Machine Learning and Artificial Intelligence Increasingly Part of SEC Oversight
Artificial intelligence may or may not be here yet, depending on one’s definition of AI. There is little doubt, however, that machine learning – the analysis of data by machines without direct human direction – has already found a home at the SEC and is growing.
"The success of today’s new technology depends on the machine readability of decision-relevant information," said agency Division of Economic and Risk Analysis (DERA) deputy chief economist and deputy director Scott Bauguess, who gave the keynote speech May 3 at the Financial Information Management Association conference in Boston. He also used his address to dispel what he described as five myths about machine-readable reporting standards.
Machine readability today is not just for numerical data, "but for all types of information," he said. "This includes narrative disclosures and analyses found in the written word. It also includes contextual information about the information, or data about the data [emphasis Bauguess], often referred to as ‘metadata.’ Today’s advanced machine learning methods are able to draw incredibly valuable insights from these types of information, but only when it is made available in formats that allow for large-scale ingestion in a timely and efficient manner."
The upshot of all this, from a machine learning perspective, is that data, when standardized, "can be combined with other relevant financial information and market participant actions to establish patterns that may warrant further inquiry," Bauguess said. "And that can ultimately lead to predictions about potential future registrant behavior. These are precisely the types of algorithms that staff in DERA are currently developing."
Sounds a bit ominous, but does this constitute artificial intelligence? The answer to that depends on a person’s perspective, said Shearman & Sterling partner Nathan Greene. It may not be at a point yet where a computer program reasons, argues and sounds like a human being, but programs looking for data patterns and making judgments on those patterns are more than mere number crunching.
Large financial firms, such as BlackRock, are already making forays into developing their own forms of inhouse artificial intelligence. Chief compliance officers at firms that make similar artificial intelligence inroads need to ask themselves a number of questions, Greene said. "These might include, ‘Will this make a CCO’s job easier or harder?’ and ‘Will the program draw on inhouse or other data that it shouldn’t?’"
Bauguess "has been a thoughtful advocate and contributor to innovation at the SEC and we would all do well to pay heed to his most recent speech," said Willkie Farr partner and former SEC deputy chief of staff John Burns. "Remains to be seen, of course, but I would mark his words as foreshadowing where the agency will be making investments down the road."
Bauguess’ speech follows up on another speech he made last year, in which he discussed the SEC’s use of data analytics and the increasing role that both machine learning and artificial intelligence were playing (ACA Insight, 7/24/17).
In recent years, "RegTech" and "SupTech," short for, respectively, "Regulatory Technology" and Supervisory Technology," have emerged, Bauguess said. "Each uses machine learning methods to lessen the burden of either complying with or supervising a wide range of regulatory requirements in financial markets. And while neither has reached maturity, both offer significant promise by way of improved market functioning and increased operational efficiencies."
The SEC has been moving toward the use of standardized data when it requires information from registrants. Standardized data allows machines to read the data. "The first rule mandating a machine-readable disclosure dates back to 2003," he said, "and more than a dozen other rules requiring structured disclosure have been proposed or adopted since then."
"The key innovation of our developing disclosure technology is making machine accessibility invisible to the rendering of a document for human readability," Bauguess said. He said this was the case with a recently proposed rule that would require SEC reporting companies to file their periodic reports in an Inline eXtensible Business Reporting Language (XBRL) format. "Currently, filers separately report a human-readable html version of a periodic report and a machine-readable version in an [XBRL] format," he said. "The proposed rule, if adopted, would combine the two requirements and create a single document designed to be read equally well by humans and machines."
"From a machine learning perspective, the financial statement data, footnotes and other key information contained in an Inline-XBRL filing can be easily and automatically extracted, processed and combined with similar data from other 10-K filings," he said.
"Sophisticated algorithms depend on this data being of high quality and being machine readable," Bauguess said. "When applied to the emerging fields of SupTech and RegTech, there is tremendous potential for enhanced regulatory compliance."
Burns noted that "the SEC has found ways to extend structured data requirements to managers (through the XML format mandated for Form N-MFP and Form 13F filings, for example) or offer the option to submit certain filings in XBRL format (such as risk/return summary information). It will be interesting to see to what extent other current or future data submissions are set in these formats."
Bauguess, in his speech, said that he has found that there are "common perceptions about data and information access that are misguided, or even wrong." He named five, which were:
Electronic access is equivalent to machine readability. "It is often assumed that if a document is electronically accessible, then it must also be machine readable," Bauguess said. "This is not true." The problem, he suggested, is because many take the term "electronic access" to mean that something is "digitally" accessible. "But just because a document can be downloaded over the internet does not mean that it can be ingested by a computer algorithm," he said. "A document stored in an electronic format, and available for download over the internet, can be impenetrable to machine processing." This is true particularly if the document is scanned, stored in a proprietary format, or is guarded by security settings. "For advanced machine learning algorithms to generate unique insights, there must be structure to the information being read."
The Commission alone develops the reporting standards incorporated in its rules. The National Technology Transfer and Advancement Act, Bauguess said, requires federal agencies to use technical standards developed by voluntary consensus standards bodies. "We borrow from standards developed and/or endorsed by external groups, whenever possible." That’s what the SEC did when it adopted XBRL for financial statement reporting in 2009, "which is an open standard format that is widely available to the public royalty-free at no cost," he said. "The standard originated from an [American Institute of Certified Public Accountants] initiative and was ultimately given its own organizational standing – XBRL International – that now has more than 600 members. And XBRL is now in use in more than 60 countries."
Retail investors don’t need machine-readable data. "It is an unfortunate but common refrain among some market observers that the average retail investor does not benefit from structured data disclosures, such as those made using XBRL," Bauguess said. This belief misses the point that structured disclosures enable third-party vendors to make the information available to retail investors at low or even no cost. "Machine-readable disclosures fuel many online financial tools popular with investors," he said. "So while it may be trued that many investors do not directly use structured data, the fact is that they do consume the data downstream. Such access would be impossible without structured data. This is particularly true for smaller SEC reporting companies."
Requiring machine-readable reporting standards ensures high-quality data. This also is not true, Bauguess said. "Despite claims to the contrary, computer algorithms can’t fix poorly reported data; they can only maximize its usefulness. Unless reporting entities comply with both the letter and the spirit of promulgated reporting requirements, a well-designed standard may still be insufficient for today’s advanced analytics to generate unique insights about market behaviors."
We don’t need the public’s views any more. This tends to come from those who know data the best, he said. They "often just assume that we know their views and will do the ‘right’ thing when it comes to implementing new reporting requirements." The fact, however, is that it is "vital" for the SEC to hear "from consumers of data, from the experts who know best how the data can be used. Because while we have considerable inhouse expertise, there is no substitute for hearing directly from the public."