Digital Financial Reporting

This is an interesting interface: Pinterest.  Imagine a Pinterest-type interface into SEC filings.

Consider this organization of SEC filing information which lists filers by sector and then lets you look at the pieces of a filing for a filer. That is one way I chose to organize the information.  You decide what sector you want to look at, then you decide what company to look at, then you can view the pieces of the SEC filing for that company, here is one piece (Link 1). That is the first sector (Agricultural Production Crops); the first company (FRESH DEL MONTE PRODUCE INC); and the first piece of their financial report (Document information).  That is one path to that information.

Now, this is a different way to organize the information; by filing component.  If you go to that link, then go to the last item on the list, number 128 Document information (291 examples); click on that link and you get a list of companies,  go to item number 105 (or search on FRESH DEL MONTE PRODUCE INC); then click on that; you end up on this component of an SEC filing (Link 2).

Notice that those two components are the same, the first (Link 1) and second (Link 2).  They are two paths to the same information.

Why is that interesting?  Well, here is why.  Before XBRL, it would be impossible to link directly to a component of an SEC filing.  The best you could do is link to a document, such as this, the SEC Form 10-K for FRESH DEL MONTE PRODUCE INC from which that information came.  Granted, the document information is not that interesting; but I could do the same for the shareholders' equity disclosure metadata (i.e. taxonomy) or the information itself (i.e. fact table, I did not do a good job providing enough room to see the entire block of information, that was not necessarily my purpose and besides, I am not a very good programmer).  I could point to other component, any component, and I could even point to a human readable rendering of that component (but I did not provide that because generating such a rendering is beyond my programming skill, the SEC does not make the information available in this manner, and I have not found anyone else who provides information in this manner.)

What is the point of all this? I provided my two paths to the information (and there is actually a third path that I provided, a list of companies by company name that does NOT break the taxonomy down into pieces but rather shows all the pieces together); but what if you want some other path to the information?  Cool, no problem.  You can provide it.

Click on one of the links in Pintrest.  Here is one of some cute puppies.  Look at the URL, the ending is just a number.  Anyone can point to that link.  Anyone can organize whatever set of information they want, include information about cute puppies, and provide a link to that spot on the internet.

As the book Everything is Miscellaneous points out, there are many different ways you can organize information and no matter how well you plan an organization scheme; you will always want other organizations of the information.  Further, consider all this in the context of Tim Berners-Lee vision of "Linked Data".

This is the power of structured information which is organized in order to enable this type of functionality.  Poor organization gets in the way of this type of functionality.

In addition to the proper structuring of the information is the metadata, or data which explains the data.  In my three organizations, this is the metadata I am using to get you to where I want you to be: the company name, the sector of the company, and the component of the financial report which exist within the financial report. 

You have to reliably and accurately give meaning to a piece of information.  You get problems if, for example, information is grouped in unusable permutations/combinationsor if the pieces are not consistently and accurately identified at the appropriate level or if the metadata is not consistent enough.

This ability to organize digital information has huge ramifications for how things can and will be organized in the future.  If you are looking at SEC filings through the lens of the existing EDGAR system and existing means of distributing all that information and cannot grasp that these will soon be legacy; then you certainly will not be able to project what this means into the processes which impact your daily life. This understanding is critical.

In my work to experiment with and further expand the Financial Report Semantics and Dynamics Theory, I analyzed a set of 291 SEC Form 10-K financial filings.  Below are the resources I used which might be helpful to other accountants trying to better understand these filings. (These were Form 10-K filings filed between December 1, 2011 and February 29, 2012.

• Select set of 291 Form 10-K filings. This resource makes it easy to explore the taxonomy and fact tables of these 291 Form 10-K filings.
• Components of select set of 291 Form 10-K filings.  This resource lets you look at the common components of each of the 291 Form 10-K filings.  For example, you can go to one place and look at each of the 291 balance sheets, income statements, cash flow statements, etc.  Great way to compare how different filers modeled similar components.
• Select set of 291 Form 10-K Filers organized by Sector.  This is the same list of 291 Form 10-K filers, but here they are organized by sector. This makes it easier to compare filings for filers which are in the same sector.
• Other resources.  Here are additional resources which have information about all 4416 Form 10-K and 10-Q filings.

