Digital Financial Reporting

For the first two quarters of SEC flings of XBRL documents I have taken a look at those documents to learn more about XBRL. What I am learning is quite helpful and the filings provide a lot of insight into the realities of using XBRL for financial reporting. I hope to document what I have learned to help those who desire to create quality XBRL filings.

Accounting Trends and Techniques, published by the AICPA, is a survey of the accounting and disclosure characteristics of corporate annual reports. Accountants have a notion called common practice which guides how some aspects of financial reporting are carried out, even though there are no laws pushing the practices.  Technical people refer to these sorts of things as best practices.  These practices are important to get the most out of XBRL. Accounting also has what is called a model financial statement which helps financial statement creators understand how to best put together a financial statement.  We need model XBRL filings also; this is the closest to such a model that I can point to today.

So I started asking myself what these practices might be, what are the criteria one might use to evaluate these SEC XBRL filings and how might those criteria be communicated.  One of the first things to pop into my mind was the Gartner Magic Quadrant.

I thought about what the axes of the magic quadrant might be.  Perhaps technical interoperability/usability on the vertical axis and information comparabilityon the horizontal axis.  What I mean by technical interoperability/usability is can best be seen by looking at one filing.  Looking at one filing ask yourself, "How well can this filing be used across different software applications."  By information compatibility look at a number of different filings and ask yourself "How well can one filing be compared to another filing."  Maybe those are not the correct axes of the magic quadrant, I am sure that the correct axes will reveal themselves as this is looked at more.

I believe that the criteria for evaluating an SEC XBRL filing can be boiled down into one key notion: investor friendliness. How "friendly" is the XBRL filing?  How easy is it for an investor to make use of to achieve what they desire to achieve?

But to really understand what I mean by investor friendliness, technical interoperability/usability and information comparability, it is best to look at some detail.  Passing the minimum hurtle of passing the SEC XBRL submission validation is really only the beginning.  First off, that minimum set of validation provides no real differentiation of XBRL submissions.  Further, the SEC and in general most everyone is still learning about XBRL and how to best use it.  The SEC use of XBRL pushes XBRL into new frontiers because of how the SEC is making use of XBRL's extensibility features. The SEC has added new validation rules each submission period, trying to understand the new rules which will most certainly be implemented is helpful.

So what might some of these criteria for evaluating investor friendliness be?

• How easy is it for an investor to compare a filing with other filings?Some things make comparing information easier, others make them harder.  For example, adding a bunch of new high level concepts to replace existing US GAAP Taxonomy concepts make comparability harder.  Not doing so makes comparability easier.
• Do the numbers add up?Trying to build a benchmarking model using information where the numbers don't add up because XBRL calculations have inconsistencies make is harder to use a filing, having consistent calculations make automating processes easier.
• Do ALL the numbers add up?It is not possible to test all computations using XBRL calculations.  XBRL Formulas are necessary to validate cross context computations such as those in a [Roll Forward]. Even if the SEC does not require or does not allow the submission of XBRL Formulas yet, it is almost a sure thing that they will.  And even if they never do, creating such validation used in the creation of an SEC XBRL filing is prudent to be certain that all the numbers do, in fact, add up.  It may be a good thing to take this one step further and make these validation rules available on the filer's web site.
• Are filer extensions consistent with the underlying US GAAP Taxonomy?What reason would a filer have for not being consistent with the base US GAAP Taxonomy? Why would the US GAAP Taxonomy go through so much effort to, say, make the [Table]s in that taxonomy always have at least one [Axis] and always have [Line Items]; but then it be a good thing for an SEC XBRL filer to break that rule?  This is not a good thing, even if the SEC does not have a validation rule requiring this practice.  The same is true for [Roll Forward]s and other modeling patterns found within the US GAAP Taxonomy.  Having a free for all in terms of how to model a taxonomy is not a good thing.
• Are extension concepts well documented?This falls into the category of consistency with the US GAAP Taxonomy but is worth specifically pointing out.  The US GAAP Taxonomy does a nice job explaining the concepts within that taxonomy using documentation and references to the US GAAP literature.  The documentation is helpful in understanding the taxonomy.  Seems logical that an SEC filer who adds an extension concept would document the concept they added and in certain cases explain why the concept they added was different than the existing US GAAP Taxonomy concepts.

