Today the U.S. Securities and Exchange Commission announced it is publishing a database of corporate financial statement information. That announcement is more consequential than it might appear.
In 2009, the SEC started requiring U.S. public companies to report their financial statements in the eXtensible Business Reporting Language (XBRL) structured data format. In XBRL, each line item and each number has a unique electronic tag, which means a structured-data financial statement can be automatically read by software.
The SEC’s adoption of XBRL was one of the U.S. government’s biggest and earliest open data efforts. In fact, it happened so early that the term “open data” wasn’t even in common use yet.
XBRL should have been the start of a complete modernization of the SEC’s disclosure system. It should have transformed the U.S. capital markets. But it wasn’t and it didn’t.
In theory, the SEC’s adoption of structured data should allow investors to make better decisions. It should support analytics tools to help SEC staff review filings and spot potential accounting fraud. And it should allow filers to automate formerly-manual compliance burdens – just as happened in the 1990s when the IRS adopted an XML format for individual tax returns that permitted companies like Intuit to roll out products like TurboTax.
The members of the Data Transparency Coalition have already developed the solutions that can do these things. Platforms like Calcbench can crunch through financial data to help investors find patterns and insights. Products like Contexxia and companies like Information Builders can help agency staff check the math of financial statements and notice potential problems. Disclosure solutions offered by RR Donnelley, RDG Filings, and Workiva can automate compliance for companies.
But the SEC didn’t take the follow-up steps that would have allowed tech companies to deliver all these benefits.
First, the agency made it very hard for investors, and the infomediaries that serve them, to use the structured-data financial statements. Each XBRL financial statement is published on the SEC’s EDGAR website separately. To make proper use of this valuable data, investors and tech companies must assemble thousands of separate files into a database. That complex step became a barrier to entry for casual coders and startup innovators.
Second, the agency continued to collect the old-fashioned document version of each financial statement alongside the new data version. To this day, companies must report the same information twice: once as an XBRL file, and again as a plain-text document. As a result, companies and SEC staff view the XBRL version as an added compliance burden and a supplemental exhibit that is distinct from the truly important, if old-fashioned, disclosure document it accompanies. (To this day, the agency doesn’t use XBRL data to check the mathematics of financial statements. Instead, SEC attorneys and accountants print out the financial statements and use calculators to do that.)
Third, for the first five years, the agency took no concrete steps to enforce the quality of the XBRL financial statements companies filed. As a result, data quality was so bad that investors didn’t trust the XBRL versions and refused to use them. Infomediaries followed their customers’ mistrust and didn’t make much use of the new data resource either. The SEC finally announced its first data quality letter to companies last summer. But it still doesn’t treat errors in the XBRL version the same way it treats errors in the document version. Errors in the document version earn a reprimand from the Division of Corporation Finance. Errors in the data version might result in a phone call from the SEC, a few times in the past five years.
Finally, after adopting XBRL for corporate financial statements, the SEC stopped right there. It didn’t continue its transformation from documents to data. Financial statements are an important piece, but only a piece, of a vast system of corporate disclosure. The SEC collects all sorts of information that could be, but isn’t, electronically searchable. Lists of each company’s subsidiaries, disclosures of each company’s stock structure, names of directors and officers, compensation details, and much more are still collected as plain text – which must be manually reviewed or expensively OCR’d in order to be useful. And without any clarity about the SEC’s future plans, our industry can’t invest in developing the next TurboTax for corporate disclosure.
Without these missing follow-ups, the SEC’s big move five years ago has, so far, been a failure. Investors don’t make much use of XBRL data; SEC staff are still using manual checks instead of software; companies still spend more time and money than they need to on compliance.
And consider the missed opportunity for U.S. capital markets. XBRL should make it cheaper for analysts to follow a wider range of stocks – which means small companies get more attention from investors. But if XBRL isn’t getting used by analysts, it isn’t helping small companies reduce their capital costs.
It isn’t surprising that Congress is frustrated. Some in Congress have called on the SEC to fix data quality and expand the universe of information collected as data. But others, unfortunately, have proposed simply getting rid of data altogether and regressing to document-only disclosures for most companies. Last spring our Coalition called on Congress to fix XBRL, not eliminate most of it.
