Thursday 2 May 2013

The case against extensions

Thinking again about Daniel Robert's recent post and the virtue of extensions... And shocked to find myself, at the recent International XBRL conference, one of the few to put my hand up in favour of extensions, I thought it might be worth exploring the other extreme - the absence of extensions.

The exclusive use of GAAP tags would give data vendors and wholesale users of data (e.g quants) what they crave - data they can normalize. Of course there are arguments that there is some data that can never be normalized but that won't stop people trying when the goal is to model the world (well maybe an economy or a sector).

So what would be the cost? Would a company's performance suddenly be mis-represented? Would the subtleties of it's economic contribution now become masked?

Isn't that the point of the label (data item description)?

For those who want a more nuanced understanding of a company's business achievements, they merely need to disregard the tags and focus their attention on the labels column. As per Daniel's piece, this applies equally to companies who wish to engage in the art of obfuscation. Isn't this the beauty of XBRL, you have enough data for everyone provided each type is used correctly?

OK so a tag may not fully describe the nature of a data value but the vendors are going to tag it with something anyway. Better that apples are tagged as fruit than risk them being tagged as oranges.

Ultimately this must be advantageous to companies, as rather than the aggregated holes into which their data will fall being decided by a data factory worker on the other side of the world, they get to decide which GAAP tag their precious data will be allocated to.

As James Claus of Moon Capital pointed out at last years US XBRL conference, investors would prefer data which has (if not down right reject data that hasn't) been classified by the companies themselves. And is the reason why Moon capital never use vendor sourced data. Companies need to get precious about their data and XBRL actually allows them to do that.


  1. Jim,

    It's a delight to see your blog. I'm learning about XBRL and by going through your archived entries I felt as if I took your entire journey compressed in 15 minutes. Seriously, I needed to see the history and I'm glad you captured it like you did, which gives great context. Off to the weeds I go. I just wanted to say thanks--there is value in preserving blogs you must know.

    Julius C.

  2. Jim,

    You have an outstanding platform for analyzing 10Q/10Ks - Thanks for all the work you've done. I currently analyze non-traded real estate investment trusts; so they report quarterly and annually but use "off-market" language that is essential for analyzing. I've searched all over but I was wondering where I can find the process used import the 10Q into the excel spreadsheet the way you've done on the XBRL to XL. Any direction would be greatly appreciated!

    Again thanks!