JP Morgan’s Reference Data Overhaul

"This is a substantial initiative, even for us," says Philip Fasano, global business executive for technology at JP Morgan Investor Services and senior vice present at parent JP Morgan Chase. JP Morgan Investor Services (JP Morgan Chase’s custody division) has nearly $7 trillion in assets under custody, of which approximately $2 trillion are global custody assets, making it the third-largest US custodian after State Street and Bank of New York and number one in terms of overall global assets because of its worldwide reach. "It required us to provide a multiyear commitment to it both in capital and people, and then ultimately it required the direction and commitment on behalf of the organization’s management to see this through," Fasano says.

The new data repository covers all asset classes--including instruments like mortgage-backed securities and derivatives, for which data is often complex to manage. "Just doing any of those is a major challenge," says Fasano. "But our world of securities is a global one, and we’re driven by our customers with respect to the types of securities that we have to warehouse in our database."

In his role as what is essentially the chief information officer of the Investor Services unit, Fasano is responsible for all of the division’s technology. Fasano, who has been at JP Morgan for roughly a year, has been the CIO at a number of other financial services companies including Deutsche Financial Services, a division of Deutsche Bank. He is also a member of the Technology Advisory Committee of the Depository Trust & Clearing Corporation (DTCC).

Reference data has been a hot topic for nearly a year now. Mark Watson, global offering manager for IBM’s Global Financial Markets division, says that centralizing reference data is a high priority right now for most firms because it will ultimately reduce costs and increase efficiencies. "Firms sometimes have hundreds of reference data systems floating around, and they’re looking for ways to consolidate to save hardware, software and support resources, and to distribute the data more easily," Watson says.

Watson says most firms are in the initial or middle stages of their reference data projects. An in-house repository that covers complex asset classes is unusual, he adds. And many firms weigh buying solutions or outsourcing to an ASP model as opposed to the costly prospect of building something from scratch.

Other financial firms have started similar reference data projects of the scope of JP Morgan’s and subsequently abandoned them because the task was too daunting and the return on investment was too slow in coming, says Tim Lind, senior analyst for The TowerGroup. He declined to name the firms in question. But those firms might return to the data reference chore soon because the products many firms initially used didn’t include the full suite of integration capabilities, data transformation tools and business rules for choosing which source of data is most reliable. The full package is increasingly successful in cutting the rates of costly failed trades.

Lind is bullish about the potential of centralized data repositories, "But short term, they suffer from what every six-figure price tag software [product] suffers [from] in this economic environment."

In fact, reluctance on the part of other firms to begin a project of this scope may one day translate into revenues for JP Morgan Investor Services. Fasano says the project had three goals at the time he found it--to build GMRD for internal use within Investor Services, to build it for JP Morgan Chase on a corporate-wide basis and to potentially sell it.

Internal use was the first priority while the second two goals were treated as equals. "Our thinking was that if we’re going to service JP Morgan Chase, it would have to take a similar form to anything that would service our clients as a product itself," Fasano says.

JP Morgan is currently discussing whether to outsource the product on an application service provider (ASP) model or to sell it as a stand-alone product. Fasano says all options are still being explored. "It is our understanding that as comprehensive as it is, this product is one of the first of its kind, so there is a definite marketplace for it, but at the same time we believe it gives us a competitive advantage," he says.

Just in the past year, two vendor-owned reference data ASP’s have appeared. Reuters and Capco teamed up to offer one of the currently available outsourcing solutions. The alliance, called Synetix, has released Reuters Reference Data Manager as its first product. Acdex, a similar ASP-type utility, is set to go live with its first customers in the third quarter.

As Investor Services’ global central repository for reference data, GMRD is designed to improve the quality and accuracy of the data and reduce operational risks associated with faulty data. GMRD first captures reference data from various sources, then centralizes, normalizes and cleanses the data to establish consistency across the firm’s core applications. A user interface enables the operations staff to repair erroneous data and also perform any necessary updates and maintenance, says Fasano.

The project cost "many, many millions of dollars" and required a team effort between the firm’s operations and technology organizations, Fasano adds. One financial industry expert, who requested that his name not be used, estimated that a project of this scope would have cost JP Morgan $40 to $50 million.

GMRD receives data updates electronically from 12 vendors to which Investor Services subscribes. The data captured by GMRD serves as a core component for the firm’s other systems that rely on securities data as a backdrop to their performance.

GMRD currently supports only asset indicative data, which is a composite of descriptive information, such as a ticker symbol or QSIP, that uniquely identifies a security or underlying asset. Over the next two years, Fasano says GMRD will phase in support for other types of reference data like support pricing data, institution information, asset servicing notifications and client account data. "Asset indicative was our first priority--essentially we viewed it as the foundation for everything else," he says.

GMRD has essentially consolidated all of the firm’s combined reference data requirements into a massive system that in turn is important to the firm’s other systems. "When you consider the types of products that we as a firm have to offer--all of our Internet capabilities, our primary custody and securities lending systems and other core systems that process our business on a day-to-day basis--they rely on GMRD," Fasano says. "We have a number of outsourcing initiatives and research products for which this is a fundamental building block--each of those areas looks to GMRD as a reliable source of reference data.

