One of the primary objectives of XBRL is comparability.
Create a standardized taxonomy and require that companies tag their financial
results to that taxonomy. This would be a universal language of business and
provide access to standardized data that was only available through expensive
subscription services in the past. And since this standardization was being
done by the filing companies, this data would be available to everyone almost
immediately. Good plan. Poor execution. The SEC’s implementation of XBRL allows
companies to add items to the standard taxonomy. So, the taxonomy isn’t really
fixed; it’s more like open source. Add items, change item relationships,
redefine the meaning of items, use items from other industry taxonomies, etc.
It’s pretty much anything goes.
I think there’s a reasonable argument to be made for extensions.
Every business is unique and sometimes there simply is not a standardized
concept that captures the meaning of the company’s line item. Fair enough. But
there’s always a related item that the extension could be linked to, and the
cost of adding those links would pale compared to the benefit to end users.
Alternatively, companies could express their uniqueness by adding dimensions to
standard concepts. For example, an ‘inventory type’ dimension would allow a
company to add custom members, all rolling up to a standard inventory item. This
is easily accommodated by XBRL’s dimension structure.
Short of that, the SEC should at least require that
extensions be rare and completely necessary. Remember that when a company tags an
item to a standard item, it also creates a unique label. So even without
extensions, a filing company is able to bend the meaning of a standard concept
to reflect its own meaning.
After mapping over 50,000 company extensions as
AsReported.com, it’s become apparent that there is little if any regulation
over whether, how or when extensions can be created. The tables below are a sampling
taken just from ‘Current Assets’. As you can see, there is little justification
for many of these extensions as almost identical items exist in the standard
For most serious end-users of financial data, comparability
is essential. By design, XBRL doesn’t require comparability, even though it
aspires to it. And its implementation turns a blind eye to completely