So those are the fundamental criteria which I might suggest as appropriate in evaluating investor friendliness of SEC XBRL filings.  There are probably others. 

What criteria do you see?  Do you see some other overarching criteria which is even better than investor friendliness?  If you were to create a magic quadrant, what would your axes be?

(Note that I have added a new category to my blog: Creating Investor Friendly SEC XBRL Filings. Watch for information on how to create high quality, investor friendly XBRL!)

This Excel spreadsheet has a matrix of the SEC XBRL filers and certain concepts which I would expect to appear in an XBRL instance of a 10-Q or 10-K SEC filing.  The spreadsheet is far from perfect, but it can help to see some things.

First, one error that I have in my spreadsheet is that I am looking for a specific period end date, in my case 2009-09-30, the ending date of the quarter.  But there is a problem with this: filers don't all have fiscal year ends of December 31. The breakdown of my set of 403 filings is this from a query I created:

--12-31      345
--06-30      10
--12-26      9
--03-31      6
--05-31      6
--01-02      3
--01-03      3
--02-27      3
--07-03      3
--11-30      2
--12-27      2
--09-26      1
--09-27      1
--10-31      1
--01-01      1
--11-27      1
--04-02      1
--03-28      1
--12-25      1
--01-31      1
--12-28      1
--07-02      1

The strange looking fiscal year ends are 52/53 week year ends commonly used by retailers.  This makes it harder to get data out of the filings because you cannot look up "all the third quater filings" as XBRL has no way to indicate "this is the third quarter.  You have to figure that out.

The other thing which mucks up the data is extensions created by filers and flexibility in the concepts being reported.  For example, whereas about 375 filers reported cash as the concept "us-gaap:CashAndCashEquivalentsAtCarryingValue", the rest did not; they used some other concept for cash which is other than what the US GAAP Taxonomy uses.

Is that flexibility a good thing or a bad thing?  Well, from my perspective, what I see is that if there is not enough consistency at the high levels of financial reporting, there will be either little hope for compatibility and/or a number of software vendors (such as data aggregators) serving as middlemen to sort the inconsistencies out using software programs.  In my view, that would not be a good thing.

I personally think that there should be a "superstructure" into which everything else fits.  How much superstructure is what needs to be figured out.  This is not about creating a fixed chart of accounts or a form, this is about exactly the opposite.  It is about creating enought superstructure so people don't get fed up with the lack of comparability due to the inconsistencies.

Not a lot, things like perhaps:

• Balance sheet (using an industry which uses a classified balance sheet):Current assets, noncurrent assets, total assets, current liabilities, noncurrent liabilities, total liabilities, total equity.
• Cash flow statement:  Net Cash Provided Operating Activities, Net Cash Provided by Financing Activities, Net Cash Provided by Financing Activities, Net Changes in Cash total, Effect Of Exchange Rate On Cash And Cash Equivalents, Net Cash Provided by Discontinued operations, and Cash an Cash Equivalents.
• Income statement: This is a little harder but maybe Net income attributable to common stockholders, Basic earnings per share, Diluted earnings per share, weighted average shares outstanding basic, weighted average shares outstanding diluted, Income from operations, and such.
• Policies and Disclosures: There are common ones, this may provide some clues. Different industries have different sets.  Certainly thing such as revenue recognition, maturities of long term debt, and other standard stuff.

Then, everything would fit into that superstructure.  You could also change the superstructure maybe, but there would be rigid rules for doing so.  It is that superstructure and the consistency it offers which enables the ease of use and comparability needed by investors and analysts.  Flexibility would still exist.

