Bank CIOs accepted a long time ago that core systems needed to be replaced. And for a while it seemed like, one by one, slowly but surely, their institutions would get around to it. But that never happened. Yes, they've made incremental changes - patched here and extended there - but increasingly they fear that this approach is hitting a wall.
In conversations with bank technology leaders, on and off the record, a picture emerges of banks at an impasse when it comes to their core systems. The need for upgrade and replacement - driven by a growing wishlist of real-time applications, ability to offer new products and nimbleness to react to new regulations with speed - is reaching all-time high levels. But the list of reasons why nobody wants to take on these projects hasn't gotten any shorter - and in many ways has gotten longer.
But CIOs aren't sitting still. They're wrapping services-oriented technology around their legacy cores, adding real-time database layers and looking into core upgrades. And in the process, putting their banks' technology in a more competitive position.
"Most of the core systems in the larger banks in the U.S. are literally 30 years old in terms of base code and they're written in COBOL and Assembler languages," observes Bruce Livesay, CIO of Memphis-based regional bank First Horizon ($25 billion in assets). "Those core foundations may have been updated, but they're still old. A lot of the small banks are trying to figure out what to do, but most of the big banks, given their history of mergers and acquisitions, haven't consolidated onto a single platform so they're dealing with multiple old platforms."
Such older systems can't handle real-time transaction processing, but tend to process checks and other transactions in batch mode, with memo posting throughout the day. Going to real time would solve industry problems around overdraft fees and being able to state a customer's true available bank account balance at any given moment in time. "If you really, truly were real-time everything, a lot of those questions wouldn't exist," says Livesay. "I think the industry will move here and I think it's important. What the consumer really wants is to know how much money they have at any given time. The next generation, Gen Y users, will be a lot less likely to write a check and a lot more likely to expect to know their balance at a moment in time wherever they are."
The obstacles to a core replacement - the bank technology equivalent of a heart transplant - are immense.
One is fear of failure. A few well-publicized core conversion projects have run over time, over budget and with major glitches. There are few success stories of U.S. banks deploying the type of modern core banking system some European and Asian banks use that real provide real-time debits and credits to customers' accounts, with a real-time post to the general ledger.
"It will happen eventually, but there haven't been any successes today," Livesay says. "A core banking replacement is like a heart transplant, it's a major initiative that's difficult to cost justify and no one wants to take that risk and be the first, especially since there have been several failed efforts."
Another hurdle is the difficulty of making a business case for these projects. The top 25 North American banks, which run home-grown or purchased core systems in-house, have been modernizing over time by adding specialty engines on top of their core system to provide functions such as workflow, online banking and bill payment, notes Jim Washburn, vice president and practice leader for banking industry consulting at Cap Gemini. "The different stacks of technology have created an integration nightmare whenever anyone tries to think about refreshing the core package," he says.
Where once, core software controlled banks' reporting, workflow, and user interface, "over the last 20 years, we've stripped the core system down to being an accounting and transaction engine and built all these specialty systems around the core, either to keep pace with the boom in electronic distribution or to create efficiencies in areas that the core systems were not able to handle or adapt to," Washburn says. "That's made the IT environment for these CIOs pretty complex and replacing the core system would be an incredible task of integration. Where do you derive the cost benefit from making an investment in that?" he asks. "Where big banks have made conversions, usually an event has forced it, such as a merger or a failure, he says.
Integration is a challenge for banks of all sizes. "What everybody seems to be striving for is a nice technology that's easy to integrate, with some kind of services-oriented architecture," Washburn says. "Most vendors provide that in their own way." However, the vendors know that where banks have already bought or created specialty engines for workflow, online banking, etc., they can't supplant such modules, but will have to cohabitate with them. "It may happen in an easier fashion with a better architecture," Washburn says. "A lot of the large banks are trying to attack that themselves, developing their own interface layers. This may not be as elegant as what some of the vendors might provide, but it seems to be workable."
A third challenge is the elusive goal of real-time transactions to provide channel harmonization, such that if the customer goes to the branch and makes a deposit, then uses the ATM, online or mobile banking, they can see the transaction reflected in real time. No U.S. vendor has this today implemented in a U.S. bank. Few U.S. banks want to be a guinea pig for an untested real-time system.
