Critics See Flaws in O.C. Welfare Fraud Findings

Last November, the Orange County district attorney’s office made a startling announcement: A six-month study of a sample of 216 AFDC cases found that recipients had committed fraud in a whopping 62% of the cases.

The finding struck a nerve in a state already wary of welfare cheats. Since it was released, lawmakers from Los Angeles to Sacramento have pointed to Orange County’s fraud rate repeatedly in pushing for expensive fraud detection programs, and Gov. Wilson cited the 62% figure in advancing his own plans for welfare reform.

But now, there is growing criticism over just how valid the welfare study may be.

To date, prosecutors have filed criminal charges in only six of the cases in the study for alleged overpayments totaling $33,529--a fraction of the $168,000 spent by the California Department of Social Services to conduct the study.

Critics maintain that the tiny number of prosecutions only shows that investigators grossly exaggerated the fraud rate in the first place, failing to take into account that people sometimes make inadvertent mistakes when dealing with a complex bureaucracy.

“For them to throw that 62% number out the way they have is intellectually dishonest,” insisted Casey McKeever, an attorney who specializes in welfare reform issues in Sacramento.

Indeed, more than a quarter of the 135 welfare recipients whom Orange County officials initially accused of welfare fraud back in November are already back on assistance today.

Statewide, officials in Sacramento estimate that the fraud rate for all AFDC cases is only about 4%.

Officials with the district attorney and the Orange County Social Services Agency, which collaborated on the project, defend their handling of the study.

They blame the small numbers of actual prosecutions on the difficulty in obtaining the evidence needed to win convictions in welfare-fraud cases, combined with a backlog in the courts.

But they assert that their “aggressive” probe was responsible for another 89 welfare cheats in the study agreeing to voluntarily withdraw from the rolls rather than face criminal prosecution, thus saving $1.2 million in taxpayer dollars over the next 33 months.

Two examples cited by investigators who worked on the study show the wide range of alleged abuses.

In one case, investigators said, a woman failed to report her husband’s earnings as a fruit picker for three months in 1992. At the time, she was receiving a $326-per-month AFDC grant for one of her two children, who is a U.S. citizen. The total overpayment: $652.

She was charged with a misdemeanor.

In another instance, a woman had reportedly withheld information that her husband was employed and living in the same house with her and their two children. As a result, she received $13,939 in AFDC and food stamps, investigators said. Prosecutors have filed felony fraud charges against the woman, who was not identified.

But critics have questioned whether state and county welfare officials deliberately targeted largely undocumented immigrants as a main sample in the study because they suspected they might find a higher incidence of fraud among that group.

The study focused on a subset within the total AFDC (Aid to Families with Dependent Children) caseload that is 68% Latino--half of whom were either undocumented or in the process of obtaining amnesty. Latinos in general make up only 38% of all aid recipients. Meanwhile, despite the fact that the authors of the study cautioned others against using their findings to make generalizations about the overall AFDC population, their research has been widely cited by lawmakers statewide to help justify spending millions of dollars on anti-fraud detection projects, including fingerprinting of AFDC and General Relief recipients.

However, analysts say the so-called Child-Only Fraud Pilot Project has obvious statistical flaws, including the method that investigators used to calculate the 62% fraud rate.

Investigators stated in the report that they had rooted out 135 cases of fraud. Based upon the original computer-generated sample of 500, the fraud rate would have been calculated at 27%. Instead, investigators threw out 233 cases for a variety of reasons, then figured the 62% rate based upon the number of cases that they actually probed--215.

Advocates for the poor have been especially critical of the low number of prosecutions--2% of the original sample of 500.

They also rejected the conclusion that the 89 people who voluntarily asked to be taken off aid during the investigation must have been welfare cheats, as suggested in the study.

Officials determined that fraud was present if there were any “substantial changes to case eligibility because of the investigation.” These changes might include a family’s failure to supply the necessary information to investigators, a change in the makeup of the household, a grant reduction, an overpayment or the discontinuation of aid.

But Mark Greenberg, director for the Center For Law and Social Policy in Washington, says there are other reasons why a person would ask to have their case closed.

“It is also possible that it is a frightening and stressful experience for anyone to be investigated in this way,” he said.

Wayne Field, chief of welfare fraud investigations for the district attorney, maintains that in all 135 cases, the people were in fact guilty of welfare fraud.

“I can tell you that in every single one of the 135 cases, it was not human error but intent to defraud. If you didn’t tell month after month that you were working under an assumed name, that’s fraud. That’s what we found,” he said. “You can get the wrong idea about the aggressiveness of the D.A.’s office, but it’s certainly not our fault that the evidentiary information is not there for us to proceed,” Field said. “The most important point is where the real savings are. We’ve got them off (welfare) and they’re not cheating. That’s where the real savings is.”

The study states that every dollar spent on fraud investigation results in $7.29 in taxpayer savings.

However, according to the Legislative Analyst’s Office in Sacramento, a nonpartisan research group that serves state legislators, the study overstates the amount of future savings.

Bill Lucia, who analyzed the study recently at the request of a legislator, whom he declined to identify, said investigators inflated the average length of time that recipients remain on aid and used higher, outdated average grant levels to determine savings.

Despite the apparent inconsistencies, Lucia said, the Orange County study figured prominently in a legislative subcommittee’s decision earlier this week to recommend that the Assembly approve $330,000 for similar fraud studies in Alameda and Fresno counties.

McKeever, directing attorney of the Western Center on Law and Poverty in Sacramento, has been a vocal critic of what he sees as the deliberate manipulation of the Child-Only Fraud Pilot Project.

McKeever said officials used the 62% fraud rate indiscriminately and irresponsibly. “But it serves the purpose of exaggerating the extent of the phenomenon to persuade the State Legislature to devote more money to anti-fraud activities.”

However, Amy Allbright, a spokeswoman for Gov. Wilson, says the Administration has been careful to specify when citing the Orange County study that it deals with a very small subset of the welfare population.

“Critics tend to attack people who bring up welfare fraud as a potential problem in public assistance programs,” she said. “But the public believes that they are being ripped off. Everybody knows somebody who is beating the system.”

However, officials in Orange County say they, too, are uncomfortable with the fact that their research has been used elsewhere to draw erroneous conclusions about the whole AFDC caseload.

“The 62% figure was for this particular study for this particular segment of the population--period,” said Angelo Doti, county director of financial assistance. “It cannot be projected on to another caseload.”

The study dealt only with those cases in which a child was living with a parent who was not eligible for aid--either because the parent is undocumented or in the process of receiving amnesty, or has already been dropped from the program for not trying to find a job.

Orange County plans to launch another $1-million study, this time examining fraud in the overall caseload. Doti said.

This time, he said, the county will hire an outside consultant to help in choosing the sample--a move designed to avoid criticism encountered after the November study.

Fraud Found

Investigators reported finding fraud in 135 of the 216 AFDC cases studied in December, 1992. Most of these voluntarily withdrew claim to benefits. Most of the reported fraud involved unreported income by those seeking benefits:

Fraud Case Status Withdrew voluntarily: 89 Criminal charges filed: 6 Under investigation: 17 Referred for re-investigation: 23

Fraud Indicated Unreported income: 77 Absent parent in home: 23 Household composition: 10 Not residents: 7 Failure to cooperate: 3 Child out of home: 3 Other: 12

Sample Difference

The sample drawn for the study differs in ethnic composition from that of the entire AFDC caseload:

Study sample White: 15.4% Latino: 64.6% Asian: 11.2% Other: 8.8%

All cases White: 27.1% Latino: 38.1% Asian: 19.6% Other: 15.2%

Source: Orange County district attorney’s office