A company of Coca Cola’s stature understands
the value of high-quality investor communications. And Workiva understands
that protecting its dominant market share requires software that facilitates
creation of the XBRL files, protects the filer against errors and omissions and
informs the process through its knowledge of relevant specifications,
taxonomies, filing rules and the tagging practices of other companies. So when
these two market leaders combined to produce Coca Cola’s latest
10Q, I expected both data quality and usability.
I got neither. The filing scored 66.0 on XBRLogic’s Quality
Score, well below the already pedestrian average of 88.7 for
all filings in Q4. Here’s what I saw.
By way of
orientation, RED is BAD. In this case, it means that these items are missing
important metadata – their relationship to other items in the statement.
Important? Extremely. This is the fabric of the extended taxonomy model created
to depict the company’s financial results. Without it, data validation is very
difficult. Standardizing custom items (shown with the blue ‘x’) is very difficult.
Ensuring that values have the correct sign is very difficult. Usability of this
data becomes very difficult.
requires that these relationships are explicitly reported in XBRL filings in an
XML file called the ‘calculation linkbase’. Most companies comply, but many do
not. And it raises three obvious questions.
- Why does Coca
Cola think this is OK?
- Why does Workiva’s software
allow this important omission?
- Why doesn’t the SEC
enforce its filing requirements?
issue is subtler, but just as important. Coca Cola created
four custom tags in this statement, marked on the previous image with a blue ‘x’. Custom
tags, or extensions, are allowed by the SEC but undermine the data’s
comparability. These 4 extensions are clearly unnecessary. Shown
below, each line item has a matching concept in the standard taxonomy.
To take one
example, ‘Purchases of Investments’ was created as a custom tag. Three
proprietary mapping algorithms yielded a single standard tag. Two approaches
were based on tags by other companies for exactly the same item in the first
case and approximately the same in second. They’re shown below.