On Tuesday, Chairman Cox announced an early release of some open source
software that allows you to analyse information that has been filed with the
SEC in XBRL format. It’s a web based demonstration environment, available here.
So, this is an interesting start. It’s a hint at some of the power that is
available when analysing XBRL information. It highlights some of the things
that still need to be done to improve the framework for filing XBRL with the
Commission (or indeed, any regulator). I think it also provides a bit of a
wake-up call to the consortium about the so-called "rendering"
Some of the obvious issues include:
Requiring meaningful labels for logical groupings of concepts. In the XBRL
specification we call these "extended link roles". At CoreFiling we
call them "Groups" because the other name is not very helpful.
Enforcing a single entity identifier.
Determining and enforcing a single versioning/taxonomy life cycle strategy so
that it’s easy to construct a time series across multiple filings by the same
Imposing an “order” constraint on disclosure segments and contexts, and equally
important, imposing a consistent framework for segment identifiers, for
individual filers, across time.
Marc van Hilvoorde is leading a working group that is coming to grips with some
of the issues to do with rendering. Rendering is a really tricky area. At one
level, accountants that prepare financial disclosures need them to look exactly
"so". Developing a specification that can provide really precise
rendering descriptions could
take quite some time be impossible. Those
consuming the data, on the other hand, are really just looking for a broad
brush approach, that will help set out tables and headings etc., so that the data can be easily conveyed to the user. It is this latter area that Marc’s
group is going to be thinking about.
I gather that that there will be another, independently developed analysis
tool, that will also be open source, that the Commission’s contractors are
still working on. Cool! Bring it on. Oh! One other point. I believe the data in
this current trial is being batched up overnight. Fair enough, it is early
days. One reason for that is that the SEC’s RSS feed is only updated every
night. What about sorting that out, guys? Once it’s been filed, it should be
available… shouldn’t it?