Q&A with an expert: Mike Willis, Assistant Director, SEC Office of Structured Disclosure
NOTE: The Securities and Exchange Commission disclaims responsibility for any private publication or statement of any SEC employee or Commissioner. This interview expresses the views of Mike Willis and does not necessarily reflect those of the Commission, the Commissioners, or other members of the staff.
Mike Willis is the Assistant Director in the SEC’s Office of Structured Disclosure (OSD). Among other things, the OSD works on the design and implementation of technological processes and tools that support the SEC’s use of structured data, such as XBRL-tagged financial information. Mr. Willis joined the SEC in 2015 after retiring as a partner at PricewaterhouseCoopers, where he worked on process enhancements involving standards-based architectures, both within the firm and for clients. He has also held several positions at XBRL International, where he has been at various times the founding chair, chair of its steering committee, and chair of its board of directors.
Dimensions, Merrill's newsletter for financial reporting, SEC disclosure and legal professionals, interviewed Mr. Willis for insights into the SEC’s growing use of the XBRL-tagged data submitted in corporate financial disclosures and the SEC’s other initiatives involving XBRL and structured data.
You head one of the specialized offices within the SEC’s Division of Economic Research and Analysis (DERA). Please describe the role of your office.
I have the pleasure of working with a very talented group of professionals: data scientists, CPAs, developers, and attorneys—a very diverse set of skills. We work to support the SEC’s efforts in making disclosures accessible and easy to use. This means that we work closely with and support other divisions and offices within the Commission in fulfilling their respective roles and responsibilities. We often lead the design and implementation of technological processes and tools for the various structured data needs of the Commission, including the design of taxonomies and schemas, validation rules, and analytical tools.
We try to be involved in the rulemaking process as early as possible to provide counsel and support on when and how structuring approaches can enhance the accessibility and usability of required disclosures, how various structuring approaches can be most efficient for filers, which requirements would help to enhance the usability and data quality of the disclosures, and how the structured disclosures can be reused for various analytics.
We also perform risk assessments via proactive, persistent monitoring and ad hoc analysis. Lastly, we engage accounting standards boards, such as Financial Accounting Standards Board and International Accounting Standards Board, and consensus market-based standards bodies, such as XBRL International and Global Legal Entity Identifier Foundation, on various structured-disclosure topics. Overall, we are working to enhance the analytical insights available to the Commission, investors, analysts, and other market participants through the increased use of structured disclosure.
You have certain goals you want to accomplish in your position at the SEC. How would you grade your progress on them at this point, after two years at the SEC?
There are some initiatives that we have worked on these last two years that are worthy of particular mention, including: the exemptive order to allow Inline XBRL filings on a voluntary basis; the proposing release on Inline XBRL, which has the potential to improve data quality; the notice that the IFRS Taxonomy is available for use by foreign private issuers when submitting their financial statements in XBRL; the posting of the Financial Statement and Notes Data Sets (which we should come back to later in the interview).
We developed risk assessment and query applications used by Commission staff, such as: the Corporate Issuers Risk Assessment (CIRA) Dashboard, which enables search queries and comparisons of XBRL data that allow for more comprehensive and timely oversight; and the Financial Statement Query Viewer (FSQV), which enables staff to search, review, and compare structured financial statement and footnote disclosures.
Meanwhile, we continue to publish staff observations, FAQs, and data-quality reminders to help filers with their XBRL submissions. That said, there is a lot more to accomplish and other topics to address—so please tell your readers to stay tuned.
When we interviewed you for the December 2015 issue of Dimensions, you praised the impressive commitment at the SEC to XBRL, which you had not expected before you started working there. How has that commitment changed or expanded?
I arrived with an initial perception that no one was using XBRL data at the Commission. What I found was a surprising number of technical experts using XBRL for a broad range of analytical applications. Certain Commission staff were leveraging the XBRL structured data, as there was no other source of “as filed” data for all registrants.
Those early adopters revealed the analytical benefits (as well as the data-quality issues) of the XBRL structured data, and their early use of the XBRL data helped spur development of query and analytical applications for use by our attorneys, accountants, and analysts in their daily reviews and research. As a result, enterprise-level tools, such as CIRA, have been developed and are being deployed for Commission staff to leverage in their work.
