Financial data headed toward easy distribution with XML

If stock exchange and money market information does not necessarily take up a disproportionate amount of space in a daily newspaper, the work of publishing these data sometimes seems to be out of proportion.

Today, most newspapers publish financial data using systems devoted to that work -- a good example would be the Associated Press' Grand Central Stocks, which uses a Macintosh and Quark XPress to get the paginated information into the paper.

But what about the future? What's on the horizon to allow newspapers to handle financial data in a more robust fashion?

The implementation of news industry data exchange standards, including NewsML and News Industry Text Format (NIFT) (see The Cole Papers, October 2000), continues to evolve with extensions to XML, such as Market Data Definition Language (Mddl) -- also known as Market Data Markup Language (Mdml) -- and eXtensible Business Reporting Language (Xbrl).

According to the Financial Information Services Division (Fisd) of the Software and Information Industry Association, "Interest in XML for market data is clearly growing."

The XML-based news media data exchange standards are NIFT and NewsML, developed by the newspaper- and wire service-oriented International Press Telecommunications Council (IPTC) of the United Kingdom, and Publishing Requirements for Industry Standards Metadata (Prism), developed by magazines and their suppliers for that segment of the publishing industry.

There are several XML initiatives related to corporate, financial and market data, such as Xbrl, for corporate reports and regulatory filings, among other things; Fpml (derivatives); Rixml (research information), and recently Mdml, all under the auspices of the Fisd. The challenge to synchronize these standards rests collectively with Fisd and the organizations and individuals involved. Dow Jones, Reuters and the Associated Press are all members of IPTC and Fisd. Dow Jones and Reuters are contributing to Xbrl; Reuters also in Rixml.

Publishers have to get financial data from different sources to get complete information that companies need. Each provider -- such as Reuters, Dow Jones and the Associated Press -- has its own small niche of information in which it specializes or is a provider. For certain classes of financial data, both financial and news executives are trying to get together to agree on interchange formats so that they can ship that information back and forth in a practical way. Yet, not every single financial table is getting tackled in all the efforts that are going on now.

Dow donates DTD
A large working group of the Fisd members has produced the Mddl specification, which was originally developed by Dow Jones and has been in production for almost two years at the Wall Street Journal's web site, WSJ.com. Three working groups were created -- the steering committee, responsible for overall strategy; the technical committee, responsible for structural design and the data-type definition (DTD) architecture, and the vocabulary committee, responsible for definition of the data elements to be included, as well as the semantics and terminology.

Kevin Roche, news systems manager and head of the XML group at Dow Jones, said that XML isn't exactly something that industry groups extend, but rather a technology they adopt and build on to meet specific needs. "XML is a meta-language," he said, "a set of rules and guidelines for building a markup language tailored to a particular domain, such as industrial parts or academia."

The main benefit of Mddl will be easier exchange of data between various players in the markets -- exchanges, banks, brokerage firms, real-time data suppliers and publishers. Roche said, "There is always a risk, or perhaps just resistance, in adopting new technology and adjusting the concept of rich data markup. But it seems clear that the advantages far outweigh the problems."

Stuart Myles, technical manager of WSJ.com, said that the work of Dow Jones and Fisd members fits with NewsML, which is a content wrapper, not a way of encoding content itself. "So, you could have a NewsML document that contains, say, a news article in NIFT about the market plus some Mddl, which offers a snapshot of various stock prices or indices."

According to Myles, Dow Jones decided to donate its Mdml Dtd to Fisd because the company heard that the consortium was working in this area, and it felt it would be helpful for the consortium to have a working DTD, used in real applications. Myles said, "We developed Mdml for our own business reasons, so it doesn't fit Fisd's spec exactly." However, he said, "we feel that there is tremendous interest in encoding market data using XML, and we have developed a simple, flexible and powerful means of doing so."

Dow Jones has already implemented Mdml in a number of systems, as have some of its business partners. "There seem to be a lot of companies -- such as Microsoft, Bridge, Reuters, the Associated Press, Lipper, Dow Jones, Factiva, Merrill Lynch and Goldman Sachs -- who are interested in something like Mdml -- an XML market data markup language," said Myles.

The rationale for Mddl/Mdml is this: facilitation of straight-through processing, ease of mapping between market data applications, and enhanced market data product functionality.

There are currently a number of independent groups working on niche areas in the financial domain, such as the aforementioned Xbrl, Irml and Fixml. However, there is no current initiative dealing with the product-price dimension of market data, which accounts for a significant portion of data passed around within firms.

Master of its domain
This is the domain of Mddl. The initial focus is on end-of-day and intra-day snapshot applications that can be extended into real-time and into more interpretative data models if appropriate. There are likely to be three stages of "vocabulary" work: identification of the broad domain (equities, currencies), identification of the types within that domain (common terms preferred for stocks, commercial or floating rate for currencies) and lastly, identification of the specific data points with the domains and types (open-high-low for equities, unit per U.S. dollar for currencies and the detail to be included at the data dictionary level).