The data that I see shows that we are actually reasonably close.  We do need to figure out the extension process, being more clear on when, where, and how to extend.

Again, the point is not to create a chart of accounts.  The point is to avoid a random free-for-all.  And actually, I don't see a free-for-all, what I see in the filings I have looked at thus far is a lot of consistency and a few bad apples.

OK, this is a good example of something which I have seen over and over in building XBRL taxonomies relating to financial reporting.  This issue is NOT just for the cash flow statement, but rather I am using the cash flow statement to point out the issue.

The question is this: How much leeway should those reporting financial information really be allowed? I am NOT saying that I know the answer to this question.  I can say that I do have an opinion.  What I have seen over and over in building the IFRS and US GAAP financial reporting taxonomies is that there are accountants who misinterpret reporting rules and create reporting inconsistencies as a result.

Here is the what I see.  The US GAAP Taxonomy for the indirect cash flow statement looks like this (this is the commercial and industrial companies taxonomy, basically all the taxonomies look like this in the area being discussed). Here is a view of the calculation relationsto come up with the net change in cash from operating, investing, financing activities as well as the exchange gain on foreign currency transactions. (OK, so here is the presentation view in that same tool.)

OK, so now I am a little confused.  If you look at line number 633 on the first report, you see discontinued operations.  These concepts appear both within operating cash flows from operating (line 316), investing (line 495), and financing activities (line 661) AND separately (see line 664, 665, and 666 on the report).

From the taxonomy, you cannot tell with 100% certainty cash flows from discontinued operations (line 667) is included in the reconciliation of the beginning value of cash to the ending value of cash, it is not shown in the calculation linkbase.  I think it should be included.

Two SEC XBRL filers reconfigured their cash flow statements.  See Citigroup and General Electric XBRL taxonomies to see how they structured their indirect cash flow statements.

The issues here to me are:

• The US GAAP Taxonomy provides a concept "us-gaap:CashAndCashEquivalentsAtCarryingValue" on the cash flow statement (and on the balance sheet) which is the cash balance being reconciled; yet some filers use different concepts in the reconcilation including: us-gaap:Cash, us-gaap:CashCashEquivalentsAndFederalFundsSold, us-gaap:CashAndDueFromBanks, us-gaap:CashCashEquivalentsAndShortTermInvestments
• The US GAAP Taxonomy has the total amount being reconciled as the concept "us-gaap:CashAndCashEquivalentsPeriodIncreaseDecrease".  Does that include or exclude discontinued operations AND should filers be allowed to CHANGE wheter that total includes discontinued operations?
• The same deal seems to be happening with the exchange gain, us-gaap:EffectOfExchangeRateOnCashAndCashEquivalents, is that included or is it not? The US GAAP Taxonomy says that it is per the calcuation it shows.  At least one filer does not include that in the computation, but rather moved the relation (and they did not redefine the concept by creating their own concept).

What I am asking is how should these situations be handled, what sort of flexibility is allowed and/or should be allowed, what sort of comparability between these is appropriate for investors, and so forth.  What level of consistency is appropriate?  At what level of a financial is consistency and comparability important?

XBRL will help see these issues.  Based on decisions which are made or not made, a certain level of consistency and comparability will exist.  Will that be the level which is desired?  Who decides these things:  CPAs who prepare these XBRL filings?  The SEC?  Investors?  Analysts?  No one?

Again, I am not providing my opinion here (I do have one), I am simply raising the question.  This question can only be addressed appropriately once it is well understood.

And remember, this set of cash flow statement issues is an example, not the only issue.  Similar issues exist in other areas of the US GAAP Taxonomy and SEC XBRL filings.  Different areas of the taxonomy will likely require different criteria for arriving at a decision, but the issues are quite similar to the cash flow statement example shown above.

