Digital Financial Reporting

The XBRL filings of SEC companies provides insight into why an information model is important. Consider this one presentation linkbase from the US GAAP Taxonomy: Statement of Income. If you click on that link and look at the page, you will notice that the presentation network of the income statement starts off something like this:

Income Statement [Abstract]             {ID1}

Statement [Table]                       {ID2}

Statement, Scenario [Axis]  {ID3}

Statement [Line Items]       {ID5}

You can click on the page and get to that presentation arrangement by collapsing or expanding the line items of that network. The IDs on the right just help you see which nodes I want you to focus on.

Why is the income statement organized in this manner, as a [Table]?  Well, there are two primary reasons reasons.  The first reason is that an income statement have a number of dimensions, for example it could be actual or budgeted, it could be for a consolidated group or for a subsidiary, or there are other breakdowns or "dimensions" which could be provided.  Those dimensions can be expressed using the "Statement, Scenario [Axis]" and the facts for which values are expressed are provided within the "Statement [Line Items]".  The second reason this [Table] is organized in this manner is because the US GAAP Taxonomy Architecture says that all [Table]s are to be organized in this manner (see the US GAAP Taxonomy Architecture, section 4.5 Implementation of Tables, page 34).

In fact, it is not that just this one [Table] is organized in this manner, ALL [Table]s in the US GAAP Taxonomy are organized in this manner.

However, look at the following sample of XBRL filings to the SEC:

http://www.xbrlsite.com/demos/console/TreeView-0001047469-09-007400.html#ID1([Table], [Axis], and [Line Items] used just like US GAAP Taxonomy)

http://www.xbrlsite.com/demos/console/TreeView-0000950123-09-030335.html#ID505 ([Table], [Axis], and [Line Items] used just like US GAAP Taxonomy)

http://www.xbrlsite.com/demos/console/TreeView-0000950123-09-029921.html(Table on balance sheet just like US GAAP Taxonomy, but no Axis on the income statement)

http://www.xbrlsite.com/demos/console/TreeView-0001104659-09-046329.html#ID1 (No table, no axis, no line items)

http://www.xbrlsite.com/demos/console/TreeView-0001104659-09-048234.html#ID50 (No table, no axis, does have line items, but not modeled using [Line Items] marker)

http://www.xbrlsite.com/demos/console/TreeView-0001310243-09-000085.html#ID1 (Table, line items, but no axis)

http://www.xbrlsite.com/demos/console/TreeView-0000004904-09-000117.html#ID1 (No table at all)

http://www.xbrlsite.com/demos/console/TreeView-0001104659-09-047977.html#ID1 (Line items, but no table or axis)

http://www.xbrlsite.com/demos/console/TreeView-0001104659-09-046313.html (No table, no axis, no line items)

http://www.xbrlsite.com/demos/console/TreeView-0000037748-09-000037.html (Table and line items, but no axis)

Why the inconsistency between the filings?  Why the inconsistency between the filings and the US GAAP Taxonomy?  The reason is that no "model" is being enforced as to how, in this case SEC filers, create their extension taxonomies.  There is no "information model" or manner specified to construct this component of the XBRL taxonomy and no automated validation to ensure that the model is being followed.

An articulated information model achieves many things, not just consistency within an XBRL taxonomy or between a base XBRL taxonomy such as the US GAAP Taxonomy and those who extend that taxonomy.

• The first benefit is that if the information model is articulated (specified) then automated validation processes can be created in order to help those creating extension taxonomies to do so in accordance with the intentions of the creators of the taxonomy which is being extended.
• The second benefit is that because of the consistency, rendering of the XBRL instance information is easier because the XBRL taxonomy is more consistent (both the base taxonomy such as the US GAAP Taxonomy and the extension taxonomies created by in this case SEC filers).
• The third benefit is that the entire process of extending can be vastly easier for users of the taxonomy.  The automated validation is only the first step.  Literally, software applications can be built in a totally different manner BECAUSE OF THE KNOWN CONSISTENCY!  Business users tend to miss this point.  Literally, the entire notion of a [Table] can be hidden from the user because software applications can be constructed to leverage the consistency, hiding the complexity of working with an XBRL taxonomy from the user.