Alan Karben, vice president for product development at ScreamingMedia Inc. of New York City, was also involved in the initial development of Mdml while working at Dow Jones two years ago. He said that Mdml solves one specific class of problems with regard to financial data interchange, and Xbrl handles a different type of problem, reporting companies' quarterly earnings.

Those earnings are "very valuable information that you'd love to have done in a standard way, because you want to compare how one company's earnings results compare with how that company did last year or how other companies did. You'd want that info in a standard format. Xbrl tries to achieve that," Karben said.

Karben's theory is that markup languages are designed to save the time of figuring out the differences oneself and having to manually make the changes. "XML doesn't actually tell you how to markup any particular type of content. XML teaches you how to write vocabularies for particular applications."

Karben said, "Let's say I have a portfolio and I own 50 shares of Microsoft, which I purchased on March 7 at $40.12. That's the basic problem that Mdml is trying to solve -- is it comprehensive enough to be useful to each company? It's useful to a certain classification of business that requires specific information," he said. It's useful for stock quotes, portfolio lookups and the like.

The way financial data gets shipped today is in little preformatted blocks of text. Karben said an example would be a financial report delivered in e-mail and all the columns are aligned, but the only reason it looks like a table is because it's in a monospaced font and the numbers line up vertically.

"That's fine for a human to look at," he said, "but you can't really present that very nicely and you can't also do sorting and searching on that sort of data. So it's much more useful to have the content tagged in a way that describes what the content is about as opposed to just preformatted in a table so that it's easy for a human to read."

Accounting information
The second acronym up for discussion is Xbrl, which will be used to create, exchange and analyze financial reporting information including regulatory filings such as annual and quarterly financial statements, general ledger information and audit schedules. The American Institute of Certified Public Accountants, in conjunction with Reuters and about 30 other organizations, has created this XML specification, formerly known as Xfrml.

"The goal," said Roche, "is that eventually, all publicly traded companies and business entities will use Xbrl for a wide variety of internal and external reports, as well as governmental filings."

Xbrl has a different focus than Mddl/Mdml -- it is an accounting language, focused on business reports. Karben said, "If you are somebody who wants to publish company earnings reports, like ScreamingMedia ... then you would love for all the companies to get behind one big standard like Xbrl."

Zack Coffin, chair of the Xbrl Liaison Committee and program leader at Kpmg Digital Media Institute, said the group is focused on three things -- membership, liaisons and adoption.

In short, Xbrl can be used by every company and in every industry around the world and Coffin believes it will be because every organization wants more timely and accurate reporting for its own internal analysis. "Xbrl marks the first attempt ever to translate financial reporting into an Internet language," he said.

Coffin said that the specification as released will make it easier for undervalued companies to get the attention "they deserve" from capital markets. For users of the information -- individual investors and institutions alike -- he said this means an "easier and more objective" way to identify business performers. "Couple these two together with Xbrl and you have one of the most significant developments in financial reporting since the birth of the Internet," Coffin said.

"What we see is that our clients need simpler and more productive ways to manage their increasingly complex information," said Coffin. Xbrl is "just that -- it brings the benefits of the Internet to business and financial reporting." Both Kpmg and Kpmg Consulting are leading members of this initiative -- serving as chair of the accounting, technology, and liaison committees -- to ensure that their clients worldwide can have the best information possible to form business decisions.

Coffin said that the group's intent for Xbrl is to have it integrate seamlessly with other XML standards in the "entire business reporting supply chain," such as NewsML, Prism, Irml and Rixml. This flow of information from company, auditor and intermediary to end-users, analysts and investor, said Coffin, will become far more efficient, and as Xbrl is at the beginning of this process, it's critical that it is widely supported and adopted.

"Thankfully, Xbrl is exploding," Coffin said. "In one year, it's grown from an initiative by a dozen companies in one country, into a consortium with 85 companies on five continents totaling over $2 trillion in market capitalization. And I'm confident the membership -- all of whom have committed to be early adopters -- will more than double in the next six months."

Xbrl, as it stands now, is usable by about two-thirds of the companies in the United States. So what are the pitfalls? It won't meet the requirements of banking companies, for example, because banks report differently than manufacturing companies. Additionally, it won't meet the requirements of a company in Germany, for example, that is reporting under German rules.

To solve some of these problems, 13 new sets of controlled terms (known as "taxonomies") are currently in progress -- one for Singapore, one for regulatory reports for banks in the United States, one for international accounting standards, to name a few.

Although the need for XML extensions continues to grow, the latest markup languages are just beginning to tackle all the requirements of the industry. Karben says that no single markup language is able to cover all data to serve everyone individually.

-- Aimee Beck, abeck@colepapers.net

The Associated Press: http://www.ap.org/

Dow Jones & Co. Inc.: http://www.dowjones.com/

Financial Information Services Division of the Software and Information Industry Association: http://www.fisd.net/

International Press Telecommunications Council: http://www.IPTC.org/

Publishing Requirements for Industry Standards Metadata: http://www.prismstandard.org/

Reuters: http://www.reuters.com/

ScreamingMedia Inc.: http://www.screamingmedia.com/

From THE COLE PAPERS, February 2001, Copyright © 2001, All Rights Reserved.