The CIRA Dashboard is used by Commission staff. CIRA enables search queries and comparisons of XBRL and other relevant data that allow for more comprehensive and timely analysis. It is useful for both reviewing disclosures and detecting possible financial-reporting anomalies, which, if corroborated with other evidence, could lead to additional inquiry. As Andrew Ceresney, former director of the SEC Division of Enforcement, noted in 2016:
CIRA provides us with a comprehensive overview of the financial reporting environment of Commission registrants and assists our staff in detecting anomalous patterns in financial statements that may warrant additional inquiry. CIRA’s multiple dashboards enable the staff to compare a specific company to its peers in order to detect abnormal, relative results, focus on particular financial reporting anomalies, and generate lists of companies that meet the criteria for further analysis. And, because this is a homegrown tool, we can customize the tool as needed, setting and refining different breakpoints, drilling down on particular corporate events.
Which divisions of the SEC are actively using XBRL and structured data?
All the divisions use structured data, but perhaps in more ways than I am aware of. Here are a few examples, in addition to what we’ve already discussed. The Division of Economic and Risk Analysis uses XBRL data for risk assessment, economic analyses, white papers, research, and consultations. The Office of the Chief Accountant uses the XBRL data for assessing disclosure scenarios across all filers or a targeted filer segment. The Office of Credit Ratings is developing analytics to identify trends and outliers to complement their exam and monitoring responsibilities with a database of credit-rating histories formatted in XBRL.
Any examples to show how quick, easy access to XBRL data is proving beneficial to the SEC?
This is a growing list, as Commission staff is rapidly increasing its use of XBRL-enabled analytical applications. Here are a few examples of the ways the staff uses XBRL:
- Looking for all filers who were early adopters for a particular accounting standard; staff can query the related elements and reveal, in a few mouse clicks, a listing of filers with early-adoption disclosures.
- Looking for a specific combination of disclosures that may reveal a risk pattern; staff can find those filers quickly with the targeted disclosure combination.
- Comparing disclosure and specific sector risk profiles across targeted filers.
- Seeking to aggregate a specific disclosure across all filers for a target period; staff can query that disclosure element and have a result in a few seconds.
- Looking for narrative sentiments that are misaligned with the numeric results and ratios; staff can see that via text analytics and sentiment analysis of the structured numeric and narrative disclosures.
- Gathering statistics or trends on a specific financial disclosure such as net deferred tax assets (liabilities) or income-tax expense.
- Performing data-quality assessment and searching for issues such as incorrect tagging, use of inappropriate extensions, and scaling errors.
The errors in XBRL filings seem to be persistent concerns. Have you witnessed improvements in the quality of XBRL filings?
Yes, we observed improvements in data quality after the Division of Corporation Finance issued the “Dear CFO” letters in July 2014, an observation confirmed by XBRL-data aggregators and analytical vendors. Further, Emil Efthimides from Bloomberg discussed in a July 2017 Dimensions article that data quality has improved over time.That said, there is more to do to improve data quality and address some of the ongoing data-quality issues that unfortunately continue to occur, such as negative values, scaling, inconsistent tagging, incorrect element selection, inappropriate extensions, and incorrect fiscal years.
What efforts is the SEC making to monitor and improve XBRL quality by filers?
DERA is leveraging the internal analytical capabilities I previously mentioned to enhance our data-quality assessments.
The increased use and transparency of the “as filed” XBRL structured data can also quickly reveal certain data-quality errors. Negative values, scaling, inconsistent tagging, incorrect element selection, inappropriate extensions, incorrect fiscal years all create anomalies in XBRL data that may distract or mislead reviewers or analysts. As a result, DERA continues to highlight these common data-quality errors with our staff observations, FAQs, and data-quality reminders to help filers improve the quality of their XBRL filings.
The OSD recently (July 20th) issued its second data-quality reminder on negative values. Any observations on why this error keeps occurring? What other common errors continue to impair the ability to use XBRL data?
Our office issues various data-quality reminders for errors that we believe are somewhat common, easy to identify, and easy to fix but that if left unresolved can impair the use and analysis of the XBRL-structured data. The reminders also include references to EDGAR Filing Manual, staff observations, or other documents related to data-quality errors, which filers can review to fix the errors. One example is when filers incorrectly associate credit balances with negative values even though debit and credit balances represent presentation attributes for the HTML document, not the underlying meaning of the XBRL element. Other common errors include filings without level 4 tagging, filers that do not submit XBRL exhibits, filings with scaling issues (e.g., the HTML presentation is in billions but the XBRL value is submitted in millions), inappropriate extensions, incorrect use of a taxonomy element, and filings with the incorrect fiscal year-end.