What am I seeing?  Well, first off; the entire set of 4416 is consistent with other analysis work which I have done and continues to show the core financial report semantics relating to the existence of assets, liabilities and equity, equity, net income, and net cash flow; plus the balance sheet does, in fact, balance.

Now, for this analysis; what I did was take specific select filings which had all of these core semantics, which where "commercial and industrial companies" (i.e. they were not in any specific industry or had no specific activities which used industry/activity specific accounting policies such as banking, insurance, real estate, etc.), they had no other obvious errors which would make me believe that the filings were of significanly poor quality, and they had the following additional semantics:

• Current assets and current liabilities were both reported (and found)
• They may not have reported liabilities, but I was able to impute the value for liabilities (i.e. Liabilities and Equity - Equity = Liabilities)
• Liabilities and equity - equity - redeemable noncontrolling interest or temporary equity - liabilities (or the value imputed for liabilities)Income (loss) from continuing operations before taxes was reported (and found)
• Net cash flows from operating, investing, and financing activities were each reported (and found)
• Net cash flow from operating activities + net cash flow from investing activities + Net cash flow from financing activities + exchange gains = Net cash flow
• A concept for revenues was found
• Income tax expense (benefit) was reported (and found)
• I was able to properly impute income (loss) from continuing operations before taxes using the difference between net income (loss) and the income tax expense (benefit)

So basically, what I am saying is that if the filing was a 10-K (and was detailed-tagged), if I found all the core financial relationships to be true, and if I found all of these basic financial relationships above to all be true and if they worked correctly; then I used the filing for my analysis.

What was I looking at? Disclosures mainly.

What did I find?  Well first off, pretty much as expected; every filer reported "significant accounting policies".  Now, they did this in many different ways, but it was there for everyone in one form or another.

Something else I became aware of that had not occurred to me in the past is that information for all disclosures is reported twice: once in the [Text Block]s or marked "(Table)" in some filings and a second time in the "(Detail)".  Two things about the (Table) stuff which is really [Text Block]s and (Detail) stuff.  First, not all sections like this are consistently marked and the is no physical connection between the (Table) and the (Detail) for the same information.  Another way of saying this is that I personally wish these were marked consistently and that there was some "hook" which explicitly indicated that they went together.

Other sections which I found in almost 100% of the filings were things such as:

• Earnings per share [Text Block]
• Segment reporting [Text Block]
• Quarterly financial information [Text Block]
• Income tax disclosures [Text Block]
• Debt disclosures [Text Block]
• Commitments and contingencies [Text Block]
• Goodwill and Intangible Assets Disclosure [Text Block]
• Fair Value Disclosures [Text Block]
• Pension and Other Postretirement Benefits Disclosure [Text Block]
• Property, Plant, and Equipment Disclosures [Text Block]
• Financial Instruments Disclosures [Text Block]
• Stockholders' Equity Disclosures [Text Block]
• Business Combinations Disclosures [Text Block]

There were other filings which had high number of occurrences but not 100%.  But there were, as was expected, lots of consistencies.  Now, the filers did not always use the same US GAAP Taxonomy concepts to express the same things, but just like "liabilities and stockholders' equity" and "liabilities and partner capital" and "liabilities and member equity" are all really "liabilities and equity"; there is lots and lots of nice consistency.

Further, if you found the [Text Block]s above; what do you think the probability is of the details of those items to likewise exist within the filing?  Pretty darn high.

Now, the disclosures work differently than the statements.  You pretty much always have a balance sheet, income statement, and cash flow statement.  What is on those statements can be much different; but what is rarely different are the fact that balance sheets exist and balance sheets have assets, liabilities and equity, equity, and the balance sheet balances.

While a few of the disclosures always exist, most of the disclosures exist only exist if specific line items exist in the statements of if a reporting entity has those items to disclose.  But, if the disclosure exists, it likewise has specific semantics.

What is my point? Semantics.  This stuff is not random.  There are reasons things show up or if they don't show up and if they do show up, other things must show up or not show up and the information ties together.  This is all like a financial reporting sudoku puzzle.

