SEC Experts Interviewed on Structured Data and XBRL Impact on Companies
Do read the interview with Scott Bauguess, Deputy Director and Deputy Chief Economist of the SEC’s Division of Economic and Risk Analysis (DERA) this week from Merrill Corp. Spurred on by what he considered to be a persistent lack of interest and understanding among certain market participants about the importance of standardised data collection and dissemination, his Machine Readability and AI speech at this year’s Financial Information Management Conference caught the eye of many.
Artificial Intelligence and Machine Learning may be two buzzwords that we are becoming ever-more desensitised to. What is often overlooked is that we all assume that the data exists and is readily available to feed these new technologies. In fact, the biggest inhibitor today in the implementation of such services is not the actual analytic methods, or even the computing power. It is the availability of good, clean, high-quality data. What many people unfortunately don’t appreciate is that in most analytics programs, the effort required to get the data into a ready and usable state is about 80% of the overall effort. Once the data is prepared, performing the analytics is comparatively straightforward.
This is where the SEC steps in within the US. Their role in helping to solve some of these problems of making information readily available and reusable for market participants has been fundamentally important. Through these technologies, financial markets now have the opportunity to learn from the disclosures that registrants are providing, using machine learning not just to crunch numerical information but also text-based information in narrative disclosures. Mr. Bauguess highlighted some of the fundamental ways in which XBRL remains necessary, including how the tagging of numerical disclosures can facilitate greater efficiencies within databases. He further advocated for the manner in which XBRL provides value through the tagging of footnotes—both in block tagging and detail tagging. A lot of the important information in a 10-K (an Annual Report) is tucked away in the footnotes and specialised disclosures. The ability to extract those footnotes systematically and organise them appropriately lets the SEC effectively apply new technologies, whether this is machine-learning text analytics on the footnotes or extracting specific pieces of information, such as rates of return on pension liabilities.
Mr. Bauguess also discussed how poor XBRL tagging can be detrimental to both the company, as well as the quality of its XBRL filings. It is hoped that Inline XBRL will help further data quality improvements and decrease compliance costs.