Terminated. It's a word that's rarely used in the world of government information technology, where projects can take on a life of their own and live forever, even if they are over budget, late on arrival or fail to operate as originally promised. But on Nov. 20, 1997, John Thomas Flynn, California's chief information officer, exercised his oversight authority and, for the first time in the state's history, terminated an IT project. The project was the Statewide Automated Child Support System (SACSS), the largest and most expensive state-run, single-unit information system in our nation's history.
Officially, the state and Lockheed Martin IMS, the project's chief contractor, mutually agreed to cancel the project after failing to reach terms on how to fix SACSS, which had been in trouble since its inception. A report issued by an independent vendor found some 1,400 problems plaguing SACSS. Less than a dozen counties out of 58 use the system after more than five years of work and expenditures of $100 million. Final costs could be as high as $345 million.
Behind the official explanation of contract conflicts with the vendor lie a number of troubling reasons contributing to SACSS' termination. They include a flawed federal program for automating child support throughout the 50 states; an antiquated procurement system; tensions between California's politically powerful counties and the state; lax project management and questionable oversight.
For California, the immediate concerns are getting SACSS on track again; agreeing with the federal government on child support deadlines; and correcting the flaws in the procurement and project management of large computer systems. More troubling, however, is the spiraling cost of, and diminishing return on, large-scale government computer projects in California and throughout the country.
In 1990, the number of child support cases in which collections were made was at a measly 20 percent. Today, after the federal and state government spent nearly $2.7 billion on automation, the collection rate remains at 20 percent, according to the General Accounting Office. Just how much longer Americans will accept such a poor return on their tax dollars for government technology projects remains to be seen.
SACSS began in 1992, when California's Dept. of Social Services (DSS) announced that the federal government had approved a $75.5 million contract for automation of the state's child support enforcement program. The contract was awarded to Lockheed Martin. At the time, DSS projected that the new system would eventually add as much as $11 million each month to the state's child support program, "more than paying for itself while collecting substantially more child support for California's children," said John Healy, interim director of DSS.
The decision to automate child support came from a federal mandate in 1988, which required that every state develop a single child support system. The feds agreed to pay an enhanced rate of 90 percent of the development costs if the project was completed by September 1995.
Most state officials point to the federal mandate, coupled with what they consider an unrealistic deadline, as a chief culprit in California's failure to build SACSS and why other states are also struggling with their automation efforts. "States all over the country are having problems with their child support systems because of federal laws imposing deadlines and penalties along with strict reporting requirements and unreasonable system transfers," said Flynn. "These prerequisites have tied the hands of states."
State Assemblywoman Elaine White Alquist, who chairs the California Assembly Information Technology Committee, also singled out federal requirements as a reason why SACSS failed. "The federal government has had an unrealistic mandate for this project," she said. "The additional requirement of having to use a system transferred from another state was also unrealistic. As a result, California didn't have any flexibility."
California's procurement system has also been repeatedly attacked as a villain in the SACSS case. Like most other states, California in the early 1990s procured complex computer systems the same way they purchased asphalt for roads -- based on low bids and rigid specifications. The results have been disastrous. Many of the project's cost overruns can be traced back to the flawed procurement process as well as the formation of the original contract between Lockheed and the state, according to a report issued in April 1997 by Assemblywoman Alquist's committee.
The report, Will SACSS Ever Really Work?, pointed out that Lockheed was in the enviable position of enforcing a contract with few protections for the state. As a result, California "was forced to assume increased project costs caused by forgotten tasks and schedule slippage." By 1995, estimates for SACSS had soared to $260 million, of which over $100 million was to be paid directly to Lockheed.
Unable to Perform
When the Health and Welfare Data Center (HWDC), the state agency charged with overseeing SACSS, attempted to install the system in Fresno County -- the largest pilot county testing the system -- they encountered significant problems. An assessment by the state found the application was unable to perform daily business operations, including a forms generation system that didn't work at all. Fresno wasn't the only county to run into problems with SACSS. Other counties stopped using the system, complaining that data popped up in the wrong place; parents could not be tracked down; and that the software, which uses more than 350 screens to manage child support, was simply too hard to operate.
According to Bill Otterbeck, a project manager with the Alameda County District Attorney's Office, the reason no more than a handful of counties use SACSS right now has to do with credibility. "We have to trust the data, and there was a real question whether that was possible with SACSS," he said. Otterbeck pointed out that with an unreliable system, counties might fail to disburse funds to a custodial parent or credit a noncustodial parent for making a disbursement.
In January 1997, HWDC hired Logicon, an independent firm, to verify and validate SACSS. One month later, Logicon filed a report listing 1,400 problems plaguing SACSS, many of which related to shortcomings with project management. Logicon concluded by expressing concern that the problems could ever be solved. One reason for the concern was Lockheed's shuffling of its staff members amongst the numerous state child support projects it contracted for, reducing the number of experienced staff working on SACSS and delaying the rollout of the system.
