Representing Information Logically
Wednesday, December 30, 2020 at 10:26AM
Charlie in Digital Financial Reporting

There is a difference between signal and noise.  Compare these three representations of one fragment of a financial report, the financial highlights:

  1. Represented with a hypercube but no dimensions | (Get to full report)
  2. Represented with a hypercube and four dimensions | (Get to full report)
  3. Represented with a hypercube and twelve dimensions | (Get to full report)

Think about something.  Logically, what exactly is the difference between the representations with 0, 4, and 12 dimensions? 

The first representation is thrown in to round out the set of examples.  I could have shown another representation with NO HYPERCUBE.  So what exactly is the difference between a representation with NO hypercube and no dimensions and with 1 hypercube and no dimensions?  What logic does the hypercube provide?  The answer is that the hypercube only provides something to reference so that you can get to the representation itself which is identified by the hypercube.

The second representation was verified to be correct mathematically using XBRL Formulas. If you look at the validation result you can see there are 46 assertions which were used 96 times to verify information and that 100% of the results are "satisfied" meaning everything checks out fine.

The third representation is exactly the same as the second except for the fact that I added those additional dimensions to the financial highlights and to the balance sheet. See the presentation relations and the definition relations.  I then ran exactly the same verification rules to check and see if the math was still correct using XBRL Formula.  If you look at those results you notice that there are still 46 assertions, 96 were fired, and 96 were satisfied.

What is my point?  Logically, the XBRL Formula processor sees both the second and third representations as being logically the same.  Proof of this is the second and third representation validation results being 100% identical.

Why is this important?  Read through the list of XBRL dimensions in the second and third representations.  How helpful to you are the additional 8 in understanding the reported information?  The first four provide explicit and helpful information: what is the reporting entity, what is the legal entity, what is the report date, and what is the reporting scenario.  You can get away without the report date and the reporting scenario; those can be safely implied.  Maybe even the consolidated entity could be implied.

But customer, debt instrument, segment, director, related party, share ownership plan, reconciling item?  Those are details that are used on some other hypercube but are not relevant to the financial highlights or balance sheet.  Basically, those additional 8 dimensions are noise rather than signal.

Dimensions in XBRL work a certain specific way.  That way is specified.  Each regulator or other party implementing XBRL does not get to reinvent what dimensions are, what they do, or how they work.  All that is specified.

My little proof shows that most of this stuff works consistently across regulators that have implemented XBRL; a few things don't quite work correctly.

For more information please see my document Representing Structures.

Article originally appeared on XBRL-based structured digital financial reporting (http://xbrl.squarespace.com/).
See website for complete article licensing information.