Originally created 12/24/01

Banks are watching what you buy



Stephen and Kara Yates discovered the banks' system monitoring credit card purchase patterns the hard way: Their card was rejected when they tried to use it during a vacation in Europe.

The card was declined when they got to their hotel in Paris. They called Chase to fix the problem.

But it happened again when Yates was in a boutique buying two leather jackets - one of them a surprise for his wife. He called again and then again when the card was rejected for a third time.

Chase had been monitoring the card for unusual spending and electronically alerted the shop and hotel that the purchases didn't fit in with the Yates' normal buying pattern and required additional authorization.

"You can't just freeze an account on someone," Stephen Yates, 32, said recently from his home in St. Petersburg, Fla. "The worst thing you can do is freeze an account in a foreign country."

But in this high-tech world, banks and credit unions review virtually every transaction on every credit card in the United States as the institutions attempt to track down stolen cards that cost them millions of dollars each year.

Purchases are monitored by computer as they occur - 24 hours a day, seven days a week - and compared to all previous transactions on a particular card in less than half a second.

"It's the most sophisticated technology in the business," said Michael Cunningham, head of Chase's fraud investigation team. "It has had a tremendous impact."

But that doesn't mean honest customers won't receive a call on occasion from a bank with questions about their purchases or, in extreme cases, have their card rejected at a store. Some industry experts estimate as much as 80 percent of flagged transactions are legitimate.

"A customer who doesn't understand why this is happening may say, 'Why would you do this?"' said Maria Mendler, a spokeswoman at Citigroup, which has 90 million cardholders in the nation. "Once they understand what's happening, they're OK with it."

While sophisticated criminals have learned to fake photos and signatures, steal cards and personal identification numbers, the one skill they haven't mastered is tracking your pattern of buying.

Banks take advantage of that.

"They can't adopt your pattern," said Wesley Wilhelm of HNC Software, a top provider of neural network programs. "That's why there is nothing that comes close to this for finding stolen, lost or counterfeit cards."

The software determines how risky a purchase is by comparing it to your shopping history with an emphasis on several factors including price; location of the purchase; time of day; day of the week; season; how many purchases are made in a row; type of store; and items being bought. It studies how long your account has been open and even your payment history.

Not only does the software know what you are most likely to buy, it also knows what a criminal is most likely to do. It knows that a typical credit card fraud consists of a small purchase at a gas station and then, if that works, high volumes of merchandise in various stores at shopping malls, said Brian Fisher of ACI Worldwide, a company that sells software to Bank of America and others.

Financial experts say the program reviews data, using complex algorithms to search for changes in patterns and scores a purchase on a scale between 0 and 999 - the higher the number, the riskier.

Though the software is developed by a handful of companies, such as ACI Worldwide and HNC, it's customized by banks and associations, such as Visa and Mastercard.

Some banks call customers and ask them about their spending to determine whether it matches up with what the card actually was used for. If a purchase is considered extremely risky - maybe a big-ticket item in a foreign country - a credit card machine will alert a merchant trying to authorize a purchase to call the bank and check a photo identification.

Most bank officials, including Chase, said stores aren't supposed to reject a purchase outright but they realize some salespeople might find it easier to just ask for another card.

Neural networking, first used in the financial industry, has made its way to a variety of fields, including health care and insurance. And one company has even announced plans to use it to combat terrorism.

The system has improved constantly since it was first used in banks a decade ago and now is able to take into account more than factors than ever before - including where you are likely to shop this Christmas.

Some customers worried about their privacy when the program was first introduced. But banks have worked to alleviate those fears by assuring them that the information would not be shared and that only a computer - not a person - was monitoring what they bought.

Though banks refused to disclose how much money they have saved since installing the system - including American Express and Discover, which wouldn't talk about security issues at all - Fisher estimates it's at least 30 percent.

Banks are eager to reduce fraud because they usually assume the cost of a lost or stolen card, though customers may incur some of the costs through higher interest rates and fees.