"The vendors will say they have real time transaction processing capabilities and some of them do in some areas," Livesay says. "But they're all based on 20-30 year old core models that rely on batch cycles. It's not just the vendors' issue, it's partly the way this whole business has evolved in the U.S., where checks are fairly entrenched. Checks have to go to the bank operations center to be handled and mostly batch processed. You'll never be fully real-time until checks are gone."
Many banks are doing check image processing, in some cases having tellers scan checks in the branch that theoretically could be processed immediately. "Due to check 21, the image legally posts as paper," Livesay notes. "You could at that moment handle the image instead of the paper. But you have governing bodies in the industry, like the Federal Reserve, where the systems are still based on batch processes and ACH windows." The Fed's process of clearing and settling checks between banks is complex and takes time. More than half of U.S. banks do memo posting because of the Fed's paper-based mechanisms, which creates problems with Reg E compliance. "It makes it more complicated to figure out what someone's account balance should be at a given moment in time," Livesay says. "Which deposits go in, which monies should come out and when, what kind of a hold to you need on the funds for how long - it creates a lot of complexity and that confuses consumers."
Regulation has been another factor stalling core conversions in the U.S., because vendors are not eager to build such complex provisions into their software. "Realistically, when you look at the number of financial institutions in the world and the U.S., where are the vendors going to devote their time and attention?" Washburn says. "They want to make more than one or two sales. They do that by creating a product that has broad appeal. If I'm a vendor and I get the number-three bank in the U.S. to use my core app, that's great - I sold a license and I may charge a lot for it, but it's a one-time event for me. I need to continue to sell that product to other geographies."
Not only coming Dodd-Frank rules, but existing laws scare off vendors, he says. "Each state has its consumer lending laws, there are federal consumer lending laws, disclosure laws, and electronic transaction disclosures," Washburn says. "This is the most highly regulated country to tackle, and many of the vendors put it in the too-hard bucket for right now."
Another reason the U.S. is a tough market for modern core software is product complexity. Deposit processing is the same around the world. But on the lending side, the U.S. has distinct products that don't exist in other countries. "Our prime lending products in the U.S., such as fixed-rate non-portfolio mortgages that we sell in the secondary market, don't exist in other countries," Washburn says. "Home equity lines of credit don't exist in other countries. Those are our two largest lending items on the consumer side." Countries with simpler banking industries are more attractive targets for core vendors.
Some banks take a middle way, in which they service-enable an older core engine, wrapping a modern, services-oriented architecture around it. "My honest opinion is the two approaches in the end achieve the same result to real time," Livesay says. "I'm not a proponent that you have to completely replace your core transaction processing engine with a new system built from the ground up. You can leverage the scalable, proven transaction processing engines that are in some of these cores today."
Livesay doesn't expect to ever do a complete overhaul of First Horizon's core, which is a Hogan CSC system. "Part of it is we believe we have a system that will scale and take us wherever we want to go," he explains. "There would need to be a compelling difference between what we're doing with service-enabling it and what you could get from a different system. In the U.S., there will be a long tail on this, there is going to be a long period of time where we'll need to continue to support old ways of handling checks and clearing and settlement. It's not going to happen overnight."
Yet full-bore core conversions can be done in the U.S. One case in point is Washington, D.C.-based NCB ($1.6 billion in assets), which recently went through a conversion to Open Solutions' DNA core application at its Ohio bank. "The major impetus for the upgrade was to get the most modern technology on the retail side, for the tellers and branches," says CIO Gerard Donlin. NCB was one of the first U.S. banks to deploy this product.
A driver behind the bank's core replacement is that Bisys/Open Solutions announced that it would be sunsetting Total Plus in a year. However, as it turns out, at the end of the year, most of Bisys' other clients hadn't migrated and the product is still being supported.
One glitch is that Open Solutions' core is retail oriented and NCB has a nationwide commercial loan business, for which the bank was using Bisys' Total Plus core solution. (Open Solutions bought Bisys in 2005.) "We're gradually getting Open Solutions to expand that functionality and capability," Donlin says.
The staff at NCB are happy with the results of the conversion. "There were workarounds, but the system is running," Donlin says. "They had some management changes, and the newer group seems to be dedicated to improving all aspects of operations and product." Perhaps there's hope for banks thinking of taking a chance on a new core.