The third benefit about is by far the most important, but it is only achieved if you also have the first two: the consistency of an articulated information model.  This points out another challenge working with the US GAAP Taxonomy: when is something a [Table]?  This problem can be easily overcome by doing what the creators of the COREP taxonomy did: make EVERYTHING a [Table].  Now, the COREP taxonomy does not call them [Table]s, they use the XBRL term hypercube. But, everything within the COREP taxonomy is constructed in one very easly to explain and leverage information model.

BOTTOM LINE

The bottom line here is this. If you articulate an information model, meaning a formal set of rules as to how that taxonomy is to be constructed, then those rules can be used to ensure that your taxonomy was created consistently and to ensure that those extending your taxonomy do so consistently. The consistency can be taken even further making using the taxonomy significantly easier if every area of the taxonomy is created in the same manner, in this case we are using [Table]s but actually [Table]s are a way to implement the multidimensional model.  If, say, every component of the taxonomy works in a similar manner, that similarity can be leveraged by software creators to make the experience of using that taxonomy easier, significantly easier in many cases.

There are a number of other entries on this blog which discuss the notion of an information model.  Also, you can read more about this in my book which will be available next month, XBRL for Dummies.

OK, now I am getting somewhere.  I have an updated Excel spreadsheetfor comparing the XBRL presentation linkbases of the 408 SEC filers which I am taking a look at.  This Excel application allows you to:

• Compare one SEC filer's presentation linkbase with another.
• Compare one SEC filer's presentation linkbase to the US GAAP Taxonomy (Commercial and Industrial Entry Point only)
• Compare two pieces of the US GAAP Taxonomy presentation linkbase (again, CI only).

You can see how the tool works here (it is really basic, but I am fiddling with approaches to embedding video in this blog, so here is the video):

I have uploaded two improved tools for viewing SEC filer company taxonomies.  Both tools are written using Microsoft Excel 2003. These tools can be found here:

• Taxonomy viewer tool: This tool allows you to select a company and view the company's taxonomy.
• Taxonomy comparison tool: This tool allows you to select two companies and view each taxonomy at the same time, helping you compare the two taxonomies.

The next tool on the list to create is a tool which allows you to compare a company taxonomy with the US GAAP Taxonomy.  Although, I will only do this for the Commercial and Industrial entry point.

I would challenge people to understand how to write Excel macros to improve these applications (i.e. add new interesting features) and send them to me and I will make them available to others.

I have created two new resources for looking at SEC XBRL Filings.  Both are for looking at XBRL taxonomies used by filers.

The first resourceis a simple little XBRL taxonomy viewer which uses HTML pages.  It is fast and it allows you to look through all the taxonomies of those who filed XBRL with the SEC.

The second resourcelets you look at the US GAAP Taxonomies. Actually, this only provides views of the presentation linkbases of the commercial and industrial (CI) entry point.  That is my focus area for looking at the filings because most filers are in this category and I don't have time or desire to provide this for each US GAAP Taxonomy entry point.

I want to point out an interesting thing about these taxonomy viewers.  If you look at the far right hand side of the concepts shown in these views, you will see a number that is in gray and looks like {ID123}.  Those numbers allow you to link to a specific concept in the taxonomy using those HTML pages.  This little trick is stolen from DecisionSoft SpiderMonkey which allows you to do basically the same thing.  For example, click hereto see what I am talking about. (i.e. you go to the ALLERGAN INC taxonomy "Supplemental Cash Flow Note" which is ID131).  Look at the link to see how this works.  (You will see why this is a handy tool next week when I use it in some other information that I provide.)

I realize that I am throwing a lot of stuff out here.  Hang tight and check back, I will provide a better organized view into all this information so those who what to explore all these SEC filings can do so.

This is a report which compares the presentation neworks of the XBRL filings made to the US SEC.  This is very interesting and somewhat unexpected results. The number of presentation networks in each filing ranges from 6 (9 filers have this number) to 55 (The Dow Chemical Company).

What is rather odd is that only three of the network roles come from the US GAAP Taxonomy.  These are the three and the number of times they are used.

I would have expected that the vast majority of the networks used would come from the US GAAP Taxonomy.  Huh.  I need to check and see what is going on, maybe I am misinterpreting or misunderstanding something.

If you want to go look at the presentation networks of a filer, you can do so here (requires IE 7 or 8) or here (see the 10th column in the table); but this is probably the best option for most people.