Digital Financial Reporting

Someone made me aware of the WICI EBR/Gartner KPI Initiative.  Their web site provides the following description of this project:

The EBRC/Gartner KPI Initiative is a market collaboration to develop standardized key performance indicators initially for the High Technology, Insurance and Retail and Consumer Goods sectors as extensions to the WICI framework. Gartner will facilitate the development of the industry sector standardized KPI concepts via private client groups to include leading company and investor/analyst participants. It is expected that the collaborative KPI effort will take approximately six months from inception.

They appear to have an XBRL taxonomy for the standardized set of KPI concepts.

Someone asked my if I could provide this information for two specific SIC codes.  So what I did was provide the information by SIC.  Here is another version of the research and development information. In this version you get a list by SIC of research and development expenses, total assets, and the percentage of the two; plus you can drill down into the SIC code to see a list of the SEC filings which make up the list.

Also, I aggregated these demos here. Keep watching, I may add some more.

This is starting to get fun! I did another experiment, grabbing the concept us-gaap:ResearchAndDevelopmentExpense from my set of test filings.  Of the 5525 filings, I got 1236 hits (22%). So for example, Microsoft is #1 in R&D Expense spending about $9 Billion between July 1, 2010 and June 30, 2011 (which is about 8% of total assets) per their 10-K filed July 28, 2011. Second is Oracle which spends about half of what Microsoft did for a similar period of time.

If anyone has any query ideas, send me an email and I will see if I can grab the information.

In order to better understand how to create SEC XBRL financial filings and XBRL instances and taxonomies in general, I analyzied 5525 SEC XBRL financial filings.  This analysis builds on another analysis which I did several months back of 1474 SEC XBRL financial filings which pointed out four common areas of confusion in the US GAAP taxonomy and therefore inconsistencies in SEC XBRL financial filings. This post helps to understand and resolve those areas of confusion.

This new analysis of the additional filers, about four times as many, shows the same consistent patterns in filings.  For example:

• The concept us-gaap:Assets is reported by 99% of all filers. The other 1% can be explained. Some filers have only us-gaap:AssetsCurrent and do not provide the concept us-gaap:Assets.  Others report using a net assets presentation format. So basically, there is a 100% complete set of the possibilities, which was expected.
• As you might expect, because the balance sheet balances, the concept us-gaap:LiabilitiesAndStockholdersEquity likewise appears with regularity, again about 99% of the time.  But because some filers are partnerships, the concept us-gaap:LiabilitiesAndPartnersCapital is reported.  Again, 100% of cases can be accounted for.
• The concept us-gaap:CashAndCashEquivalentsPeriodIncreaseDecrease (net change in cash on the cash flow statement) appears with regularity, 95% of the time.  The 5% balance are filers which either do not report a cash flow statement for some reason, use a different concept such as us-gaap:CashPeriodIncreaseDecrease or they create an extension concept for some strange reason or erroneously use the wrong concept.  Again, 100% of all possibilities can be accounted for.
• The income statement is less consistent, but still more consistent than not.  The highest area of variability where one might expect consistency is the reporting of net income (loss).  There are three concepts which are used somewhat inconsistently due to ambiguity in the US GAAP taxonomy it appears.  There are three possible "candidate" concepts which could be chosen, different filers are reaching different conclusions as to which is the appropriate choice.

So what does this mean? My interest is not really the analysis of financial filings but rather the creation of financial filings.  However, as the ability to use of the information within the financial filings is a driving characteristic of financial information; I am indirectly interested in the analysis of the information.

But what I am more interested in is the consistency of the semantics for the purpose of using that consistency to drive the creation of financial reports.  That has been my interest in XML in financial reporting since back in 1998 when I first started to get people interested in XBRL.

That is why this SEC XBRL financial information set is so interesting to me, it basically proves my long held hypothesis.

However, one thing is a bit disappointing to me.  I actually believed that the "individual investor" could and would actually make use of the SEC XBRL information directly.  That will simply not be the case. Despite what many XBRL zealots say, I don't see the SEC XBRL information as being usable by individual investors any time soon, if ever for the following reasons:

1. Because US GAAP is not a "form".  There is simply too much variability allowed for the average person to overcome.  At the same time the variability is not so great that it cannot be overcome with some work. The SEC could improve the situation by revising regulation S-X. Others have called for the SEC to make such revisions even before the creation of the US GAAP Taxonomy. While the SEC does not need to turn US GAAP into a "form", some structure is necessary in order to help the average person use this SEC XBRL financial information.
2. Other variability in financial reports.  One of the biggest is the fiscal period.  The fact that periods can end at different times and things like the 52/53 week year end make using information harder.  There are other things such as this.
3. US GAAP taxonomy ambiguities.  There will likely always be ambiguities in the US GAAP taxonomy. That is just the nature of the beast.  Some will be addressed, others never will.

Those are the realities. You can wish the world were different, but it is not. While I am no expert, my database skills, XML skills, Excel skills, Access skills are likely better than the average investor.  This does NOT mean that the SEC XBRL information is a waste of time. What it means is simply that these average investors are going to need some help by others to organize that data set to make it easy enough for them to use.  That will definitely happen.  Maybe these denormalized and more useable sets will be free, maybe users will be charged some fee; hard to say, probably both.

So there are two lessons here. 

The first lesson is that using SEC financial information can be challenging and the challenges really have little to do with XBRL.  Some, but not really that much.

The second lesson is that if you want to build a system and you want your data set to be easy to use, dig deep into the SEC XBRL financial information because it offers many, many clues which you can use to make your system work better.

I proposed a little experiment on the xbrl-public list.  This is my first stab at providing the results.

The information represents the total assets of the 5,523 (by my count), to the SEC between 2011-04-01 and 2011-08-19 inclusive, 10-Q and 10-K filings only. Total assets is the total per the filing and may not be the same date for each filer.

The experiment is to see if others can get the same result, or rather agree on the parameters and see if we get the same results.

It took approximately 4 hours for my computer program to go through each of the filings and grab the information.  After reading the information, I loaded the information into a database to do the queries.

One interesting observation is that it took my program about 4 hours to run through the full set of filings. To do this by hand I calculate that it would have taken me about 92 hours (about 1 minute per filing, clearly it would have taken more time most likely and would have been an incredibly boring task). Hard to come up with an accurate computation, but I estimate that I saved 96% in the time it took to gather this information.

Frankly, I wonder if anyone has ever been able to do that computation, or would even endeavor to do so. XBRL makes this both easy and possible. It will be interesting to see the many new ways people find to look at this SEC XBRL information set.