Another troubling aspect in the demise of SACSS has been the power struggle between the counties and the state over the direction of the system's development. In California, the district attorneys of the 58 counties administer and operate child support through their family support offices. Originally, SACSS was intended to link all 58 counties and connect them with the database systems of state agencies -- including motor vehicle, employment and tax revenue databases -- to increase the chances of locating absent parents and their assets.
California's counties have traditionally sought a more decentralized child support system; one that allows counties of different sizes to create systems that meet their specific needs. From the state perspective, this kind of multijurisdictional system, versus a single statewide model, is difficult to implement. California already has struggled with the counties over the direction of its Statewide Automated Welfare System (SAWS), another statewide project that has had a troubled past.
In fact, Lockheed stated that part of the problem with implementing SACSS had been that counties repeatedly asked for changes in the system. Leora Gershenzon, an attorney with the National Center for Youth Law, an organization that has kept close tabs on SACSS, blames the system's failure directly on the state/county power struggle for control of SACSS. "What killed the system the most was that the state allowed politically powerful counties to customize it to the point where it collapsed," she claimed. "It was the biggest nail in the coffin."
Assemblywoman Alquist doesn't believe counties exert too much power on IT projects; but other critics of SACSS have pointed out that the majority of the states with the most trouble automating their child support systems have politically strong county governments.
What does trouble Alquist has been the oversight, or the lack of it, on the project. "It was obvious to us early on that the project was not working well, yet it took DSS a while to admit something was wrong," she pointed out. "In February 1997, both the vendor and the state gave no indication of how bad the situation had become. Even as recently as one month ago [October 1997], at an oversight hearing, we weren't told how bad the situation was."
Alquist's committee report criticized Flynn's office, the Dept. of Information Technology (DOIT), for its slow reaction to the growing crisis over SACSS. "How bad does a particular project need to become before the state's information technology czar considers termination?" asked the report when it was published in April 1997.
When Flynn pulled the plug on SACSS in November, Alquist and others praised the CIO for finally taking decisive action. Flynn explained that the decision to terminate the SACSS contract took so long because there were so many issues at stake. In particular, he pointed out that "one-third of the counties had the system up and running. We had to take that situation into consideration," said Flynn. "Termination was the last resort."
The Next Step
With California and Lockheed unable to come to financial terms on how to save SACSS, the state has had approximately 90 days in which to come up with an alternative plan for federal approval and financing. One option is to adopt a system from one of the counties where child support has been automated.
The dilemma for California is to come up with a design that meets the vastly different needs of the counties without automation and still have a viable, single system that meets federal requirements. Unlike SAWS, which currently involves several consortia of counties building different systems, child support automation must be centralized, according to Michael Kharfen, a spokesman for U.S. Dept. of Health and Human Services. "We believe in the principle of a single statewide system because of the requirement for a national directory [containing names of noncustodial parents]," he said. Without single systems in each state to make data sharing more manageable, building a national directory will be extremely difficult, according to Kharfen.
Alquist hopes that DOIT and HWDC get the ball rolling again. She said the state must move quickly on a plan that addresses functionality, time and cost. In addition, the Legislative Analyst's Office urged the state to take steps that ensure a new contract is based on performance; breaks the outcome down to smaller, more manageable projects; requires the vendor to share in the financial risk of the project; allows payment to the vendor only upon acceptance of deliverables; and requires that contract and project management be carried out by qualified, experienced personnel.
In Alameda County, one of the few counties with a working child support system, automation has vastly improved how the county manages its 60,000 cases. "Automation is absolutely core to child support," said Otterbeck. "It has meant the difference between an excellent program and one that is barely serviceable."
For other counties that have waited for SACSS, the situation doesn't look good. Welfare reform has added more reporting requirements to the beleaguered child support offices. Counties (under state oversight) must meet these requirements to be eligible for Temporary Assistance for Needy Families (TANF) block-grant funding. With one estimate for completing a new system at three years, counties could be in for a long wait. Federal penalties could make the situation worse.
According to Kharfen, HHS doesn't want to take money away from states, "but penalties are the only control we have. We want good systems, but we want them done on time," he added.
When the federal deadline of Oct. 1, 1997 for state automation came and went, only 17 states were certified as meeting all the requirements of the federal mandate for developing a statewide child support enforcement system. To state officials, the numbers were further proof of how unrealistic federal deadlines and requirements have become.
The General Accounting Office (GAO) has also criticized HHS for what it calls "limited leadership" that has inhibited states' progress with child support automation. In particular, GAO cited HHS for its failure to: require states to follow a structured, disciplined approach to systems development; to assess whether a sufficient number of systems were available for transfer to other states; and for putting states in the position of having to present inaccurate and sometimes impossible schedules showing that they would meet federal deadlines.
What remains unclear is the public reaction to ongoing expenditures of taxpayer funds for information systems that fail to work as originally intended or not at all. While the complexities of computer technology may be difficult to grasp for many citizens, they do understand the value of money. If systems such as SACSS continue to stumble, then citizens and Congress may begin to question why so much is spent with so little to show for it.
February Table of Contents