One of the issues which the financial reporting community will need to address can be demonstrated by looking at the disclosure of significant accounting policies in SEC XBRL financial filings.  The issue is a general issue, it relates to many areas of a financial report. There really is no "right" or "wrong" answer, there are just different approaches and each of those approaches has "functionality" which it delivers. You may, or may not, see this as a "change to financial reporting" or a "change in US GAAP".  That is not the point of making this information available.  The point is to help accountants to understand the issue.

The issue relates to the difference between unstructured information and structured information. With legacy approaches to creating a financial report the information disclosed is basically unstructured an therefore there is no "box" that information must fit into.  You can understand "the box" by realizing that when you move from unstructured to structured information, you basically take the unstructured information, structure it in some way (thus creating the box), and you put the information into a box.

The "box" is not good or bad, it is just a box.  It is not that unstructured is good and structured is bad; or that structured is good and unstructured is bad.  They are just different.

So here is what I mean.  If you understand financial reports, then you know that within a financial report, such as within an SEC financial filing, you have to disclose significant accounting policies.  If you look at SEC XBRL financial filings (which I have, more info later) you will see that 100% of the 10-K filings disclose significant accounting policies.  Reporting rules require this.

But, filers structure this disclosure using XBRL in different ways.  Here are the primary ways I see this done (this is looking at only the [Text Block] or (Table) which every SEC filer provides in their SEC XBRL financial filing:

• Significant Accounting Policies (us-gaap:SignificantAccountingPoliciesTextBlock) is used most.
• Basis of Presentation and Significant Accounting Policies (us-gaap:BasisOfPresentationAndSignificantAccountingPoliciesTextBlock) is a distant second
• Business Description and Significant Accounting Policies us-gaap:BusinessDescriptionAndAccountingPoliciesTextBlock) is next
• Basis of Accounting (us-gaap:BasisOfAccounting)
• Organization, Consolidation, Basis of Presentation, Business Description and Accounting Policies (us-gaap:OrganizationConsolidationBasisOfPresentationBusinessDescriptionAndAccountingPoliciesTextBlock)
• Organization, Consolidation and Presentation of Financial Statements Disclosure and Significant Accounting Policies (us-gaap:OrganizationConsolidationAndPresentationOfFinancialStatementsDisclosureAndSignificantAccountingPoliciesTextBlock)

Now, some filers (very few) decide that none of those concepts work for them and decide to create extension concepts. Those are obviously errors and one of the existing concepts should have been used.

But, other filers combine different things together and do feel obliged to create an extension concept and it creating such a concept can be justified.  For example, one filers created the concept Summary Of Significant Accounting Policies And Recent Accounting Pronouncements [Text Block].  They combined two things which both have concepts which exist in the US GAAP Taxonomy; but is this the right thing to do?

THAT is the issue.  Basically, it is possible to come up with all sorts of permutations and combinations of information.  Each permutation/combination needs to have a "box" or concept created so that the SEC filer can put the information inside that box.  This is the way they have always reported.

But, the filer creating such a concept basically makes comparing information significantly more challenging.  You can still do it, you just need to map the filer extension concept to some other concept which is defined to include significant accounting policies.

Or, alternatively, the filer could unbundled the information into the two concepts which exist; separating "Significant accounting policies" and "recent accounting pronouncements" into two separate boxes.  This reduces the permutations and combinations.

So, it seems that the spectirum of options is as such:

• Provide lots and lots of permutations and combinations, and still allow a filer to create more permulations and combinations
• Provide lots and lots of permutations and combinations, but DON'T allow the filer to create other possible permulations/combinations
• Require SEC filers to unbundled their disclosures, and also their financial statement line items, into discrete disclosures/line items (i.e. get rid of the bundles)

Like I said, there is not necessarily a right or wrong answer here; it is just a choice which the financial reporting supply chain needs to figure out.  What would be good is to understand the pros and cons of each alternative, all things considered.

And I point out again; this is not just an issue with significant accounting policies; it is a general issue for which I am pointing out with this significant accounting policies example.

Which alternative do you feel is the best choice, all things considered?