"It’s insuring accuracy of the larger-scale systems that are the foundation of our operation--it’s at the heart of every one of those systems," Fasano says. "For example, as a custodian, we clearly process quite a number of trades, so trade data that comes in would then be checked against the securities indicative data to be sure that all of the information incumbent on the primary system was there and actually correct."

Another system that relies on the reference data is the JP Morgan Concentrator, a browser-based application that will link firms to the virtual matching utilities (VMUs) as part of the industry’s straight-through-processing initiative. The Concentrator will route transactions, perform third-party standing settlement instructions (SSI) maintenance and manage basic reference and statistical data. Its first phase is due out this month, covering the Global Straight Through Processing Association (GSTPA). The second phase, connecting to Omgeo and third-party message routing, is due out in December.

"In the world we live in, we talk a lot about STP," Fasano says. "We believe having a resource like GMRD within our firm enhances our ability to operate on a straight-through basis, and straight through includes reduction of failed trades, ability to optimize our own environment on behalf of our customers and our ability to really process their trades seamlessly for them."

GMRD operates on a batch basis with some real-time functionality. It receives data on a real-time basis, from internal as well as the vendor systems. The data is cleansed automatically by GMRD, and operations staff work in real time to immediately resolve any issues the system picks up and kicks to the user interface, Fasano says.

"As a database, we have an awful lot of cleansing that goes on with the information that’s coming in from the vendors, and we do a good deal of reference checking across our portfolio to make sure that everything’s within tolerance," he says. "Anything that’s not within tolerance gets bounced to a repair queue so that our operations people can take a look and fix it quickly."

Using the system’s front end, the operations staff determine whether the data in question needs to be repaired and then re-enter it into the database so that it can be used by the firm’s other systems, Fasano adds.

Investor Services has already seen increased operating efficiencies as a result of GMRD. "As a consequence of using GMRD, we’ve found that we’ve really minimized, compared to what we had to do historically, the challenge to the operations staff of dealing with repair queue issues," Fasano says.

Investor Services developed GMRD to provide the rest of the firm’s systems with a single source of enriched, accurate reference data. "Our systems critically rely on this data and the integrity of it so that our customers can be relatively guaranteed that this aspect of our business is absolutely world class and that we’re capable of delivering the information they seek in the various reports that we generate for them," Fasano says. "Our technology organization felt it imperative to create this capability for the firm as a fundamental part of our overall global architecture."

Investor Services has hundreds of systems firm-wide, with its primary systems supporting the custody and securities lending businesses. "Looking at custody and securities lending alone, that’s a very, very substantial infrastructure and franchise in those businesses globally," Fasano says.

These systems were traditionally built with their own security masters and reference data master files. "Before GMRD, these databases were maintained by disparate groups of operations staff as well as disparate groups of people in the technology organization," Fasano says. Different systems might be using different versions of the same data. GMRD provides one single source of data for all the systems to draw upon.

"The foundational issue and one of the reasons we thought it very important--within the framework of our overall architecture--to develop a single common reference data resource was to eliminate some of that portfolio and maintenance backlog with respect to the number of systems warehousing this data," Fasano says, "But more importantly we wanted to provide that single source that would be definitive within JP Morgan Chase and to feel comfortable that we’re maintaining it as professionally as we can on behalf of our customers."

Because of its vast size and scale, Fasano believes GMRD provides Investor Services a value-added service and a competitive advantage. "We have a number of major initiatives underway currently that rely on this being a key component of their build-out, so the groundwork has been laid by implementing this system, which we think raises the bar substantially on what our competitors are doing," Fasano says. It’s also a potential additional revenue source.

Fasano says implementation of GMRD was quite involved as Investor Services owns a combination of disparate legacy and newly implemented systems. "I think what most folks who attempt to do this underestimate, is the resource commitment you have to put in place to permit yourself to implement this kind of a system and do it with minimal post-implementation risk," he says. "I was actually surprised and relieved at the fact that we had almost no post-implementation issues."

Thorough testing before implementation was key to its success, Fasano says. "We had done our homework and tested these systems and their potential output in concert with our many production systems to make sure the points of integration had been bedded down such that after the implementation was complete our operations went on pretty seamlessly," he says.

GMRD is built on IBM mainframes, using IBM’s DB2 family relational database management system products as a core database. The application framework is C++ and the Web front end is IBM Web Sphere. "We have built this out across the DB2 platform with the intent to leverage it more broadly within the firm and then ultimately to have the option to put it out there as a product that we could sell," Fasano says. "We built it to be the foundation of a larger architecture both for our firm and for anyone else who would look at it potentially as a product."

Fasano says the large scale of the project and of the firm itself required more traditional architecture tools. "When it comes to products that are part of our core operation, and this would fall within that, we tend to build those out on the large-scale pieces of equipment that fall within the IBM framework," he says.

GMRD’s centralized data management unit, which provides 24-hour global operations support, is located in Dallas. JP Morgan also leverages the Dallas office on the systems side for application and production services, Fasano says.

Also, JP Morgan used outside vendors to help build GMRD, Fasano notes. "We involve our technology partners as a matter of course on all our major initiatives," he adds.