XBRL data quality is
not improving. Not from the investor's perspective. Here are 5 examples of
filings produced using Certent Disclosure Management software. Certent
scores near the top of XBRLogic’s Quality Score, so its clients make
marginally fewer errors than the competition. But is their software, and
compliance software in general, smart enough to:
- Guide
companies to produce complete, well-formed reporting taxonomies, consistent
with relevant US-GAAP sector taxonomies
- Encourage standardization by presenting industry filing data to
inform tagging, labeling and table formatting decisions.
- Produce financial data sets that can be easily and cheaply
accessed by investors without validation, modification and re-tagging of items?
Let’s look at the 5 examples,
then decide.
EXAMPLE
I. CONOCOPHILLIPS
- The
company tagged an item to an element in the bank taxonomy. CononcoPhillips is
not a bank. This requires the investor to replace this tag with one from the
commercial taxonomy. XBRLogic’s tagging solution searches prior filings in the
commercial sector for similar items and provides a menu of choices. SUGGESTION: Restrict tagging to standard elements
in the companies target taxonomy and common elements.
- Anytime
an investor is confronted with an extension (custom tag), he must find a
standard element and replace the tag. In this case XBRLogic’s expert system
tagger was used. Remember that companies are giving wide latitude to shade the
meaning of a closely related standard element by using custom labels. Since a
related item is almost always available, extensions should be rarely needed. SUGGESTION: Provide filers with a menu
of potential tags for each item that reflects the sector or peer group consensus
for like items?
- Here
an unclassified element is tagged in a classified balance sheet. The investor
is left to guess whether the item is current or noncurrent. SUGGESTION: Only allow tags to classified
elements if the target statement is classified.
EXAMPLE II. READING
INTERNATIONAL
- Segments should be represented in the company’s taxonomy as dimensions.
And ‘segment’ should be broadly defined as company-specific components of an
element. In the case below, the filer correctly tagged revenue segments as
dimensions, and incorrectly tagged cost of revenue segments as extensions. This
is obviously inconsistent and asymetric, creating matching problems for the
investor. SUGGESTION: Guide filers to use dimensions instead of extensions whenever and
however possible. It preserves the comparability of the data.
- Unnecessary extension. Note that XBRLogic’s tagging
algorithms include an expert system using taxonomy blocks, exact matching of
both labels and qualified names, and partial matching using NLP, shown here.
EXAMPLE
III. HALLADOR ENERGY
- Here’s an
example of a simple input error, transposing subtotals causing a mismatch
between labels and qualified names. Human error is understandable, but failure to
engineer checks to identify and correct human error is not. SUGGESTION: Enforce your validation checks. If an error is flagged, the filer
should not be able to continue with file preparation until the exception is
corrected.
EXAMPLE
IV. CENTURY COMMUNITIES
- Relationships
between elements are the connective tissue of XBRL taxonomies. Knowing the
arithmetic of the company’s taxonomy is key to understanding and utilizing the
data. This filing has omitted almost all of the summation-total arcs for the
income statement. This math can be deduced, but it’s one more hurdle for
investors. SUGGESTION: Do not allow filers to create incomplete or
poorly-formed taxonomies.
EXAMPLE
V. REXAHN PHARMACEUTICALS
- This filer
created 5 extensions for its cash flow statement. At least 3 are unnecessary. Shown
are acceptable standard elements generated by XBRLogic’s tagging process. The
two deferred lease items could also be tagged using a ‘type’ dimension against ‘Amortization
of Deferred Leasing Fees’.
- Sibling
errors are a common. They occur when the filer tags two elements as siblings in
the company’s taxonomy, but those elements are parent-child in the standard
taxonomy. ‘Proceeds from Issuance or Sale of Equity’ cannot be a component of
itself, yet that’s how it’s represented. SUGGESTION:
Require consistency between the filer’s taxonomy and the standard taxonomy.
Every inconsistency imposes a burden on the investor to reconcile the data between
models that are not 100% compatible.
These
observations and suggestions are based on using XBRL data, rather than from a
direct knowledge of the compliance software of Certent or any other vendor.
Judging from the data produced, compliance software is NOT smart enough to
produce a world-class data product. It wasn’t smart enough to identify and
correct the issues in these examples. And from my experience, it’s not getting
better. Companies continue to file 10Q's and 10K's using flawed or incomplete
taxonomies quarter after quarter, The absence of hard enforcement by the SEC
and soft enforcement by compliance software allows this to continue.