But today’s announcement addresses the first of the four missing follow-ups. By making the corporate financial data it receives more readily accessible – no database assembly required! – the SEC will facilitate more use by investors, companies, and innovators.
Today’s announcement also provides some hope for the future – both the short term and the long term. The agency really is serious about making data work.
In 2015, the SEC could put itself back on a path to modernization by taking a few more short-term steps: finishing the publication job it started today; exercising better data quality enforcement; adopting inline XBRL; and starting to use the Legal Entity Identifier (LEI). In other words, modernization at the SEC does not need to begin with expensive, expansive IT projects.
Here is what our Coalition is hoping the SEC will do next year – and what might help persuade Congress that the whole XBRL project should be fixed, not mostly ended.
- Publish the rest of its existing XBRL-formatted financial statement data. Today’s announcement covers the “face” financial statements – in other words, companies’ balance sheets and income statements will now be available as a consolidated database. Today’s announcement promises that the footnotes, which are also filed in XBRL, will be published as a consolidated database in 2015. To finish the job it started today, the SEC should do this.
- Switch from documents-plus-XBRL to inline XBRL.
The SEC has already developed the necessary technology to stop collecting a document version plus an XBRL data version of each financial statement. A format called “inline XBRL,” or iXBRL, would allow companies to submit a single version, both human-readable and machine-readable. The SEC’s staff has already prepared a draft rule to replace documents-plus-XBRL reporting with iXBRL reporting; filing vendors are ready to support this change at no cost to registrants; Financial Executives International (FEI) supports iXBRL; and Chair White has confirmed she is considering the change. But until the SEC acts, public companies continue to file two versions of every financial statement. The need to confirm that these two versions agree creates additional compliance costs for filers.
- Better enforce the quality of existing XBRL-formatted financial statements.
Until July 2014, the SEC had taken no action to enforce the data quality of XBRL-formatted financial statements. As a result, investors and markets do not trust, and have been slow to begin using, XBRL data. In July 2014, the SEC issued a Dear CFO letter outlining common XBRL errors and informally contacted a few companies whose XBRL submissions included a large number of errors. But until the SEC incorporates XBRL data quality into its regular review process, and addresses errors as part of comment letters from the Division of Corporation Finance, data quality will remain questionable.
- Adopt the LEI for Exhibit 21 to the 10-K.
The Legal Entity Identifier (LEI) has been adopted by financial regulators in over 60 countries, plus the Commodity Futures Trading Commission (CFTC), to identify regulated entities and their parent entities. The U.S. Treasury Department has called for all U.S. regulators to use the LEI instead of their existing non-compatible identifier codes. Universal use of the LEI would allow investors and tech companies to automatically match an SEC registrant’s data with information filed by the same entity with other regulators. It will take time to replace legacy identification numbers with the LEI across the SEC’s hundreds of forms. However, one immediate change would save investors a great deal of time and money: incorporating the LEI in the existing Exhibit 21 to the annual report on Form 10-K, the list of subsidiaries. Exhibit 21 is currently submitted as plain text. Variations in company names make it nearly impossible for software to match Exhibit 21 with other data reliably. Replacing the plain text list with a simple electronic form, and requiring the LEI of each subsidiary, would instantly allow software to automatically match every public company’s subsidiaries with other records. This simple step would also be a first move toward the total modernization that the disclosure system desperately needs.
Over the long term, of course, there is much more work to be done. The SEC needs to embrace a total transformation from documents to data throughout its disclosure system. Standardized data is better for investors, agency staff, and registrants. If the SEC switched its disclosure requirements from documents to data, investors could use the data for better decisions, agency staff could deploy Big Data analytics to find errors and potential fraud, and registrants could automate their compliance tasks. The SEC should adopt consistent data formats for all existing disclosures and consistent identifier codes for public companies, firms, officers/directors, and other entities.
But we aren’t asking for all that to happen right away – or even in 2015. If the SEC is serious about open data, these four steps in 2015 will put it on the road to total modernization.