Several weeks ago I published pieces of the analysis which I had done on the SEC XBRL filings where were being submitted to the SEC.

I am doing a similar analysis for the second quarter of SEC XBRL filings and you can find information on that here on my blog.  I am doing this a little differently this time, learning from the past analysis.  You can go look at the information, it is all summarized on that link above and you can drill into detail if you want.  I will provide some information below to help you figure out how to navigate and read this information, see that below.  I do want to thank XBRL Cloud for the base of my analysis.  Without that list of all the filings and without some of the heavy lifting which is being done there, I would not be able to do as much as I can.  So thank you to XBRL Cloud.

If you are looking for summaries, this is what you want to check out the following items on this page, which is the same link as above:

• (A). Analysis Comments: Summary of analysis by filing. Shows interesting observations, issues, and specific errors identified within a set of 403 SEC XBRL filing. This is a complete list of SEC XBRL filings analyzed. (Note that this is updated periodically, so check back.)
• (B). Reported Calculation Inconsistency Summary:  Of 403 filings analysed, only 90 produced XBRL calculations inconsistencies. This shows a list of each SEC XBRL filing which has calculation inconsistencies reported by XBRL Cloud. This does NOT necessarily mean that the calculation are not valid, but rather point to filings which are being investigated further to determine if calculations are valid. See the analysis comments for a list of any confirmed calculation inconsistencies.
• (C). Reported Test XBRL Formula Issues Summary: Of 403 filings analyzed, only 55 produced XBRL Formula validation results which were different than the expected results. This shows a list of each SEC XBRL filing which had an unexpected result for one XBRL Formula which was used to test the cash flow statement roll forward. This does NOT mean that the filer had an error, but rather it identified where a filer used different concepts then were provided for this roll forward in the US GAAP Taxonomy, or they changed a computation in some manner, or the added additional concepts to the computation for some reason. See the analysis comments above for resolution of these issues.
• (D). Validation Errors Reported by XBRL CLoud Summary: Of 403 filings analyzed, only 28 produced XBRL Cloud validation errors. These errors were not detected by SEC validation upon submission for a number possible reasons including different interpretations of the EFM by XBRL Cloud and the SEC validator, different interpretations by a software vendor and XBRL Cloud, and other such issues. These issues should eventually be resolved as the SEC validation conformance suite continues to be built out, software vendors resolve these interoperability issues, etc. This list does NOT mean that the filer had an error, but rather it identifies possible software interoperability issues which should be resolved. See the analysis comments above for resolution of these issues.
• (E). Information Model Summary: Of 403 filings analyzed, a hand full appear to have used the [Table], [Axis], [Line Items] concepts consistently with the way the US GAAP Taxonomy was created and consistent with the US GAAP Taxonomy Architecture (see section 4.5 Implementation of Tables, page 38). There appears to be no guidance coming from the SEC as to how to properly model [Table]s, thus the variety. This analysis is making no judgment as to what is correct or incorrect, only pointing out the wide variety that [Table]s are being used. It is likely there are similar issues relating to the [Roll Forward] and other patterns within the US GAAP Taxonomy.
• (F). Excel Data Analysis Demo: This Excel spreadsheet allows someone to extract information from the set of 403 SEC filings. The macros used provide examples of how to get at the information in an SEC XBRL filing without using an XBRL processor. (Clearly you can do vastly more with an XBRL processor, but the point of this is to see what you can do without an XBRL processor.)
• (G). Excel spreadsheet with XBRL Cloud Error Information Sliced and Diced: This Excel spreadsheet slices and dices the information reported by XBRL Cloud. It focuses on errors and calculation inconsistencies. No informational messages, warnings, and best practices.

I am thinking about taking this information, condensing it, and providing it in the form of a high level summary of this detailed information.  Not sure if I will, or will not, undertake that task.  If you have an opinion, let me know.

