The red flags kept piling up. On his welfare application, the man indicated he lived in Anoka County and didn’t make enough to file taxes. Then he accessed some of his medical benefits in North Dakota.

The inconsistency sparked a county investigation and led to the discovery that Robb Bastian actually ran his own company in North Dakota and had bragged about its success on social media. Last fall, Bastian pleaded guilty to wrongfully obtaining assistance — a felony — and was ordered to pay nearly $26,000 in restitution.

Some welfare cheats, like Bastian, misrepresent employment or residency. Others inaccurately report assets or household composition. County and state officials say they aim to spot these red flags before benefits are doled out, preventing ineligible applicants from bilking taxpayers out of potentially hefty sums.

“Anything we can stop on the front end is certainly beneficial all the way around,” said Judy Grandel, of Hennepin County’s fraud investigations unit.

Across Minnesota, the stakes for recipient welfare fraud are high, despite making up only a fraction of public assistance cases overall. Last year, investigators completed more than 7,300 investigations and halted or reduced benefits in nearly half, according to the state Department of Human Services (DHS).

Since 2012, about $24 million in overpayments — benefits paid to which recipients were not entitled — have been identified through the state’s Fraud Prevention Investigation (FPI) program.

The state works with county agencies to curb public assistance fraud through the DHS Office of Inspector General, which was created in 2011 to increase the department’s focus on these kinds of investigations. Minnesota’s FPI program traces back to 1989 and covers 74 counties, providing grant funding to 12 individual counties as well as 14 regional investigation units. Last year, this grant funding totaled about $3.1 million.

County by county, these programs look wildly different, both in staffing and structure.

“Each county handles their cases very uniquely,” said Vicki Kunerth, deputy inspector general of fraud investigations.

Despite this variability, each county-level program tracks totals related to its fraud prevention work. From caseload sizes to turnaround times, these numbers help the state gauge a program’s efficiency.

County officials say key measures include cost savings by heading off improper payments, identification of overpayments, and the program’s cost-benefit ratio, which the DHS has described in training documents as the standard it “monitors most closely.” (DHS says it is revising its FPI manual to reflect that the ratio “is just one measure of efficiency.”)

State data from the past five years shows that these numbers can vary significantly from year to year. Investigators say their fraud totals largely hinge on the cases involved and the types of referrals coming from county workers.

Recipients claiming to raise children alone at the same address may trigger a referral, for instance. In one case, investigators were tipped off to a woman’s true marital status by a Star Tribune article.

Lavish assets can also raise suspicions, as in the case of a Blaine couple who got food support and medical benefits despite an income that exceeded $400,000.

Each county’s yearly recipient welfare fraud numbers raise questions about how it’s prevented, at what point it is caught, and which methods do a better job for taxpayers. But DHS officials say they have “never identified a need to rank all FPI programs in the state” and have not “developed criteria for doing so.”

By the numbers

In the past five years, Hennepin County has topped the seven-county metro for estimated cost savings, or fraud caught before benefits are given out.

Anoka County, by contrast, has consistently outpaced its larger neighbors in identifying overpayments. Anoka County officials say less than 1 percent of public assistance involves fraud. But state data show that even that sliver of cases resulted in more than $820,000 in overpayments last year.

Anoka County’s relatively high overpayments don’t necessarily mean that more fraud goes on there, county officials said.

“We’ve always taken a very aggressive stand [against fraud], and that may be why you see the numbers you do,” said County Board Chairwoman Rhonda Sivarajah.

For years, Anoka County has also used DNA testing to cinch cases that involve paternity questions and is among the first nationally to do so.

Recovering money that has been overpaid can take years, with some never paid back. Collecting funds may involve payments, reducing ongoing benefits, tapping income tax returns or even drawing from lottery winnings, county officials said.

Only death or a court order will discharge unpaid debt in Anoka County, according to Dee Guthman, deputy county administrator. Records show the county has averaged about $420,000 a year in overpayment recoveries over the past nine years.

DHS officials said counties may have spurts of large overpayment totals due to cases going undetected for long periods, or involving costly benefits like medical assistance. DHS officials said they “are not concerned with this variability.”

How effective?

The cost-benefit ratio was devised as a key measure of a county program’s anti-fraud work. The ratio calculates how many dollars are saved for every dollar spent. Counties aim to stay above a $3 state-imposed benchmark. Like other performance measures, the annual ratio varies, influenced by factors like staffing, according to the DHS.

Some in Washington County attributed its low ratio last year — $1.54 — partly to staff turnover, said Sgt. Sara Halverson.

A low cost-benefit ratio may also reflect a drop in the number of case referrals, said Todd McMurray, who oversees Chisago County’s anti-fraud program. Over the past five years, Chisago County’s cost-benefit ratio has averaged just below $2, among the lowest in the state.

“If you don’t have a lot of referrals, you don’t have a lot of chances for savings or overpayment situations,” McMurray said.

When the ratio dips low enough, the state requires a county to draft a corrective action plan, McMurray said. This plan may propose more staff training, for instance.

Sivarajah, of Anoka County, said the fraud prevention program ultimately is about ensuring that the right benefits go to the right people.

“When people are committing fraud and not being honest about income and assets and living situation, we are going to take that very seriously,” she said.