One of the most underappreciated hardships banks face is what to do about duplicate deposits. Each day, there are hundreds of checks or electronic payments processed two or more times, the result of a client who uses remote deposit but later absent-mindedly drops the scanned checks off at the branch or a data-center worker who processes checks already cleared by the previous shift.
Whatever the root cause, these duplicates are becoming harder to manage and increasingly expensive to weed out, perhaps between $100 and $125 per item, say payments officials who are looking to tame this uniquely Check 21-era problem.
"In the old days, duplicates rarely occurred" since checks were nonreplicable, says Tim Dillow, a senior vice president and regional operations support manager for BB&T in Winston-Salem, N.C. But after image replacement documents came along, "we were dealing with a lot of duplicates. In that environment we impacted our clients negatively, and we were exposed to losses."
These days, BB&T handles duplicates through an automated process that singles out these items before they are posted, matching them against a running file of previously cleared items. But analysts say BB&T is one of the few major banks doing this, with most others catching duplicate checks only after they've been posted and double-dip into a customer's account.
The inability to catch double-posted items also feeds a prevalent form of check fraud-duplicate presentment. Banks and industry groups worry the problem could worsen as the volume of consumer remote capture and mobile payments rises.
At BB&T, which processes 40 million check, card, commercial and automated clearing house payments a month, the bank usually finds 900 or more "suspect" duplicate items. Three-fourths are "false positives" of actual first-run deposit items. But the remainder are duplicates that commonly come from a remote-deposit client who brings in checks that have already been scanned, or even BB&T's own tellers who mistakenly re-scan old checks from image-teller stations.
"Today, our biggest source of duplicates are for the most part our own BB&T branches," says Dillow.
Analysts say the best way of catching these items requires real-time identification on day one. BB&T uses a duplicate-detection system from Conix Systems, which creates a database file collecting details of all incoming payment items from the preceding 45 days in order to flag any matching items that get re-posted.
Dillow says BB&T stops about 4,400 duplicate items a month, most in the receiving stage when they are being reconciled in the batch prior to posting. Many institutions use second-day detection techniques, which often will require action to reverse the posted item and issue a return. Because of the growing use of remote capture services, many banks would like to move to day-one detection.
But branch operations frequently don't have access to daily items deposited via ACH, through which converted checks arrive. Channel isolation creates the "possibility an individual item may get mistakenly, or intentionally, submitted through different channels," says Aite Group senior analyst Wesley Wilhelm.
Another complexity is how accidental duplicates differ from the fraudulent type. Banks usually isolate duplicates through account numbers, check numbers, dollar amounts and payee names. "However, when you start looking at the intentional duplicates,' it's not necessarily a good idea to make it a rock-solid, hard definition of what a duplicate is," says Wilhelm. With a fraudulent item, the payee or check amount may be altered, or the check may be deposited outside the 30- to 45-day window most banks would be tracking.
Analysts say duplicate check fraud has grown into an estimated $500 million problem for banks-half the estimated $1 billion in total check fraud in the U.S. A major gap in stopping duplicate fraud is knowing when a check has been deposited at multiple institutions.
One image-exchange industry consortium, the Image Industry Interoperability Group, has recommended the creation of a new communications link outside the normal method for handling return items. "It'd be nice if somebody who drops off a fraud item could be detected right off the bat," says Steve Fortson, vice president for Conix Systems.
While there is impetus to improving day-one detection technology, there are technical barriers. Matthew Bowen, a senior vice president with Fidelity National Information Services, says cross-channel duplicate detection would require a massive database to limit the timeframe of item comparisons. "The problem for a big bank is because their volume is so high, they can [only look back] a day or a week," Bowen says.
Detecting duplicates is sure to become a more pressing problem as the rapid expansion of remote deposit capture continues, especially in the mobile space. Dillow fears that the explosion of smartphone deposit-capture and peer-to-peer payments applications will greatly increase the number of redundant items coming from those users who'll re-send payment requests, unsure if the first one got through. "Looking forward, there are all kinds of potential for new sources of duplicates," says Dillow.
