An interview with Scott Bauguess, SEC Division of Economic and Risk Analysis, and Matthew Slavin, SEC Office of Interactive Disclosure
In July 2014, the SEC's Division of Corporation Finance issued a sample of the "Dear CFO" letter the Division had sent to certain public companies, calling attention to omissions of calculation relationships in the companies' XBRL-tagged financial statements. At the same time, SEC staff released observations on the rate of custom tagging. For insights into these SEC actions and other recent XBRL developments, Dimensions spoke by telephone with Scott Bauguess, the Deputy Director and Deputy Chief Economist in the SEC's Division of Economic and Risk Analysis (DERA); and Matthew Slavin, the IT Program Manager in the SEC's Office of Interactive Disclosure. Dr. Bauguess was the acting co-Director and Chief Economist of DERA over the summer, before Mark J. Flannery started as the Division's new Director and Chief Economist in September.
Part 2 of this interview will appear in the December 2014 issue of Dimensions.
What led the SEC to issue the “Dear CFO” letter about the XBRL requirement for calculation relationships?
Mr. Slavin: We cannot comment on what led the Division of Corporation Finance to issue the letter, but DERA certainly monitors XBRL filings as part of our efforts to improve data quality. We certainly believe that the required calculations are helpful to provide context and interpretation of the custom tags that folks use. It also helps reduce the incidence of incorrect numbers. It is an important area.
Of all the issues with XBRL quality, why did you choose calculation relationships first for a “Dear CFO” letter?
Mr. Bauguess: It is an easy topic to comment on, because it is straightforward to measure. We also offered some perspective on the custom-tag rate usage at the same time. In that area, it is a lot harder to say whether or not there was an error, because there is a lot of subjectivity in the use of custom tag rates, but what it did show was that the use of custom tags varies across filers and has changed over time. It is not always just black and white with the quality of XBRL filings.
Mr. Slavin: Whereas with the calculations, if they are missing when they are required, it is much easier to assess that.
Did the SEC seek a balance in issuing both of these at the same time?
Mr. Bauguess: That is one way to look at it. We look at many facets of the filings, and these offered two big issues that we thought were relevant to discuss.
Mr. Slavin: We have mentioned other issues in our past staff observations on XBRL tagging that we have posted on the SEC website. But it was decided these two were the ones to start with. Previously, there have been staff observations listing some of these same issues-and others, such as incorrect signs, scaling issues, and things like that.
Did any companies have to restate their financials because of these or other XBRL errors?
Mr. Slavin: Each company makes its own decision about whether it corrects past filings. But we have seen amendments that list XBRL-related issues as reasons for that amendment.
What XBRL errors bother you the most? Are there recurring mistakes that have been surprisingly persistent?
Mr. Slavin: As part of our efforts to improve data quality, we focus on a lot of these different issues-and in particular, issues that affect data quality are of concern. These include errors such as incorrect values and tag selection (which includes custom tags as well). Anything that affects the use of the data is certainly of interest to us.
Mr. Bauguess: “Bother” seems like a very negative way of saying it. We always recognized, even in our original 2009 release, that there would be some learning that would occur in the market-that the market would need to adjust. When each company makes a filing, there may be some specificity about the information they are trying to communicate, and they may have to decide between a standard tag and one that may communicate different information-it is not a straightforward decision. I think what you see is that, over time, the filing community, particularly large and mid-size companies, is becoming more comfortable with the standard tags and is learning how to integrate and use them. From my perspective, the learning in the market is quite natural, and we are seeing improvements over time in the rate of use for extension tags.
Could XBRL errors ever be egregious enough to involve the SEC's Enforcement Division?
Mr. Bauguess: I do not know of any enforcement cases that have been taken. If there is a misrepresentation in the financial reporting, it will likely permeate many aspects of the filing and I do not know that there has ever been a translation error that has resulted in an enforcement action. But certainly if there were a misrepresentation that manifested in some sort of harm-that is for the Enforcement Division to answer.
Do you expect the SEC to issue more letters to filers about XBRL requirements?
Mr. Bauguess: The Division of Corporation Finance [staff] may take actions at their own discretion (“Dear CFO” letters and the like). At DERA, we intend to continue to publish information and guidance on the quality of XBRL filings and provide updates to what we have observed based on what we have already put out. That is certainly on our radar: ongoing communication.
