Standards Convergence: Will it Ever Happen?
Everyone knows that standards are important; that’s why we have so many of them: SWIFT FIN, ISO15022, EDIFACT, FIX, ISO20022, FpML, XBRL, TWIST, ACH and RTGS formats, and OMGEO, for example. There are historical and political reasons for this proliferation. Most of today’s established standards arose in isolation within disparate user communities. For instance, SWIFT FIN came from the international banking community with the original intention of automating the telex; while FIX was a product of the securities industry, originally developed by Salomon Brothers and Fidelity Investments as a bilateral electronic protocol for pre-trade communication.
Typically, users of a standard want to maintain a controlling interest in its content and technical direction because the standard determines, to an extent, what business they do and how that business is processed. Attempts to merge or rationalise standards often falter because the communities that created them are unwilling to concede control to another group and, having already implemented the standards themselves, see little immediate benefit in changing them.
During the 1980s and 1990s, after the pioneering standards were first established, the original user communities expanded and the range of activities covered by each standard grew. As usage and coverage spread, standards first began to touch and then to overlap with one another. Today, the origination, confirmation, settlement and reconciliation of even a simple transaction may involve the use of several standards, each with different rules, syntax, format, protocol and communications requirements. With the advent of XML and the simultaneous evolution of cheap and secure connectivity, yet more standards have arrived – each expressing the needs of the business community that created it.
Service-oriented architectures (SOA), the latest fashion in application design, depend on loosely coupled services that produce and consume messages that are also described by standards, whether proprietary or public. Financial institutions need to maintain the systems and expertise required to track standards and manage the standardised data flows that are the lifeblood of their business.
In the meantime, the industry talks about standards convergence. There are isolated examples of converging standards, but these usually come about from the consolidation of business infrastructures (for example, the union of principal European securities settlement systems under Euroclear). It is rare that different standards bodies manage to converge when the users they represent remain distinct.
The ISO20022 International Standards Team Harmonization Group has made good progress on customer-to-bank payment initiation messages, but this is one of the simpler functional areas that financial messaging standards cover, and none of the converged standards were well entrenched before the project began. Other initiatives to converge standards, such as ISO15022, have yielded little practical benefit in terms of standards convergence, despite the warm words and public support they received from the major bodies in the financial standards community.
This is a major headache for financial institutions. Institutions may not analyse what it costs to stay current; it is seen as an ongoing overhead – part of the cost of doing business and rarely looked at in isolation. But when the numbers are added up they make sober reading. For instance, Misys estimates that up to 25 per cent of institutions’ IT budgets is spent on tracking and implementing standards.
If standards convergence remains a dream, and manually tracking standards remains a nightmare, what is the alternative? Standards interoperability is one answer to this question. The idea is that rather than all standards converging, each retains its own syntax and format, but measures are taken to ensure that conversion between them can be performed automatically. The key to this is to ensure that each standard shares the same fundamental ‘view of the world’, at least the ‘business world’ covered by the standard.
The most comprehensive attempt to codify this ‘world view’ is the ISO20022 Repository. This is an attempt to build a model (an ontology) of financial industry business in a message format and syntax independent representation. The representation chosen is based on unified modelling language (UML), an abstract design language more commonly used to describe interactions and relationships between classes in object oriented software systems. Because they are derived from the same formal ontology, messages based on the ISO20022 Repository should all share the same factors in terms of the meanings attached to data elements, and should, therefore, be readily interoperable.
So far, only the core payments kernel payment initiation messages and investment funds distribution messages have been produced in this way, but for new standards, at least, ISO20022 offers some hope of automating the mapping between messages from different functional areas, reducing if not eliminating the need for laborious manual mapping.
When work on the ISO20022 Repository began (it was originally known as the financial dictionary) the aims were bolder. Rather than permitting interoperability between standards designed from scratch around its definitions, the belief was that if existing standards could be mapped in a one-off exercise to the Repository, then all mapped standards could interoperate with one another. For one reason or another this never came about. Principally, this was because the old standards were not themselves designed around a formal ontological framework and without a stable and unambiguous reference to something outside the context of the informally described message definitions, it is difficult to see how reliable mappings could be constructed.
The Semantic Web is a project of the World Wide Web consortium that aims to move the content of the World Wide Web from formats mostly comprehensible by people (HTML) towards formats ‘comprehensible’ by machines. XML is a key enabling technology for the Semantic Web, but several other initiatives have been launched under this umbrella including the definition of a standard formal mark-up language for ontologies (OWL) and tools for managing and querying ontologies, such as SPARQL.
This work is currently the subject of much academic research (some of it EU and US government funded). A particular research focus is ontology mapping, i.e. mapping information expressed by one ontology into a form defined by another. Research aims vary, from providing mapping tools or program language bindings to allow mappings to be constructed manually, to using machine learning techniques to discover mappings automatically. All of this is of course highly relevant to the problem of standards interoperability in achieving the goal of machines being able to ‘understand’ the content of messages and automatically convert them from one standard to another, and perhaps perform other functions too.
Generally, standards convergence only happens when market infrastructures merge and their individual users become part of the greater group. In other instances, standards interoperability seems to be a more fruitful way forward. Automated interoperability between legacy standards is fraught with technical difficulties; mapping between these standards will remain the province of manually configured middleware. For newer standards that are based on the same formal ontology, automated interoperability should already be achievable. In the future, as research from the Semantic Web bears fruit, it may also be possible to map automatically between ontologies. In the meantime, however, financial institutions need efficient middleware.