Here is a little demonstration of what one can do with XBRL. Take a look at this link.  The link is to an interface which I created which lets you do cross company comparisons of information filed by a set of 403 SEC XBRL filers. Here I am focusing on text blocks used to report policies or disclosures.

What I did was look through the filed data, found text blocks which were reported a lot by filers, and then picked a set of them to show in this demo.  I had to be a little selective to get the demo to work right.  Some of the comparisons resulted in rather large files (22 MEG, for example with 332 filers reporting "us-gaap:SegmentReportingDisclosureTextBlock", so I did not use those large files). Rather than try and make this demo scale to work with large files, I simply did not use the large files.  Again, this is a demonstration.

Using the XBRL in this way is not limited to analysis of the SEC filings by investors.  Consider this the following other use case.

If you are an accountant, you have probably heard of the AICPA Publication Accounting Trends and Techniques. If you are not familiar with that publication, it is basically a survey of the reporting practices of about 600 filers, created by grabbing SEC filings and reorganizing the information to compare companies.  Basically this is a form of best practices in financial reporting.  I worked at Price Waterhouse years ago and we had an internal publication which was similar.

Leveraging XBRL, imagine being able to build an interface into the SEC XBRL filings and be able to automatically pull information from the filings.  Imagine being able to organize this by industry (i.e. SIC code maybe).  Imagine not being limited to the 600 companies selected for the AICPA publication due to the manual nature of that process. Both attorneys and accountants who create SEC filings use this information to help figure out how to create disclosures.

Further, this process is not unique to SEC filings.  There are many, many other forms of reports and filings where a similar workflow occurs.  Being able to slice and dice this filing information allows for all sorts of interesting possibilities.

* This is the list of text blocks I am using in my demo.

