Even as check volumes decline, high-volume check issuers still have to find ways to protect against fraud.
Hoping to rally the defense, the Federal Reserve Bank of Philadelphia has been investigating the possibility of using seal encoding technology to detect fraudulent items among the approximately 250 million U.S. Treasury checks annually presented for payment.
Seal encoding invisibly stores data such as the payee and the check amount within the issuing bank's logo or in another area on the check. Therefore, if someone intercepts and alters a government payment check, the amounts written on the check would no longer match the data contained within the seal. This technique can also foil basic forgeries using a laser printer, as the encoding employs a proprietary format presumably unavailable to counterfeiters.
The technique would help the Federal Reserve with its check processing responsibilities performed on behalf of the Treasury. After imaging Treasury checks as usual, those check images would be sent to a detection server.
"You don't slow down your sorter," said Arun Jain, vice president, retail payments, Federal Reserve Bank of Philadelphia. "Once the images are flowing to your detection server, you're doing two things-you're decoding the server, and you're digitally interrogating the payee and the amount."
Since printing errors and teller ink stamps can obscure the seal encoding, the detection software has to allow for some degree of error tolerance. Otherwise, an excessive numbers of false positives would drive up the cost of manual intervention.
An element of user education also comes into play. First, tellers have to be trained to avoid stamping on top of the small seals, and thus obscuring their contents. Second, to ensure image quality at the point of scanning, check processing operations personnel have to be willing to take a time-out for quality.
"When you're imaging checks and it's falling outside of some standard tolerances-well, stop the sorter," said Jain. "When you're relying on image exchange and you're relying on image detection technologies, you have to have image quality as a requirement."
Any checks flagged for review would then be sorted into various categories, including encoding errors, unidentified seals, and altered items. "Some may be true fraud, and some might be encoding error, so you want to have your people look at it," said Jain.
Implementing the new process would quicken the Fed's response time upon detecting an item suspected as fraudulent. That's a slow process as it stands today. "We send the information to the Treasury, and they will take a couple of days to post those items to their issue file," said Jain.
Only then can the Treasury determine which items were never properly issued, and request that the Federal Reserve take action. "They'll come back to us, and we'll do the research and send an adjustment back to the depositing bank," said Jain.
The turnaround would be significantly shorter through seal encoding. "If we detect this item right away in our shop, and we know that it's a fraudulent item, we can send that item back to the depositor that day," said Jain. "It can save several days in that process."
Although the initial implementation of seal encoding would be limited to internal validation, the seal decoding technique could conceivably take root further upstream in the chain of collection, such as at the bank of first deposit or check cashing agencies. "It can get to that point eventually, but there it gets more involved," said Jain.
One important factor that seal encoding has going for it is that it doesn't require the physical presence of the check itself. Given the recent industry push towards check truncation, it's essential that any new security measures work well within an image exchange system. "People want to know that whatever security features are put in are image-survivable," said Jain.
Nevertheless, the Fed is also conducting a research project involving digital watermarking. With this technique, the paper used for checks would be treated with chemical inks in a secure environment, thus making it more difficult to counterfeit a check. But it would also make checks treated in such a fashion more expensive to process, since verification would have to take place at the point of first deposit.