XBRL-based structured digital financial reporting - Blog: Digital Financial Reporting (using XBRL)

I had this idea for an analysis to perform on the SEC XBRL filings in order to see what they were looking like and to test some things I was fiddling with. The analysis vastly exceeded my expectations in terms of information it provides and it is incredibly useful to seeing the types of issues accountants are going to have to deal with. This blog post summarizes the results of that analysis. I have a detailed document which provides additional insight, if you want a copy of that send me an email and I may be able to get that detail to you.

The analysis looks at 408 SEC filings of 10Q's all filed after July 1, 2009. This is the complete list of filings: http://www.xbrlsite.com/demos/console/console.htm. This list came from XBRL Cloud's list of SEC filings.

I established a hypothesis that if I ran an XBRL Formula through each SEC XBRL instance (10Qs) that the formula (business rule) should either support the fact that cash reconciles or there should be some reason that the formula does not support that and there would be a specific reason as to why. Another way to state this is to say that "beginning balance of cash + changes in cash = ending balance of cash". The US GAAP Taxonomy states this, although implicitly; you can see that here in the US GAAP Taxonomy CI entry point. (What I mean by implicitly is that there is no official business rule which states this and there is only the implicitly way that a [Roll forward] is created which unofficially implies this, it could be explicit though.)

So, I created this business ruleusing XBRL Formulas. One of the first thing I recognized was that filers were using two different US GAAP Taxonomies (2009 and 2008) so I had to create another XBRL Formulafor the second taxonomy. I then used the complete list of SEC filings (mentioned above) and created a little batch file which grabbed the SEC XBRL instance from the list, applied the business rule, and spit out a validation report in XML which I then converted to HTML. I did this using the UBmatrix XBRL Processing Engine. This is a summary of the raw validation results.

The really interesting thing is that of the 408 filings, 358 passed validation first time through! I was truly amazed. That mean that 358 of the 408 filers provided the concepts specified by the US GAAP Taxonomy, that the numbers added up correctly, and that there were two periods reported. This part took very little effort to set up, it is just creating the formula, getting the list of filings to run the formula against and generate the batch file to automate the entire process.

The hard part was figuring out why the other 50 did NOT validate. That was the really interesting part o this analysis, but also rather time consuming.

Before I get to the observations there is something that I want to point out. While I did this analysis for the cash flow statement, the same type of analysis could be performed in other areas of the taxonomy. But if the analysis were to be performed, I speculate that the types of issues you would discover would be quite similar to the issues uncovered by my simple analysis. Keep this in the back of your mind as you read these observations.

High level observations

Here are the observations which I made:

Of 408 filings, the cash flow statement was extremely consistent at this high level. For example, all but 6 filers of the 408 total filings had used the concept "us-gaap:CashAndCashEquivalentsPeriodIncreaseDecrease" specified by the US GAAP Taxonomy for net increases or decreases to cash reported on the cash flow statement.
Of the 408 filings, the vast majority of the filers followed the US GAAP Taxonomy which calls for the concept "us-gaap:CashAndCashEquivalentsAtCarryingValue" to be used on the cash flow statement. (358 plus, did not to an actual count)
Filers who did NOT use the concept "us-gaap:CashAndCashEquivalentsAtCarryingValue" generally used another US GAAP concept which generally included one of the following: us-gaap:Cash (5 filers), gaap:CashCashEquivalentsAndFederalFundsSold (1 filer), us-gaap:CashAndDueFromBanks (8 filers), us-gaap:CashCashEquivalentsAndShortTermInvestments (4 filers).
Only 5 filers created extension concepts for the balance of cash. Of those, 4 created extension concepts which were different for the beginning and ending balance. (i.e. one concept for the beginning balance, a different concept for the ending balance). This is a big no-no in my book and the fact that only a hand full did this and that this is NEVER done in the US GAAP Taxonomy itself is pretty good evidence that creating two different concepts for the beginning and ending balance is not the way to go.
The vast majority of filers followed the US GAAP Taxonomy calculation and included the changes in cash related to exchange gains and discontinued operations within the concept "gaap:CashAndCashEquivalentsPeriodIncreaseDecrease"; however, there were a hand full which did not include cash changes from exchange gains or discontinued operations within that concept, rather they were additional reconciling items in calculating the total change. In is hard to believe that calculating this total change in cash should be done in two different ways, but that my just be me.
One filer had an additional reconciling item for VIE related changes to cash (PAPA JOHNS INTERNATIONAL INC). It may be the case that this concept needs to be added to the US GAAP Taxonomy.
Two filers (DTE Energy, PPL Corporation) has "Assets reclassified as held for sale" included in changes to cash. This may need to be added to the US GAAP Taxonomy. I do recall from my work with the IFRS taxonomy that this concept is part of the cash flow statement as these to filers reported.
Two filers (Symantec Corporation, NVIDIA Corporation) had a 52/53 period year end and used the wrong startDate which did not match to the instant reported, therefore the XBRL Formulas did not fire. When the startDate was corrected and the validation was rerun, the validation worked correctly. These are pretty obvious errors.
Of the 408 filings, only 2 filings had "rounding issues" which caused the computations to not add up correctly for this one formula. Personally, I belive that filers should make all these adjustments BEFORE they put their numbers in the XBRL instance because sure, it may be some additional work for the filer, but it will be even more work for all the analysts who make use of this information, trying to deal with these rounding issues. Again, that is my personal view.

It seems that comparability would be extremely easy to achieve for the high level of the cash flow statement. The biggest issue to comparability are (a) the use by a minority of filers of different concepts to represent the balance of "cash" than the US GAAP Taxonomy calls for and (b) where the exchange gains and discontinued operations really should be in the computation of the net change in cash. These are both accounting issues, not technical issues.

Further, looking one level deeper into the cash flow statement, a high percentage of filers (over 95%) report the concepts: us-gaap:NetCashProvidedByUsedInInvestingActivities, us-gaap:NetCashProvidedByUsedInFinancingActivities, us-gaap:NetCashProvidedByUsedInOperatingActivities. I may do some additional work to see more precicely what is going on with these concepts.

Example of Issues Accountants Will Have to Deal With

OK, so now to the example of issues accountants will need to deal with. I have my personal opinion on these, but clearly this is not my call. Someone will need to deal with these issues however. Not doing anything is an option, but I don't really thing it is the best option.

Take the issue of where the exchange gains and losses realating to cash are totalled, there seem to be the following options: (a) Force everyone to use the US GAAP Taxonomy computation, (b) allow for filers to change the computation but clearly document that change, (c) do nothing which is really implying that option "b" is the right choice. What sort of comparability should be provided? Who decides?
If someone does move the computation of "us-gaap:CashAndCashEquivalentsPeriodIncreaseDecrease", should they create a new concept or simply change the definition of the existing concept? This debate has existed for years in the XBRL community.
Should prepares be required to report the concept "us-gaap:CashAndCashEquivalentsAtCarryingValue" in their cash flow statement, even though they use some other cash concepts on their balance sheet and don't show that concept at all?
Should the concepts for the changes in assets held for sale and VIE related changes be added to the US GAAP Taxonomy?

Again, those are just a few examples. And remember from the point I made above, while the concepts you might be looking at and the considerations will be different in different areas of the US GAAP Taxonomy, these issues from the cash flow statement are likely quite similar to issues which will be discovered in other areas of SEC XBRL filings. These issues are not just for specific concepts, but more over arching principles which should be applied throughout the entire US GAAP Taxonomy.