What general advice do you give filers to improve their XBRL quality?
Filers are strongly encouraged to review their XBRL structured-data filings, specifically for the common errors we’ve already discussed. We advise filers to review relevant accounting guidance, Interactive Data rules, and the US GAAP and IFRS taxonomies before selecting elements. We also encourage filers to review our website for staff observations, FAQs, and data-quality reminders. Our staff members frequently attend conferences and webinars to discuss the topic of XBRL data quality and answer questions from the public. Filers are also encouraged to send any questions on structured data to our email inbox (firstname.lastname@example.org), or call us at 202-551-5494.
You’ve explained that the SEC considers the inclusion of structured-data requirements when adopting any new form types. How does DERA/Office of Structured Disclosure get involved in a rulemaking process to determine the role for structured data and the best potential formats?
As I mentioned at the start of this interview, we try to be involved in the rulemaking process as early as possible. When relevant, we provide counsel and support on when and how structuring approaches can enhance the accessibility and usability of required disclosures, how various structuring approaches can be most efficient for filers, which requirements would help to enhance the usability and data quality of the disclosures, and how the structured disclosures can be reused for various analytics.
Beyond XBRL-tagged data in financial statements, investors and analysts have expressed interest in XBRL-formatted data for earnings releases, proxy-statement executive compensation, and MD&A. What is the SEC’s timeframe for deciding how to broaden the structured-data requirements to these disclosures?
The Commission has received comment letters from various parties that support the expansion of structured data for MD&A, corporate actions, and a broader range of filer disclosures. We in DERA appreciate their interest in XBRL-structured disclosures and will continue to support initiatives by the Commission to broaden the structured data requirements to other disclosures. We encourage the public to send comments on this topic to the Commission so as to help inform policymakers of the market considerations for the increased or decreased use of structured data.
The SEC’s concept release on modernizing financial disclosures in Regulation S-K discusses scaled or layered disclosure that is tailored for different investors. In your view, how can the use of structured data help to make this happen?
In the paper or unstructured electronic filing environment, layered disclosures can result in both duplicate and aggregated disclosures within a single filing. Structured disclosures enable automation of disclosure aggregations as well as rendering of a single disclosure within multiple layers or locations of the document (e.g., “gross revenue” as a disclosure concept is entered once and rendered many times), thereby eliminating some disclosure duplications that commonly occur in the unstructured filing formats.
Further, structured disclosures are easily accessed by a wide range of analytical applications and can be reorganized or aligned to the preferences of the user, whether an investor, an analyst, a peer company, or Commission staff. In substance, XBRL is a supply-chain standardization effort that empowers consumer customization of disclosures into how they view filer disclosures via various adjustments and reclassifications that the consumers perceive to be appropriate.
Supply-chain standardization frequently offers enhanced business practices, such as the now commonly accepted ideas of improved refresh speeds and quality within the grocery supply chain, courtesy of the UPC/bar code. By empowering consumer customization, XBRL provides similar supply-chain standardization for the financial-information supply chain.
We are just beginning to observe this impact in the financial-information supply chain. In a Q&A article in the July 2017 issue of Dimensions, Emil Efthimides at Bloomberg mentions that “the use of structured data enables investors to access the reported data more quickly.” Further, with structured data, investors may dissect the data more easily. For data aggregators, structured data enables them to extract, analyze, and compare data across filers and different periods in a more timely and efficient manner.
How are smaller companies helped (rather than burdened) by XBRL?
With XBRL, smaller companies, like their larger counterparts, are able to make their disclosures immediately available to the public in a structured format. Before XBRL, larger companies had an advantage, as data aggregators and analysts commonly tended to cover them instead of the smaller companies. With XBRL, smaller companies could benefit from increased analyst and investor coverage, because XBRL increases the availability or reduces the cost of collecting and analyzing corporate financial data. As a result, XBRL may reduce some of the information barriers that make it costly for companies to find appropriate sources of external financ[ing], thus lowering their cost of capital and increasing the efficiency of capital formation.
In addition, for smaller companies that may be looking for disclosure best practices and information on peer-group companies, XBRL enables them to have access to those disclosures in a freely available standardized disclosure format.
