XBRL reporting was first mandated by the SEC in June 2009, with a gradual phase-in period lasting 3 years. The object of the regulation was to improve comparability of financial information across all public companies, making information more accessible and useful for investors.
But has that really happened? A recently released CEASA study reports that less than 10% of investors find XBRL data useful. Investors’ concerns mostly fall into two categories: a high rate of data error, and a large number of company-specific taxonomies, or “extensions”. Lack of audit requirements for XBRL was also cited as one of the limitations. The picture was even worse for footnote tagging, where inconsistencies are even more widespread.
Simple errors — incorrect totals, omitted thousands, missed data points — have simple solutions: better analytical tools will catch those errors and improve accuracy. But not all the difficulties are as easily dealt with. Company-specific extensions, while some may seem redundant, in many cases are used because the standard taxonomy is either not clear or not adequate to capture the essence of a company-specific transaction. To make things worse, some companies change their taxonomies from report to report, which makes analysis or comparison between several reports particularly difficult.
There is also a growing concern that, as smaller reporting companies enter the XBRL filers circle, error rates will increase. Smaller companies normally have fewer resources, and tend to see XBRL requirements as additional burden. In a recently released FERF survey, XBRL is cited as the biggest bottleneck in the SEC reporting function across the board. It has been also noted that many companies are dissatisfied with their XBRL filing providers, and plan to in-source these services.
It’s still too early to tell how widespread XBRL-based analytical tools are going to be. In addition to the limitations discussed above, many types of analysis require at least five years of financial information, while XBRL so far provides only three. And although XBRL solves some problems, it still fails to resolve some comparability issues that stem from changes to accounting principles and implementations. Even with XBRL, such filings still require manual analysis.
On top of all that, the fact remains that the most interesting data is sometimes the hardest to standardize. Most interesting classes of analytical data come from the less-structured parts of SEC filings (e.g. MD&A and Proxies), which tend to be exempt from XBRL requirements. There are efforts to extend XBRL requirements to these elements as well — but given the questionable success of XBRL even for the most structured data sources, it’s unclear how successful these efforts will be.
For the foreseeable future, chances are good that XBRL will remain more or less what it already is: a supplemental tool, but hardly a replacement for more traditional analytical resources.