Digital Financial Reporting

So I am in brainstorming mode, trying to figure out how to communicate the information I am trying to communicate.  Here is a disclosure analysis summary which contrasts how different public companies reported the Level 3 Disclosure Text Block and the Level 4 Disclosure Detail for various disclosures.  Currently I only have 2 disclosures, but I will build that up to around 75.

This is the summary of which disclosures I will be testing in machine readable form. This blog post explains things in more detail.

I will leave it at that for now.  More information coming soon.

So here is a documented case where XBRL contributed to detecting and correcting an accounting error. Click on the image below to get a larger image after you read this.  In this public company's March 31, 2017 income statement (on the LEFT) they ordered their income statement line items one way.  Then, in the June 30, 2017 income statement (on the RIGHT) they ordered them a different way.

The NEW way is consistent with literally all public companies that report these line items and consistent with the US GAAP Financial Reporting XBRL Taxonomy.  The OLD way was not.

(Click image for larger view)This inconsistency was detected using the fundamental accounting concept relations validation continuity cross checks. Now, they still don't have their XBRL-based representation correct.  They have some concept selection issues that need to be fixed.

A measurement of XBRL-based public company financial reports to the SEC that I have made indicates that those reports are about 88% consistent with what I would expect.  That said, the measurement is per a sampling of 65 disclosures for all 6,023 public company 10-K filings for 2016 and I cannot call this a verifiably scientific set of results.  However, if anyone has a better measurement I would love to see that measurement.

Here are the details:

(Click image for larger view)I am just starting a journey that will be very similar to the measurement of the fundamental accounting concept relations that I have been taking.  Those measurements are more precise and indicate that 99.19% of those relations are consistent with expectation.  There are currently 5,271 of 6,000 reports or 87.9% of all reports that are consistent with all 22 of the fundamental accounting concept relations.

Right now I am calling this measurement the "Disclosure Mechanics" of a disclosure.  If you want to understand this in detail, please go read Understanding the Mechanical Rules of Disclosures.  Here is the crash course.  First, I create logical, structural, and mathematical relations information in machine-readable XBRL format.

That rule in human-readable form looks like this:

(Click image for larger view)There are similar rules for 65 disclosures.  Software uses the rules (currently there are two commercial implementations) to check disclosures.  Here is the explanation mechanism provided by one of the commercial implementations:

(Click image for larger view)The results are summarized and either a disclosure is consistent with expectation or inconsistent with what was expected.  There are THREE REASONS an inconsistency might occur:

1. The filing has an error that needs to be corrected.
2. The US GAAP XBRL Taxonomy has an error that needs to be corrected.
3. My test or the software algorithm has an error that needs to be corrected.

How do I know I am getting the results right? Observed empirical evidence from the XBRL-based reports themselves.  Here is an example of that for the inventory components disclosure.  I have some really good tools that I have created and perfected.  Also, there are commercial tools that are becoming available.  I helped XBRL Cloud create this tool for checking Disclosure Mechanics rules.  Tools will continue to improve and evolve.

Over time these errors are corrected and the consistency rate increases, just like they did for the fundamental accounting concept relations.  Harmony increases in the system, information quality goes up.  What is the quality goal?  In my view, the goal is Sigma Level 6 which is 99.99966% consistent with expectation. Right now we are between Sigma Level 2 and 3.

If you want to know more about creating zero-defect XBRL-based reports, please read Blueprint for Creating Zero-Defect XBRL-based Digital Financial Reports.

If you have some tool ideas, please send me an email.

It is not so much that XBRL is anything really revolutionary.  XBRL is an idea similar to ideas that has been used by many industries to structure information to make things more efficient.  For example, electronic medical records and digital blueprints are similar ideas that have existed for years.

XBRL is enabling the institution of accountancy to catch up to the present.

Now, will XBRL and other technologies change the institution of accountancy?  Absolutely.  Ask yourself a question.  "What if all financial reports were digital?"  What exactly would that mean?

If you want to catch up to the present, consider reading Getting Ready for the Digital Age of Accounting, Reporting and Auditing: a Guide for Professional Accountants.

Old school approaches to creating financial reports have been inefficient for years.

I created a disclosure analysis prototype.  This analysis tool prototype looks at the current inventory components roll up.  It compares the Level 3 Disclosure Text Block representation and the Level 4 Disclosure Detail representation.  You can see and compare the text block and the detail side-by-side.

The rules for the disclosure are useful not only for verifying that disclosures were created correctly, but also for querying information from reports.  I realize this prototype needs more work to make it more useful, I am just trying to get all the basic pieces of the puzzle working.

Currently, the working prototype has information for the S&P 500 companies from their 2016 form 10-K.  All of the information and renderings are provided to me by the XBRL Cloud Edgar Report Information Web Service API.  With permission, I created copies of the renderings, stripped out the detail that you can normally drill down to (you have to subscribe to XBRL Cloud to get that functionality), and use that information to provide the comparisons.

Additionally, you can navigate to the publicly available XBRL Cloud viewer to look at details.  I also link you to the SEC Filing Page of the XBRL-based financial report submission so you can go look at that.

Here are some observations that I have. Of the 500 companies analyzed:

• 158 companies provide both the Level 3 Disclosure Text Block and Level 4 Disclosure Detail that I would have expected.
• 189 companies do not provide EITHER the Level 3 Disclosure Text Block OR the Level 4 Disclosure Detail; basically they don't have inventory.
• 15 companies report the details of components of inventory directly on their balance sheet and therefore per SEC filing rules they are not required to provide the Level 3 Disclosure Text Block
• 26 companies reported the Level 3 Disclosure Text Block, but I did not find the Level 4 Disclosure Detail.
• 37 companies reported the Level 4 Disclosure Detail, but I did not find the Level 3 Disclosure Text Block and the company did NOT report the detail in the balance sheet.

What about the other 75?  Well, I am figuring that out.

Here are some specific observations:

• FLOWSERVE CORP made the mistake of combining the representation of two disclosures into one making the Level 4 Disclosure harder to read.
• RAYTHEON added an Axis to the facts and then assigned each of the inventory component facts to that Axis, which achieves nothing.
• Advanced Auto Parts put the inventory components roll up together with the valuation allowance roll forward, making the Level 4 Disclosure Detail harder to read.
• Allegion plc combined multiple disclosures into their Level 4 Disclosure Detail representation making the representation virtually impossible to comprehend.
• ANADAKRO PETROLEUM CORP used the general Level 3 Disclosure Text Block, but then used a specialized concept for the total of the inventory components roll forward.
• DOW CHEMICAL CO created Level 4 Disclosure Detail presentation relations that do not match the Level 3 Disclosure Text Block and they appear to have not created any XBRL calculation relations.
• Honeywell International uses an different approach to representing the subtotal before the LIFO reserve.  Honeywell uses us-gaap:InventoryGross as the subtotal.  FMC Technoligies INC uses the same us-gaap:InventoryGross concept, but they use a different LIFO and valuation adjustment concept. FORD uses a different concept than Honeywell.  Dr Pepper Snapple Group, Inc. uses a different concept than Honeywell. DUPONT uses a different concept than Honeywell.  (Huh, I am not sure if these companies are making different disclosures or are providing the same disclosure using inconsistent approaches. More investigation is necessary.)

Anyway, I will leave it at that for the time being.  Interesting stuff.  This would be a great resource for teaching intermediate accounting.

If you have any ideas that would help turn this into a useful tool, be sure to send them my way.