We would like to accelerate those communications-not take so long between communications. We saw a lot of very positive feedback from the market when we posted some of our analysis a couple of months ago. We would like to do more of that.
So companies can expect to see more public communications about XBRL errors?
Mr. Bauguess: They can expect to see more communications on these types of issues from DERA.
The SEC staff has also issued observations on custom tag rates. What is the concern with custom tags?
Mr. Slavin: They have an impact on the usability of the data for comparative analytics. Whenever you have a custom tag, you need to understand what that custom tag represents. When all filers are using a standard tag, you can compare; but when someone creates a custom tag for something that obviously has a standard tag, that creates a problem because there is no way to compare it as apples to apples. You have to examine it and understand what it is, try to figure it out-and that requires some level of manual intervention. It is harder to perform automated analysis with custom tags. Certainly, custom tags are needed because there are unique items that filers occasionally have, or information specific to a certain industry that perhaps the standard taxonomy does not yet
fully cover. But custom tags are there because US GAAP allows for variety in reporting. The idea is to use custom tags as infrequently as possible-and only when they are absolutely needed.
Mr. Bauguess: What is the optimal level of custom tags for a filing? That is not yet an answerable question. We are still learning that. The big data vendors that sell this information do a standardization or normalization exercise. They have to make assumptions about how to define the items across companies so that they can allow comparability in a database. Now we are asking companies to do that by selecting either a standard tag or a custom tag. So they are making a decision: “If I do not use a standard tag, I get perhaps more precision in what I am communicating (which is good), but maybe that makes it less comparable (which can be bad).” Some of the features in the tools are benchmarking tools. If I am an airline company, I may want to see what other airline companies are doing, and I might want to be compared to them or be like others in my industry. So there is perhaps some convergence in how companies report. Ultimately, I think we will find out empirically what the optimal number of tags is just by observing what the market settles on.
Mr. Slavin: There is an interesting connection between custom tags and calculations. The calculations help someone consuming a custom tag better understand what it is. If you understand how it calculates up to, say, its parent, perhaps total current assets, you will know that this item is at least a current asset. That is another reason why those relationships are important. At the same time, unique tags that are justified can help to provide information about a company and how it may differ from other companies.
Do you expect to issue other staff observations that have data from actual filings?
Mr. Bauguess: We are deliberating about the next analysis to do and put out publicly.
What is your assessment of the quality of XBRL filings?
Mr. Slavin: All data has errors or issues, and it is important to minimize those. We have seen certain areas of improvement and certain areas where improvement is still needed. I would say that some of the tagging in the primary financial statements has improved over time, but there is still a lot more work to be done, particularly in the disclosures.
Commercial data has errors in it as well. Do you think that too much is being made about XBRL quality?
Mr. Bauguess: There have been some academic articles and other studies that have shown it is not just the XBRL data that has quality issues-it is also some of the vendor data. This is very common.
Mr. Slavin: We occasionally see quality issues in the various vendor data sources that we consume.
Does the SEC plan to adopt XBRL for 1934 Act filings and other items currently not mandated to be tagged, such as earnings releases, proxy statements, MD&A, corporate actions, or financial information outside the financial statements?
Mr. Bauguess: We have embraced the idea of a culture of smart disclosure. That does not entail just XBRL. It entails any structured-language format. Any time we have a rulemaking that is going to require the disclosure of information that is a candidate for structuring, we are having dialogues about what that structuring could look like, how to do it, the lowest-cost way of doing it.
Mr. Slavin: It is about the right technology to use. Data is structured at different levels. XBRL lends itself well to the semi-structured nature of financial statements.
Mr. Bauguess: We spend a lot of time within DERA interacting with policy groups to make sure that solutions are available for structuring problems as they arise.
How has the SEC responded to the letter about XBRL that the House Committee on Oversight and Government Reform sent in September 2013? Has the letter prompted SEC changes or shifts of focus?
Mr. Bauguess: We cannot comment specifically on the letter. The response was not public. Generally speaking, we are aware of all the voices out there in the community-not just those who want us to do more with XBRL and enforce quality but also those who want us to do less and not require XBRL. DERA considers ways to minimize costs while increasing the usability of the data and educating the public to satisfy both camps. I think there is a nice middle ground where structured data can end up. If we can just make it through the next couple of years, I think a lot of this noise will subside.
Click here to access all Dimensions eNewsletters