Written by Sunir Kapoor Posted on December 7, 2009
Sunir Kapoor is the President and Chief Executive Officer of UBmatrix. He is also currently serving as a member of the XBRL US board. Sunir previously held executive roles at Oracle, Microsoft, Novell, and a number of Silicon Valley startups.
For all the energy expended today on XBRL document creation by filers, the market tends to focus on data “tagging,” losing focus on XBRL’s other unique aspects. If XBRL were just about tagging or “bar coding” data, many other tagging approaches/technologies/standards in the market would deserve our consideration. What actually sets XBRL apart, though, is the wealth of semantic information that can be associated with tagged data, and the ability this provides to automatically process that information.
XBRL’s distinguishing benefit is that it enables the flow of information that is syntactically correct, complete, accurate, and consistent. This empowers XBRL documents’ creators and consumers to leverage that information to reap the full benefits of XBRL.
Today, let’s focus on one of XBRL’s key features: validation.
Everyone in the reporting information supply chain benefits from the validation enabled by XBRL’s calculations and formulas semantics. XBRL’s validation reduces the number of errors introduced into this reporting process, helping address the classic “Garbage In / Garbage Out” challenge. For example, filers can use XBRL calculations and formulas to validate their filings before submission.
XBRL validations can take various forms. The Calculation Linkbase allows filers to check their XBRL documents for simple math errors. The Formula Linkbase allows the filer to validate their XBRL documents against a set of business rules — and XBRL is extremely powerful in its ability to support a wide range of such rules. Examples include checking whether calculations are within a given tolerance; checking ratios amongst reporting items and raising a flag if the ratio is out of norm; and checking for a valid filing, for example, if concept A is reported, concept B must be reported.
To allow everyone in the information supply chain to clearly understand what defines a valid information exchange, XBRL’s Calculation and Formula Linkbases can be shared between the consumer and creator of XBRL documents. Furthermore, because these business rules live in the taxonomy and are not hardcoded into the applications that either create or consume XBRL documents, it is easy and quick to update business rules and communicate these changes to all parties.
Formulas have already demonstrated value in production systems. While everyone talks about the FFIEC/FDIC as among the first successful large-scale projects, what isn’t often discussed is the key role that XBRL formulas played in its success. The XBRL formulas written by the FDIC have allowed banks to pre-validate their filings before submission, with dramatic top-line results. Submissions with correct math have increased from 70% to 100% of filings. The number of “clean” submissions have increased from 66% to 95%. With respect to the SEC’s XBRL mandate, formulas now can be used to check XBRL filings against the EDGAR Filer Manual. Charlie Hoffman has been using formulas to apply additional validation checks against SEC filings.
A current example that highlights how XBRL validation could improve the reporting information supply chain is in recent news regarding American Recovery and Reinvestment Act (ARRA) stimulus reporting in the United States.
To qualify for stimulus money, recipients are required to fill out detailed federal forms, but with 130,000 recipients and 99 data fields to enter, there’s much room for error. Government Accounting Office analysis of the first round of reports from recipients showed 3,978 reports that “showed no dollar amount received or expended but included more than 50,000 jobs created or retained.” The GAO also found more than 9,000 reports that showed no jobs created or retained for more than $1 billion spent. Reports have circled describing money being spent in non-existent congressional districts.
Simple XBRL formulas could have checked the validity of submissions like these — e.g., checking for a correct congressional district — and flag outliers that need review, such as the ratio of money expended vs. jobs created.
The market needs to figure out how to exploit XBRL’s key benefits -- taking advantage of validation, one of its key differentiators. The risk of failing to do so is that XBRL would end up just one among many data exchange technologies, and the information supply change would miss out on sorely needed improvements.
This message on Validation needs to be promoted much more widely as it eradicates many of the anti-arguments put forward by the anti-XBRL lobby, some of whom, paradoxically you might think, are very active in the XBRL community. It always reminds me of the line in the Godfather: keep your friends close, but your enemies closer!
I would argue that instead of "The market needs to figure out how to exploit XBRL’s key benefits" the market needs to be shown how powerful a tool XBRL validation can be
The second thing that is needed is for other software vendors other than UB Matrix and Fujitsu to add the formulas functionality to their XBRL product offering.
Otherwise it will remain a too expensive option
Colm;
I fully agree with your comment.
XBRL formula validation is providing the XBRL key benefit: Automated validation of 1) Correctness, 2) Completeness, 3) Accuracy and 4) Consistency of busines information (financial and non-financial).
I hope to have the opportunity to meet with you - again - soon!
Cordially
Frederic
Colm;
I agree with you also about the benefits of validation, particularly XBRL Formula validation, as being a way to show how powerful XBRL can be. This will come when there is more XBRL instance information available, such as the SEC XBRL filings.
I would point out another often missed benefit of XBRL as being the ability to exchange those XBRL Formulas. What we have seen at UBmatrix has been that people tend to not see the utility of validation because they build all the validation within their software systems or solutions. They assume they must do this. But with rules based validation which is not system specific or a one-to-one validation solution, there is no need to build the validation systems. Users can just buy an off the shelf valiator (such as UBmatrix or Fujitsu as you mentioned) and then just write the rules.
Further, because the validation is not system specific (i.e. you can exchange those rules) there are additional benefits.
It takes time, but people do eventually realize this.