To mark the fourth anniversary of the Securities and Exchange Commission’s (SEC) interactive data mandate, Columbia Business School (my alma mater) and its Center for Excellence in Accounting and Security Analysis (CEASA) published a review of the current state of eXtensible Business Reporting Language (XBRL) that notes the manifold issues that plague this promising technology. From its perspective, three key issues hamper greater use of XBRL. The first is the high error rate in the tagging process and the tendency of companies to use too many non-standard tags, both of which substantially reduce the usefulness of the data to practitioners. Second, they believe technologists, not regulators and accountants, should be more involved in developing software that makes it easier to consume XBRL-tagged data. Third, companies should spend more effort improving the quality of their data than on trying to kill the mandate.
I agree with all three points, but I arrive at slightly different recommendations. Chiefly, people have been slow to pick up on using XBRL because there’s no free, easy way to make use of the technology. Consequently, too few people are aware it exists, and even those who do may not care because there’s no way for them to benefit from it. There’s a bit of a chicken-and-egg situation here; without software to consume XBRL-tagged data usefully, there’s limited demand for consuming XBRL-tagged data, which in turn limits demand for software that works interactively with XBRL-tagged data. One part of the solution is for companies to stop complaining about the tagging requirement and create value from this mandate by reusing and repurposing tagged data on their investor websites. XBRL and other existing tools can create electronic documents that are snazzier than any glossy paper-based annual report could ever hope to be.
As for the first point raised in CEASA’s paper, data quality is a problem, but it’s partly because there’s a learning curve involved and partly because securities analysts have not been as involved in the issue as they should have been. There are two issues here: tagging errors and the use of non-standard XBRL tags. Errors are currently running at about a 1 percent rate. Depending on your perspective, this is either surprisingly low or completely unacceptable. I’m in the latter camp. Applying a six-sigma standard for accuracy would mean only a dozen or so SEC filers would have a single tagging error in any given year across their multiple periodic filings. This may be too much to hope for, but a 0.3 percent error rate (that is, three standard deviations) is not unreasonable. Transgressors could be fined, as the paper suggests, but other factors and measures may accomplish the same thing, such as a published table identifying companies with tagging errors. I also expect that if companies believed that more investors cared about the accuracy of their tags they would be more careful.
The lack of comparability of tags used by companies is another data quality issue. Minimizing the use of custom tags is a must in this respect. There’s a relatively easy fix for this issue because it’s likely that out of ignorance or misplaced priorities the people who are doing the tagging are opting for a custom tag if the financial statement wording is even slightly different from the off-the-shelf XBRL caption that is functionally equivalent. This could have been prevented and may be remedied by greater involvement by professional investors in helping companies determine the proper tags. The CFA Institute could be more active in organizing securities analysts who can help define how companies in specific industries maximize commonality in how they report and tag their results to achieve the highest degree of comparability within and perhaps even across industries. This would not be a one-time event but an ongoing effort. There’s a precedent for this. Beginning in the early 1970s, the SEC mandated something called full disclosure, which was the first major step in expanding the breadth of information available to all investors in US-registered securities. Analysts (often represented by industry-specific splinter groups of the New York Society of Securities Analysts) were heavily involved in creating business segment reporting guidelines.
Yet it’s worth noting that there’s also a limit to the degree of comparability that can be achieved even when companies use identical XBRL tags, so buy-side investors may not always choose XBRL-tagged financial statement numbers over information provided by investment data providers such as FactSet and Standard & Poor’s. Regardless of how a taxonomy is structured, there is no requirement in Anglo-Saxon accounting for uniformity in presentation. Thus, Company X might include some things and exclude others in a captioned item on its income statement or balance sheet that Company Y treats differently. A simple, common example is that some companies report selling, general and administrative expenses as a single item while other break out sales expense on a separate line. Investment data providers massage reported numbers so that it’s possible to compare X and Y side by side, which can be great for portfolio managers but not for Wall Street analysts and anyone else who needs the as-reported figures.
A second important reason for limited take-up of XBRL-tagged data to date is the lack of useful free tools available for consuming XBRL-tagged information. The original plan was for the SEC to fund the development of an interactive data viewer for its EDGAR public company data website, but what’s there is not terribly interactive. When the XBRL mandate was under discussion, I assumed that ad-supported finance sites such as Yahoo Finance might fund the development of an XBRL viewer, but it hasn’t happened. I expected retail investor-focused online brokers such as Fidelity and Charles Schwab would give their clients access to a free tool. Not yet. I also expected Google to step up, but no. XBRL International is fostering competitions to encourage the development of such tools. I reviewed the winners of last year’s competition, and while I think they represent a good first effort, much more is needed.
Creating a good, full-featured interactive viewer is a difficult and expensive undertaking. It’s unrealistic to expect large enterprise software companies to jump into the professional viewer market because it’s a pretty small niche. The problem with free XBRL viewer software is that there’s no money in it for the developer. This is why XBRL International has been sponsoring the competition. It might be possible to apply the Adobe Acrobat business model (charge for creating interactive XBRL documents but offer a free viewer), but the universe of companies willing to pay for the creator side is considerably smaller than the universe of companies that publish web documents, so I’m not sure what company would be willing to fund such a project.
As to the Center’s third point, I believe public companies ought to stop complaining about having to tag their SEC filings and begin using their XBRL-tagged data to create interactive investor websites. It’s time for corporations to reimagine external financial reporting using currently available technology. Rather than treating electronic filing as a compliance requirement of little business value, they should use already available technology to repurpose their mandated reports into more effective financial and corporate communications. Companies can expand their mandated interactive financial disclosures into corporate communications that explain their strategy, performance, opportunities, financial position and stewardship. Doing so would naturally extend the glossy annual report into the digital age, harnessing technology to be interactive, using audio and video for impact and providing insightful analytical tools for investors. It’s important to begin now, given the ever larger audience of people who grew up using computers and for whom an electronic document is a more natural medium than paper.
One other point worth noting: It’s also harder to get professional investors working with XBRL when there are still so many holes on the data collection end. For instance, most earnings releases are not yet tagged, which means that one still has to manually enter the data into some spreadsheet to update a company model or create a table quantifying the differences between actual results and projections. This is a time-consuming process for stock analysts in the precious minutes between the initial release and the conference call discussing the results. It’s worse for buy-side analysts who cover many more companies, especially if a company lacks research coverage. Although the XBRL taxonomy for US-GAAP is well-developed (and in my judgment the most comprehensive in the world), it still needs industry-specific measures (such as revenue passenger kilometers for airlines or kilowatt hours produced for utilities), and it would be nice if one could grab company-specific product line data (for instance, revenues of specific drugs for a pharmaceutical company) as a tagged item. There’s also a threat to the use of XBRL in public company reporting should there be full adoption of the IFRS accounting standard by the United States. The US-GAAP taxonomy is well-developed but the IFRS one is not. I suspect one reason that’s the case is that the IFRS governing body fundamentally does not understand the difference between a taxonomy and an accounting standard.
I believe XBRL has a bright future, but CEASA is correct in pointing to serious shortcomings in how it’s been deployed in the service of investors. Solutions to the issues require effort and imagination, but they should not be too difficult to implement. To date there have been more words than real momentum behind making XBRL a practical tool for investors. I expect that is going to change over the next several years.
Robert Kugel – SVP Research