“I ran software validation on my XBRL files, so they must be right.” Correct? Unfortunately, no.
Although software validation is a key component to getting XBRL right, there are many items in XBRL files that require human involvement. Relying on XBRL software alone to catch errors and produce high-quality XBRL will result in diminished XBRL quality in SEC filings.
It is important to understand where human involvement is necessary. Of the seven components necessary for achieving a high-quality XBRL filing, some can be conquered by software; some cannot; and some require a combination of the two. We can examine why this is the case by exploring how either automation or human involvement is needed to achieve each component of XBRL quality.
Key components of XBRL quality: automation or human involvement for each one?
In What Is Quality? Key Components of XBRL Quality in SEC Filings, we explained the seven critical components for XBRL quality:
- EDGAR Filer Manual
- US GAAP Taxonomy implementation guides
- XBRL best practices
- XBRL Specification 2.1
- US GAAP Taxonomy
- SEC Staff Observations/FAQs
- Accounting knowledge
EDGAR Filer Manual
Section 6 of the SEC's EDGAR Filer Manual (EFM) specifies the instructions for properly preparing XBRL submissions to the SEC. Some of these instructions can be automated by software, and some require human involvement. For those that can be automated, software developers have included them in the XBRL validation software programs that exist today. But a significant portion of the EFM instructions cannot be automated for XBRL compliance. For example, an EFM instruction which isn't included in software validation is the instruction to choose the element with the narrowest definition when there is a choice among elements that have definitions which fit the fact being tagged. This takes human expertise to determine first which elements have definitions that fit, and then which one of them represents the narrowest definition.
Adhering to the EFM requires a strong knowledge of the many EFM instructions, primarily because an individual needs to know not only how to comply with those rules that cannot be automated but also how to interpret and correct the findings of validation software for those rules that can be automated.
US GAAP Taxonomy implementation guides
The FASB's US GAAP Financial Reporting Taxonomy Implementation and Reference Guides provide insight and guidance on how to use the US GAAP Taxonomy. These guides are easy to read and understand, but they are not conducive to automation because they do not consist of a set of rules; rather, they show items such as the appropriate dimensional structures and elements for tagging specific disclosures. Compliance with these guides comes from reading them, comprehending them, and implementing the concepts where applicable. Each of these steps requires the involvement of XBRL-knowledgeable individuals.
XBRL best practices
The resolutions of the XBRL US Best Practices Committee consist of best practice recommendations, some of which can be written into automated validation rules and some which cannot. For those that can be automated, certain software programs (such as the XBRL US Consistency Suite) have included these rules in the software, thereby
allowing you to determine if your XBRL is in compliance with those best practices. For best-practice recommendations that cannot be automated, it is important for the XBRL preparer to read, understand, and apply the best practices where appropriate. Complying with many best-practice recommendations requires XBRL-knowledgeable individuals, such as when determining whether a revenue amount pertains to a customer, a counterparty, a disposal group, or a related party, and then selecting and structuring the applicable combination of tags for the specific situation.
XBRL Specification 2.1
XBRL software can determine if XBRL files are in conformance with XBRL Specification 2.1 as well as the Dimensions 1.0 specification. If an XBRL file is not in compliance, most all XBRL preparation and validation software will let you know promptly. No human intervention is necessary, other than to fix any issues identified by the software.
US GAAP Taxonomy
The US GAAP Taxonomy is a dictionary of more than 15,000 tags; consequently, there are many properties of tags and relationships between tags that can be automated into software to determine if these relationships and properties have been applied appropriately. Rules that can be automated include items such as “if Tag A is used, then Tag B should never be used” or “if Tag A is used, then Tag C should always be used” or “the value for Tag A should always be entered as a positive amount.” These rules are currently written into certain XBRL software programs, such as XBRL US Consistency Suite, allowing the user to be alerted to any violations.
Proper application of the US GAAP Taxonomy, however, also has an aspect that simply cannot be automated. There is no replacement for familiarity with the US GAAP Taxonomy by a knowledgeable individual to properly select, apply, and structure the tags. Properly communicating a disclosure, including making decisions on when to create an extension element or whether to use multiples axes, is best handled by a knowledgeable individual with familiarity of the US GAAP Taxonomy.
SEC Staff Observations/FAQs
Many of the XBRL issues in the SEC Staff Observations and FAQs can be identified by software and, consequently, many are included in XBRL validation programs. Items from the staff observations for which software can check include some-but not all-unit types, positive/negative signs, invalid axis/member combinations, and improper calculations. On the other hand, staff observations that cannot efficiently be checked by software include improper extensions when a GAAP tag exists, improper use of a GAAP tag when an extension should have been created, and improper modeling of tags to represent a disclosure.
High-quality XBRL requires that the preparer of the XBRL understand the accounting meaning of the financial disclosures. In general, this knowledge is not easily automated and, therefore, this component of quality requires individuals with accounting expertise to be involved with the XBRL. Although some basic accounting disclosures can be automated for purposes of auto-tagging, this is very limited. Tagging accounting disclosures requires accounting knowledge and those XBRL files that do not use the knowledge of accounting individuals with XBRL experience are evident, thanks to the poor XBRL quality.
Both software and human involvement are required: the importance of judgment
Getting XBRL right requires both software validation and human involvement. Preventing errors also means realizing that, of all the errors that can exist in XBRL submissions, there are those errors that can be detected by software and-the difficult ones-those that cannot. Everything from selecting elements to determining proper calculation relationships requires judgment and experience. Of course, to the extent that automation can assist with XBRL preparation and review, it should be used. But most aspects of XBRL preparation require judgment, and for those items that can be automated, judgment is still necessary to determine if the automation has handled them correctly.
Unfortunately, there are too many cases of over-reliance on XBRL software, which leads to a poor-quality XBRL filing. Understanding the limits of XBRL software validation, and the areas where there is a need for involvement by knowledgeable individuals, is critical to quality results.
How Merrill assures quality XBRL filings
When we handle an XBRL submission, it is prepared or reviewed by our experienced CPAs. With deep XBRL knowledge, we understand the intricacies of XBRL and the SEC rules, and can apply that to the XBRL documents. Equally as important, we are able to effectively communicate that understanding to the filing company's personnel. Since the filing company is ultimately responsible for the accuracy of the XBRL files, it is imperative that meaningful conversations occur between clients and consultants until a consensus is reached that the XBRL files are correct. These discussions can educate the company personnel on how information translates into the proper XBRL structure and data entry. We have clients who have become quite expert in the requirements of XBRL because they learn through our process and build on their knowledge from quarter to quarter.