Welfare AFIS

After much delay, state and local governments are beginning to use finger-imaging technology to curb welfare fraud and save dollars.

by / March 31, 1996 0
On June 3, 1991, Los Angeles County used finger-imaging technology for the first time to curb fraud in its General Relief welfare program. Within six months, the county stated that it had saved $5.4 million, thanks to the technology.

Just about everybody in state and local government sat up and took notice. Yet today, almost five years later, only a handful of counties are running finger-imaging programs for welfare applicants and just one state is finger-imaging applicants on a statewide basis.

One vendor called welfare finger-imaging a "no-brainer," adding that the systems pay for themselves in less than a year. Yet the vendor, like others, has been clearly frustrated by the slow growth of civil finger-imaging in the government sector.

For those who track technology in the public sector, the situation with finger-imaging is all too familiar. A new technology comes along and quickly delivers the kind of benefits that businesses in the private sector would jump at. Yet, for a number of reasons, government is slow to accept the technology. "It reminds me of the slow growth for electronic benefits transfer," remarked Rick Ferreira, government affairs representative for human services policy at EDS.

Electronic benefits transfer took so long to grow because of the inordinate amount of time states needed to develop the necessary RFP process and to leap some federal regulatory hurdles. The same thing is happening with finger-imaging, according to Ferreira, but with the added problem of states having to overcome the social concerns -- some might call it timidity -- about fingerprinting welfare applicants.

Whatever the reasons, state and local governments now willing to consider finger-imaging for welfare programs will find that the technology is more robust and less expensive than when it was first introduced five years ago. And the quick paybacks remain.

Los Angeles County actually began manually fingerprinting welfare applicants in 1986, when the county's Department of Public Social Services (DPSS), the second largest county-operated welfare department in the U.S., was required to provide housing benefits to homeless applicants without identification. Fearing abuse, the DPSS began manually fingerprinting and photographing applicants as a way to deter fraud in its General Relief program. Within three years, the county had a stack of 50,000 fingerprint cards and little in the way of results. Realizing that the existing system wasn't working, the county began to investigate automated fingerprint identification systems currently in use by law enforcement agencies.

The question of fraud -- or, more precisely, how much fraud -- has always been a contentious one in the welfare arena. Nationally, welfare experts believe that fraud -- and waste -- could be costing the American taxpayer more than $40 billion per year. But few states or counties can point to any actual figures on fraud when asked why they need finger-imaging.

"It was just the normal suspicion that people were trying to cheat welfare," said Bob Modell, San Diego County's data processing manager, when explaining why his government is investing in finger-imaging for welfare.

In Massachusetts, Gov. Weld has proposed finger-imaging the Commonwealth's 92,000 families who receive Aid to Families with Dependent Children (AFDC). Weld administration officials have no estimates on how many cases of fraud or double-dipping exist to justify investing in a system that will cost taxpayers $3 million per year. However, they counter by saying that finger-imaging discourages cheating and fraud from the outset. And even if the state's welfare rolls are reduced by just 25 to 50 cases per month, the system will still pay for itself.

While states can proceed with finger-imaging AFDC recipients statewide, counties must receive a federal waiver if they plan to finger-image the welfare recipients on their own, according to Ferreira. Currently, only Los Angeles County has such a waiver. Food Stamps is less of a problem, however. Counties as well as states are allowed to fingerprint Food Stamp recipients.

When it comes to finger-imaging, two facts stand out: welfare officials are not fingerprint experts and computers are never 100 percent accurate. These two points set the stage for how finger-imaging systems have to be developed in welfare agencies. First, the system must be capable of matching prints without the aid of a fingerprint expert, because unlike the law enforcement field, there aren't any in welfare agencies. As a result, finger-imaging software must automatically process the matching for the welfare agency.

Most welfare finger-imaging systems require an applicant to place one or both index fingers on an electronic scanner, which captures an image of the fingerprint, and feeds it into a computer that attempts to match the prints against a database of the entire pool of welfare cases. But no system is foolproof, according to Steve Yeich, director of business development at Printrak International Inc. "Because of the mathematics of fingerprint matching, it's impossible to build a system that's 100 percent accurate," he said.

To resolve this problem, welfare finger-imaging systems have to capture what Yeich calls non-minutia data -- information about the applicant that goes beyond the details of their fingerprint. Non-minutia data can be demographic information about the individual or a photograph. Finger-imaging systems also have to be fast. Most agencies look for a response time that's within five minutes.

Printrak provides the matching engine hardware and software used by EDS in Los Angeles County. According to Yeich, Printrak's technology is designed to handle massive databases, ranging from 5 million to 40 million prints in size.

Los Angeles County's AFIRM (Automated Fingerprint Image Reporting and Match) system incorporates two leading-edge technologies. One is the client/server architecture that gives the system its flexibility to expand according to need and budget. AFIRM also uses symmetric multi-processing computers that use several microchips simultaneously to process the fingerprint database searches as quickly as possible. While AFIRM is the oldest welfare finger-imaging system in the country, its core technology remains up-to-date.

As in several other California counties and in New York state, AFIRM runs on the UNIX operating system, has a server with the database and matching software at a central location, and communicates with the district offices, which are equipped with a workstation or PC, a scanner and -- in some cases -- a digital camera.

In Los Angeles County, communications are done over 56Kb phone lines. San Diego County, which has been pilot testing a system from Unisys Corp., uses T1 lines to link seven county offices with the central server.

Figuring the costs for finger-imaging technology is not easy. The few programs in place all vary. In Los Angeles County, EDS, which owns the central site computers, charges the county a flat rate per month for its services. The county has a lease-to-purchase arrangement for all the remote site equipment.

