock payroll involving about 6,800 monthly paychecks until it is satisfied that corrections made by Convergys after the January test of biweekly payments will work.
The Convergys Florida operations chief – which has the seven-year, $278 million contract for outsourcing the state’s personnel systems – said the miscues identified in the dry run have been fixed. Chris Emerick, the Jacksonville-based vice president of Convergys Employee Care, said the company is ready to simulate a monthly payroll sampling.
And a spokesman for the state Department of Management Services, which chose Convergys two years ago and is shedding the state’s 30-year-old computer system once the privatized system is up and running, told the newspaper that the agency is confident Convergys will work the bugs out of the personnel project.
Emerick said the simulated payroll drills will provide a comfort level “above and beyond the testing required by the contract” before Convergys takes over personnel systems.
“It was expected this parallel process would identify various differences due to the number of manual processes that have been automated,” he said. “It was also expected that the additional edits built within the People First system would highlight data issues. All differences identified to date have been reconciled.”