us-gaap:InventoryDisclosureTextBlock (138 Filers)
us-gaap:ScheduleOfSubsequentEventsTextBlock (138 Filers)
us-gaap:SignificantAccountingPoliciesTextBlock (111 Filers)
us-gaap:BusinessCombinationDisclosureTextBlock (104 Filers)
us-gaap:DisposalGroupsIncludingDiscontinuedOperationsDisclosureTextBlock (89 Filers)
us-gaap:LongTermDebtTextBlock (66 Filers)
us-gaap:PropertyPlantAndEquipmentDisclosureTextBlock (65 Filers)
us-gaap:RelatedPartyTransactionsDisclosureTextBlock (53 Filers)
us-gaap:CashFlowSupplementalDisclosuresTextBlock (51 Filers)
us-gaap:DisclosureOfShareBasedCompensationArrangementsByShareBasedPaymentAwardTextBlock (49 Filers)
us-gaap:EquityMethodInvestmentsDisclosureTextBlock (43 Filers)
us-gaap:ScheduleOfVariableInterestEntitiesTextBlock (41 Filers)
us-gaap:OtherIncomeAndOtherExpenseDisclosureTextBlock (39 Filers)
us-gaap:LoansNotesTradeAndOtherReceivablesDisclosureTextBlock (35 Filers)
us-gaap:ScheduleOfGuaranteeObligationsTextBlock (32 Filers)
us-gaap:ScheduleOfGoodwillTextBlock (30 Filers)
us-gaap:ScheduleOfLossContingenciesByContingencyTextBlock (26 Filers)
us-gaap:AvailableForSaleSecuritiesTextBlock (25 Filers)
us-gaap:AssetRetirementObligationDisclosureTextBlock (23 Filers)
us-gaap:MinorityInterestDisclosureTextBlock (21 Filers)
us-gaap:AccountsPayableAndAccruedLiabilitiesDisclosureTextBlock (20 Filers)
us-gaap:IntangibleAssetsDisclosureTextBlock (20 Filers)
us-gaap:ProductWarrantyDisclosureTextBlock (19 Filers)
us-gaap:ScheduleOfLineOfCreditFacilitiesTextBlock (18 Filers)
us-gaap:ScheduleOfAccountsNotesLoansAndFinancingReceivableTextBlock (17 Filers)
us-gaap:FairValueByBalanceSheetGroupingTextBlock (16 Filers)
us-gaap:ScheduleOfBusinessAcquisitionsByAcquisitionTextBlock (16 Filers)
us-gaap:ScheduleOfSegmentReportingInformationBySegmentTextBlock (16 Filers)
us-gaap:ScheduleOfTreasuryStockByClassTextBlock (16 Filers)
us-gaap:TransfersAndServicingOfFinancialAssetsTextBlock (16 Filers)
us-gaap:ScheduleOfStockByClassTextBlock (15 Filers)
us-gaap:ShortTermDebtTextBlock (14 Filers)
us-gaap:OtherLiabilitiesDisclosureTextBlock (13 Filers)
us-gaap:MarketableSecuritiesTextBlock (12 Filers)
us-gaap:CollaborativeArrangementDisclosureTextBlock (11 Filers)
us-gaap:PublicUtilitiesDisclosureTextBlock (11 Filers)
us-gaap:InterestIncomeAndInterestExpenseDisclosureTextBlock (10 Filers)
us-gaap:PostemploymentBenefitsDisclosureTextBlock (10 Filers)
us-gaap:ScheduleOfRestructuringAndRelatedCostsTextBlock (10 Filers)
us-gaap:CommitmentsDisclosureTextBlock (9 Filers)
us-gaap:ScheduleOfDefinedBenefitPlansDisclosuresTextBlock (9 Filers)
us-gaap:ConcentrationRiskDisclosureTextBlock (8 Filers)
us-gaap:DetailsOfImpairmentOfLongLivedAssetsHeldAndUsedByAssetTextBlock (8 Filers)
us-gaap:ScheduleOfDerivativeInstrumentsTextBlock (8 Filers)
us-gaap:ScheduleOfOtherAssetsNoncurrentTextBlock (8 Filers)
us-gaap:AccountingChangesAndErrorCorrectionsTextBlock (7 Filers)
us-gaap:DeferredCostsCapitalizedPrepaidAndOtherAssetsDisclosureTextBlock (7 Filers)
us-gaap:EnvironmentalLossContingencyDisclosureTextBlock (7 Filers)
us-gaap:PartnersCapitalNotesDisclosureTextBlock (7 Filers)
us-gaap:UnusualOrInfrequentItemsDisclosureTextBlock (7 Filers)
us-gaap:DisclosureOfLongLivedAssetsHeldForSaleTextBlock (6 Filers)
us-gaap:FullCostMethodOfAccountingForInvestmentsInOilAndGasPropertiesDisclosureTextBlock (6 Filers)
us-gaap:RealEstateDisclosureTextBlock (6 Filers)
us-gaap:ScheduleOfRestrictedCashAndCashEquivalentsTextBlock (6 Filers)
us-gaap:CashAndCashEquivalentsDisclosureTextBlock (5 Filers)
us-gaap:DiscontinuedOperationsPolicyTextBlock (5 Filers)
us-gaap:EarningsPerSharePolicyTextBlock (5 Filers)
us-gaap:EquityMethodInvestmentsTextBlock (5 Filers)
us-gaap:FairValueOptionTextBlock (5 Filers)
us-gaap:PropertyPlantAndEquipmentTextBlock (5 Filers)
us-gaap:QuarterlyFinancialInformationTextBlock (5 Filers)
us-gaap:ScheduleOfDividendsPayableTextBlock (5 Filers)
us-gaap:ScheduleOfEarningsPerShareDilutedByCommonClassTextBlock (5 Filers)
us-gaap:ScheduleOfFinancialInstrumentsOwnedAndPledgedAsCollateralTextBlock (5 Filers)
us-gaap:ScheduleOfOtherNonoperatingIncomeByComponentTextBlock (5 Filers)