Third, the standardization and automation enabled by XBRL may be particularly beneficial to smaller companies where resources may be more constrained than with larger filers. Companies with manual aggregation, assembly, and validation processes may find that these manual processes can be significantly automated and thereby enhanced via XBRL standardized taxonomies for both reporting as well as ledgers.
Bottom line: Smaller companies may realize a significant benefit via standardization and related process automation, despite their relatively limited resources.
Smaller companies’ data is crucial, as data aggregators and users have emphasized the need for high-quality data for all filers, regardless of size. We’ve heard from the analysts’ community that they use the XBRL data from large accelerated filers as well as from smaller reporting companies.
With the SEC proposal for companies to use Inline XBRL, what are the expectations for improvements in data quality that will help investors?
As outlined in the proposed rule, the use of Inline XBRL is intended to improve data quality and thereby benefit investors and other market participants and data users, and to decrease, over time, the cost of preparing the data for submission to the Commission.
Separately, the freely available open-source Inline XBRL viewer has a range of filters that can help filers more quickly identify disclosures associated with common data-quality errors. For example, with the Inline XBRL viewer, filers and investors can see how disclosures were tagged, if the values were entered as positive or negative, or if there are any scaling errors.
For XBRL submissions made using Inline XBRL, the Commission’s open source Inline XBRL viewer allows the public to identify disclosures by “topic” rather than the more traditional “word search” approach. This capability enables any user to quickly identify all disclosures for a particular topic (e.g., “stock compensation”) within an Inline XBRL filing. Topic search can improve disclosure-checklist processes for both filers and their internal and external auditors.
What further plans does the SEC have to encourage the use of XBRL data by the public, especially investors and financial analysts?
We make structured data available via EDGAR and RSS Feeds for all XBRL filings. Further, we have been posting Financial Statement Data Sets (FSDS) for the past few years and recently posted Financial Statement and Notes Data Sets (FSNDS).
The posting of the FSNDS is the first time that the “as filed” data for all filers and for all financial-statement disclosures has been aggregated and made freely available in a combined file from any source. This data set provides a unique analytical opportunity for investors, analysts, academics, and other market participants to access and evaluate the comprehensive disclosures from all filers for a broad range of perspectives, including data quality and relationships with indicators of risk, potential fraud, and inappropriate accounting. As a result of the breadth and depth of the FSNDS, we expect that it will eclipse the outstanding popularity of the FSDS in terms of market demand and downloads.
To encourage filer awareness and attention to data quality, we will probably continue publishing staff observations, FAQs, and data-quality reminders. Further, we are working on additional capabilities and features of analytical tools that expand the use of XBRL data.
Does the SEC try to synchronize its activities with the structured-data requirements and usage in other countries and markets?
Information-sharing is key for regulators in understanding the market situation surrounding structured data. Regulators, including the SEC staff, from a broad range of countries participate in the IFRS Taxonomy Consulting Group (ITCG). SEC staff also presents on structured-data topics at XBRL International events. Additionally, SEC staff periodically meets with staff from the European Securities and Markets Authority on a broad range of regulatory matters, including structured data.
What do foreign issuers using IFRS taxonomy now need to know about XBRL filing?
As of March 1, 2017, foreign private issuers that prepare their financial statements in accordance with IFRS and are subject to Rule 405 of Regulation S-T have been able to submit their financial statements in XBRL using the IFRS taxonomy specified on the Commission’s website. They must submit their financial statements this way as of their first annual report on Form 20-F or 40-F for a fiscal period ending on or after December 15, 2017. Filers should be aware that there is no phase-in based on issuer filer status.
We encourage foreign private issuers to visit the IFRS Taxonomy FAQ page on our website.
DERA has a Twitter feed that’s both informative and witty, which is perhaps unusual for a government department. In what other ways does your SEC division have fun?
Your readers are encouraged to follow DERA on Twitter. Interested parties can also subscribe to the Office of Structured Disclosure’s email distribution list.
All SEC staff participate in and support charitable causes, such as Feds Feed Families. Our office staff recently won a pie-eating contest during one of the Division’s challenges to raise awareness for the Feds Feed Families program, as well as a related canned-food-sculpture competition. We also enjoy presenting to other Commission staff about different topics under our office’s expertise, such as Inline XBRL and machine learning. And of course, we love to analyze structured data.
This interview was originally published in the September 2017 issue of Dimensions. Download a copy of the full newsletter.