Digital Financial Reporting

I took that disclosure analysis tool and further modified it.  Not, it is customizable and you can change the concepts being looked for by simply adding a text block concept to a row in a spreadsheet.  Here are three prototypes:

• Basis of reporting
• Inventory components roll up
• Property, plant, and equipment components roll up

Here is a link to 65 disclosures that I am analyzing during my campaign to improve disclosure quality.  I will create some documentation and provide that.

For now, if you go into the column with the stuff that looks like HTML (Column "L") and you double click the cell, a form opens up and the HTML for the Level 3 Disclosure Text Block so you can read the disclosure.  The point of the Excel application is to example how public companies are using each text block to see if they are using them consistently.

In their Harvard Business Review article, The Competitive Landscape for Machine Intelligence, Shivon Zilis and James Cham summarize the state of machine intelligence (which they say is a more neutral term for artificial intelligence).

The article provides a PDF titled The State of Machine Intelligence, 2016 which summarizes somewhat of a machine intelligence "stack".

Here are two points made by the article:

Machine intelligence is not just about better software; it requires entirely new processes and a different mindset. Machine intelligence is a new discipline for managers to learn, one that demands a new class of software talent and a new organizational structure.

Every employee can use machine intelligence to become more productive with tools that exist today. Companies have at their disposal, for the first time, the full set of building blocks to begin embedding machine intelligence in their businesses.

Business professionals, just like professional accountants, need to gain a few new skills to understand that this change is real and it will be significant.

I have modified the Excel spreadsheet which I use to extract fundamental accounting concepts from the XBRL-based financial reports of public companies.  Using that same information extraction scheme I can now determine of the four required disclosures have been provided for.

• Here are the results of the information extraction (i.e. there are no macros in this spreadsheet)
• Here are the results and the code you can run yourself (i.e. contains macros)
• Information to Understand Four Required Disclosures (PDF)

So, I took the Excel spreadsheet that I had, modified it to grab the CONCEPT name rather than the value, and then I had it look for concepts related to four required disclosures: nature of business, basis of reporting, significant accounting policies, and revenue recognition policy.

Because the required disclosures are the same for everyone, I put the URL of every public company 10-K for 2016 (last year, not this year yet...that is coming) into the "List" in the Excel spreadsheet.  I pressed the start button, went to yoga, and when I came back two hours later the process was complete.

The results showed that 3,705 XBRL-based reports or 61% behaved as expected and all four required disclosures were found.  But that left 2,318 or 39% that were missing one or more of those four required disclosures.

Now, there are two specific reasons why one of these four required disclosures were not found:

1. The disclosure exists in the report, but for some reason the wrong concept was used or an extension concept was created.  That is an XBRL error.
2. The disclosure does NOT exist because the disclosure was not provided. That is an ACCOUNTING error.

The most interesting aspect of this little analysis is that I found that the Accounting Standards Codification (ASC) was "squishy" in terms of stating the requirements for these disclosures.  I would have figured that the ASC would be more "crisp"; providing clarity that these discourses are required.

There is specific wording that makes it clear that the nature of business, significant accounting policies, and revenue recognition policy are required.  But the basis of reporting disclosure is not specifically mentioned.  However, several things make it obvious that the basis of reporting is a required disclosure:

• Empirical evidence shows that 95% of all public companies provide this disclosure. What would the rational for having 95% make this disclosure and 5% not be required to make this disclosure?  I am not saying that there is not a specific reason that could be provided.  I am saying that if you don't provide a specific reason why it is not required; then by default it seems like it would be hard to justify making the disclosure if 95% of other public companies do.
• The basis of reporting essentially states explicitly the basis that was used to create a report.  US GAAP is just one basis.  IFRS is another basis of reporting.  Other Comprehensive Basis of Reporting (OCBOA) is also a basis.  There are others.  It is also good to include if the report is prepared using the accrual or cash basis, that the currency units are US Dollars.  That is pretty important information and the significant accounting policies state that important policies must be disclosures.  This is pretty darn important.
• Every audit report provides the basis of reporting.  All of these 10-Ks are audited and I would suspect that every audit report provides this information.  However, the audit report is from the auditor, not the reporting entity.  This information in the audit report does not satisfy this disclosure.
• There is not one professional accountant that I have asked which has said that the basis of reporting is not a required disclosure.  Not one.  Further, every disclosure checklist that I have seen includes the requirement to disclose the basis of reporting.

Do you disagree?  Do you have specific circumstances that you can point out when these disclosures are not required.  (Please read the PDF first, we point out known exceptions but none of them apply to the public companies that I am analyzing as far as I have seen thus far.)

Again, I am going to go look for specific reasons just out of curiosity.  But, I am going to wait until March 2018 when the 10-K season is completed to give the public companies an opportunity to fix as many of the errors as possible.

Finally, what is interesting is that the terms SHALL, MAY, SHOULD, and other terms are really not defined in the ASC.  Contrast this to RFC 2119 which is provided by the Internet Engineering Task Force (IETF) that explicitly define these terms in order to make things work better.  There is a big difference between SHALL and SHOULD. This is another good document on the topic.  And this is an interesting document, Shall We Abandon Shall? Interesting stuff.

This is a very useful XBRL Taxonomy Library provided by CoreFiling.  What would be even better is if they included all metadata for each of these taxonomies, including business rules.

What would even be better than that would be to be able to edit (i.e. extend) the taxonomies via some sort of a "wiki" mechanism.

If you understand how to make accounting process automation and process robotics work; then you recognize the need for metadata to make that automation work.  (If you don't understand how to make this work, then you will want to check out Closing the Skills Gap.)

If you want to get started with accounting process automation and process robotics; here is some high-quality, proven metadata that you can use.  There are basically two sets of metadata.

• Fundamental accounting concept relations continuity cross checks: That set of metadata is what makes these quarterly quality checks work.
• Reporting checklist and disclosure mechanics: That set of metadata is what makes this work for XBRL Cloud and Pesseract which is a noncommercial working proof of concept. (For more information on this, see here.)

Here is the metadata:

• Fundamental accounting concept relations continuity cross checks: (the OLDER version is not pure XBRL; the NEWER version is pure, standard XBRL)
  • Older version: (Human readable | Machine readable)
  • Newer version: (Human readable | Machine readable)
• Reporting checklist: (pure standard XBRL)
  • ONLY version: (Human readable | Machine readable)
• Disclosure mechanics: (pure standard XBRL)
  • ONLY version: (Human readable | Machine readable)

Available implementations:

• Commercial: XBRL Cloud
• Noncommercial: Pesseract (working proof of concept)
• Noncommercial: Excel prototypes for fundamental accounting concept relations
• Commercial: FlexRule(only certain reporting styles of fundamental accounting concept relations)

You can implement accounting process automation and process robotics using proprietary approaches such as a proprietary rules engine, the semantic web stack of technologies, or the XBRL stack of technologies.  My personal preference is the XBRL stack of technologies extended with additional non-standard functionality to overcome the known deficiencies of an XBRL Formula Processor.

This business logic gap analysis can help you understand which approach might best suit your needs.

To understand more possibilities, see the document Blueprint for Creating Zero Defect XBRL-based Financial reports.  If you want some more tips and tricks, see Putting the Expertise into an XBRL-based Knowledge Based System for Creating Financial Reports. If you want all the details, then you will want to check out Intelligent XBRL-based Digital Financial Reporting which basically has an accumulation of my notes for the past 20 years of working with XBRL.