I will probably edit this post and change the title at some point.
Well in July we moved house. We now live in Luxembourg (again) and a lot of July was spent organising the appartment and going on holiday.
I have started prototyping the data import for the XBRL data. Prototyping is necessary as (as of the time of writing this post) I am on the 4th prototype. My thoughts on XBRL are that it is a great concept, but there are a number of problems.
A problem with it is that as there is little or no (sorry SEC & Edgar Online) validation of the structure of the data . There is a standard taxanomy for the different sorts of company (real estate, bank, insurance etc..) for the different types of report ( balance sheet, income statement etc..). A taxonomy defines a structure for presentation and data calculation purposes and references a schema of us-gaap approved line items (e.g. Assets, Accounts Payable etc..).
With each submission (10Q or 10K) the companies submit all the facts along with their own flavour of taxonomy. This is because they like to structure the report in their own way and even add their own company specific facts. In addition some companies will take an element from another report and use that when there is a perfectly good element in the report.
Due to the hierarchical nature of the data one would expect a “folder item” e.g. Current Assets would be the some of all the items within. I started working on this premise using just a presentation hierarchy. The problem with that is that in some cases there is a summary line item which is the sum of the items in the “folder” and it is included within the folder. Thus if you try adding up the folder contents you get double the correct amount. And it is not simply a case of just ignoring the summary item as some time it has the same name as the folder and some times not.
I then thought, I know what, I will use the standard taxonomy and import the company data into that and just add it up. Well the problem with that idea is that not all the company data finds a home due to naming. I thought not a problem, I will just add the missing items. Well that works sometimes but not always and when it doen’t you end up with duplicate items.
My latest prototype which is achieving the greatest satisfaction has reutrned to the company taxonomy only and using what I have learnt with the calculation hierarchy to calculate the summaries where necessary and then copy them over into the presentation hierarchy. This works but some of the presentations are absolutely diabolical. The image to the right shows in the top box the presentation hierarchy for Caterpillar (CAT) and in the lower box the calculation hierarchy. As you can see, I think the lower (the calculation) looks better in that there are 2 top level items:
“Assets” and “Liabilities & Stockholders Equity” and within the “Liabilities & Stockholders Equity” there are 2 items “Liabilities” and “Stockholders Equity”. That to me is more logical than the presentation hierarchy (on the top) whih has “Assets” as a top level item followed by a whole load of “Liabilities & Stockholders Equity” entries presented over multiple lines. It all balances which at the end of the day that is all that matters, but in terms of presentation I prefer the calculation hierarchy.
I would like to say it is like that for all companies but it isn’t. Have a look at the SEC filing available on Edgar Online (image on the left). At least my prototype is showing the hierarchy ;-). Go here for the original.
Summary of issues (opportunities?):
- Companies not adhering to the standard taxonomy structure.
- Companies adding in facts from other reports rather that using the correct fact.
- Duplicate entries doubling up the values.
- Weird looking presentation hierarchies.