The county's contract with EDS for the General Relief program started in June 1991, runs to December 1998 and will cost $17 million. The county is paying EDS another $23 million (which includes $17.5 million in system, support and labor costs) for finger-imaging its AFDC applicants. That contract runs from April 1994 to September 1996.

In New York state, the vendor, North American Morpho, is charging the state $3 for every search and also is leasing the computers to the state. The system is expected to serve an estimated 1.5 million welfare applicants when fully deployed.

San Diego County acted as a test site for the beta version of Unisys' finger-imaging system and, in return, had use of the system free of charge for one year. When the agreement ends this month, the county plans to purchase its own finger-imaging system. According to Modell, the county believes it can acquire a system that serves the county's entire general relief program for approximately $400,000 to $500,000.

Clearly, the reason counties and now states are willing to invest millions of dollars in finger-imaging systems is the tremendous cost savings they can generate when welfare recipients start dropping from the rolls once they hear about the new procedure. What's less clear is why these people stop collecting benefits. Were they actually engaged in double-dipping? Or were they driven from the program by the fear and stigma of fingerprinting?

The most comprehensive figures on savings come from Los Angeles County, where finger-imaging has been used the longest. Through the end of June 1995, AFIRM has generated gross savings of $17 million for the county's General Relief program. Much of the savings, $5.4 million, was achieved in the first six months of operation when "3,021 cases were terminated and another 242 cases failed to comply with AFIRM's requirements," according to DPSS.

When DPSS began finger-imaging AFDC applicants in 1994, over 30,000 people failed to show up for their fingerprint appointments. When these people were scheduled for an interim interview, 8,030 failed to attend, resulting in the termination of their cases. These results became part of a study on the cost-benefits of finger-imaging conducted by the accounting firm Ernst & Young. Based on the study's calculations, the county expects to achieve a net savings of $66 million over a three-year period from the denial and termination of AFDC cases.

New York City, which began finger-imaging General Relief applicants in the summer of 1995, expects to see enrollment drop by as much as 50 percent with the help of the technology. New York state might save as much as $26 million with its new system.

As Los Angeles County has learned, and other governments are now discovering, finger-imaging doesn't catch fraud, it deters the crime from happening. As one vendor said, "the last thing you want to see as a result of these systems is the government prosecuting some poor family in court because of welfare fraud."

New York City's finger-imaging system uncovered 43 cases of double dipping out of 148,502 recipients. In Los Angeles County, more than 350,000 people who receive AFDC have been finger-imaged by AFIRM, but only a total of 104 suspected fraud matches have been identified by the process.

When finger-imaging was first proposed in New York, the proposal was bitterly contested by the two sides of the state Legislature before the green light was given to use the technology. Virtually everywhere that state or local governments have proposed using finger-imaging, the technology has been opposed by civil liberty and welfare advocacy groups, who claim the programs both stigmatize the poor and treat them as the scapegoats for a system that suffers more from waste than from fraud.

Los Angeles County discovered that uncertainty, possibly stigma and fear of fingerprinting can unnecessarily drive welfare recipients away from benefits to which they are entitled. When Ernst & Young evaluated finger-imaging for AFDC, their original cost savings projections of nearly $116 million had to be revised downward nearly 26 percent because clients, who first refused to be fingerprinted and were dropped from the AFDC rolls, later returned to the program. However, the Ernst & Young study also monitored client reaction to finger-imaging. According to Dawn Wilder, EDS project manager for AFIRM, the survey results showed overwhelming support from the clients for the system. "They thought it would be successful in preventing fraud," said Wilder, who added that in the five years the county has been finger-imaging no group has filed any lawsuits against the process.

According to Modell, the fact that resistance to the technology has been so mild in Los Angeles County and the five other California counties that use finger-imaging has made it that much easier for San Diego County to go ahead with its project. Another reason may be the fact that the counties have been stalwart in refusing to share the finger-images with law enforcement agencies.

Whatever the case, states are seeing fewer battles in their legislatures over the issue of finger-imaging and are readying plans for their own systems. Besides the state of New York, Massachusetts, New Jersey, Pennsylvania, Texas, Connecticut and California are in various stages of research, plans and implementation for statewide finger-imaging.

Another issue that might spark more interest in finger-imaging and speed government plans for the technology is welfare reform. Should Congress go through with its plans to convert the nation's support for the poor into a block grant program, then people like Jim Vaden, national sales director for NEC Technologies Inc., expect more activity. "There's no doubt that welfare block grants will have a big impact on finger-imaging," he said.

Meanwhile, in places like Los Angeles County and the state of California, government officials are looking at ways to expand the use of the technology to include other benefits, such as Medicaid, to perform other functions, such as tracking people who have time limits on their benefits, and to expand their use into interstate tracking systems to deter double-dipping across state lines.


When the federal government agreed to allow Los Angeles County to become a demonstration site for finger-imaging AFDC applicants, they required that the county have an unbiased third party study the costs and benefits of the Automated Fingerprint Image Reporting and Match (AFIRM ) system.

The county's Department of Public Social Services hired the accounting firm of Ernst & Young, which used the federal government's methodology to calculate the amount of possible fraud and the potential benefits. The methodologies, known as "random assignment" and "experimental design" involved a control group of several thousand applicants who were never notified nor asked to be finger-imaged and an experimental group of applicants who were notified and subjected to finger-imaging. With finger-imaging as the only factor that differed between the two groups, Ernst & Young came up with both a fraud rate and savings rate. According to Dawn Wilder, EDS project manager for AFIRM, Ernst & Young originally calculated a fraud rate of nearly 4 percent. That number has now shrunk to 2 percent because of client recidivism. Los Angeles County uses an even more conservative figure of 1.5 percent.

Despite these low numbers, the study projects gross savings of $86 million from finger-imaging, with net savings to be well in excess of $66 million.