BLOG: Digital Financial Reporting
This is a blog for information relating to digital financial reporting. It is for innovators and early adopters who are ushering in a new era of digital financial reporting.
Much of the information contained in this blog is summarized, condensed, better organized and articulated in my book XBRL for Dummies and in the three documents on this digital financial reporting page.
This is probably one of the best examples of the problems caused when a machine is forced to imply information. It is an example of where one error can lead to another error which masks the first error. It is also an example of the consequences which occur when a filer in essence redefines the meaning of a concept by using a concept in a different way than it was intended.
I will use this AT&T filing to make my point. I am not picking on AT&T, it is just that I happen to run across this in my tuning of the fundamental accounting concept impute rules.
If you look at the income statement of AT&T, focusing on the operating income section:
The income statement looks straight-forward enough. Note the line item "Total operating expenses". If you examine the concept using the SEC interactive data viewer, you see that the filer used the concept us-gaap:OperatingExpenses to represent that fact:
However, if you examine the US GAAP XBRL Taxonomy, you notice that the concept Operating Expenses does not include Cost of Revenue. The total of Cost of Revenue and Operating Expenses is the concept Costs and Expenses:
However, if you take a look at the AT&T income statement again, AT&T included the line item "Costs of services and sales" within the line item Total operating expenses.
So, the problem is that AT&T should have used the concept us-gaap:CostsAndExpenses to represent the line item they refer to as "Total operating expenses". While the label is similar to the concept they did use, the meaning of what they are representing matches the concept Costs and Expenses which includes Cost of Revenue.
The problem is that my fundamental accounting concept relations conformance testing did not pick up this error. The impute algorithm imputed a value incorrectly. So, the error made by the filer who used the wrong concept is masked by an error in the impute algorithm which imputed a value incorrectly. The results of my validation report did not show any errors.
But I found this when I looked at the validation report. Now, a human can figure out that an error occurred by looking at the information, but a machine stumbles over this and does not realize that it made an error. But if I never happend to look at the report, I would not detect the mistake.
This is why it is better to explicitly provide information rather than force a machine to try to imply a fact.
XBRL International released a document, XBRL Taxonomy Guidance Document, which is a very good step in the right direction when it comes to building XBRL taxonomies. The focus is using the XBRL technical syntax correctly. That is only a portion what a taxonomy author needs to understand when they are representing information about a business domain in machine-readable form.
Over the years I have run across the following books which are extremely helpful in trying to understand digital financial reporting. I strongly recommend that for anyone who wants to understand digital financial reporting well or who want to build rock-solid products/solutions to read the following books:
- Data and Reality, by William Kent: (First and last chapter are best, entire book is useful) The primary message of the Data and Reality book is in the LAST CHAPTER, Chapter 9: Philosophy. The rest of the book is EXCELLENT for anyone creating a taxonomy and it is good to understand, but what you don't want to do is get discouraged by the detail and then miss the primary point of the book. It provides something useful. The goal is NOT to have endless theoretical/philosophical debates about how things could be. The goal is to create something that WORKS and is USEFUL. A shared view of reality. That enable us to create a common enough shared reality to achieve some working purpose. Actually SEEING it work (i.e. prototype) PROVES that it works, let you see and understand HOW it works, and help one see how to make it work even better.
- Everything is Miscellaneous, by David Wenberger: (Entire book is useful) This is very easy to read book that has two primary messages: (1) Every classification system has problems. The best thing to do is create a flexible enough classification system to let people classify things how THEY might want to classify them, usually in ways unanticipated by the creators of the classification system. (2) The big thing is that it explains the POWER of metadata. First order of order, second order of order, and third order of order.
- Models. Behaving. Badly., by Emanual Derman: (First half of the book is useful) If you read the Financial Report Semantics and Dynamics Theory, you got most of what you need to understand from this book. But the book is still worth reading. It explains extremely well how it is generally one person who puts in a ton of work, figures something out, then expresses extremely complex stuff in terms of a very simple model and then thousands or millions of people can understand that otherwise complex phenomenon.
- Semantic Web for the Working Ontologist, by Dean Allenmang and Jim Hendler: (First two chapters) This is an extremely technical book, but the first chapter (only 11 pages) explains the big picture of "smart applications". It is awesome. It also explains the difference between the power of a query language like SQL (relational database) and a graph pattern matching language (like XQuery). Querying can be an order of magnitude more powerful if the information is organized correctly. That is why picking the correct data storage format is important.
I would recommend reading the books in the order listed. The investment in time to understand this information will avoid going down the wrong path.
Back in April 2014 as part of the minimum criteria testing I was doing, I observed that the percentage of public company filings which conformed to all of the fundamental accounting concept relations was 25.6% (1711 filers).
Last month I pointed out that the percentage of public companies which conformed to all of these relations was 53.1% (3863 filers).
I ran the same test for the last filing of every public company which reports to the SEC today and the percentage that conform to all of the fundamental accounting concept relations is now at 61.9% (4334 filers). You can see the graphic below which breaks this information out by generator (filing agent or software used).
This rapid growth rate is expected to continue for three reasons:
- These relations between fundamental accounting concept relations is not really controversial and it is increasingly easy to understand and correct the errors.
- I have provided a substantial amount of information here which helps those who want to correct these issues. This recently added analysis shows that each of these tests can precisely detect not only nonconformance to these relations, but also generally the reason for nonconformance.
- I know of at least one commercial implementation of software which can be used to detect and therefore correct these issues, XBRL Cloud. This testing will appear on XBRL Cloud's Edgar Dashboard at some point.
If you look at the numbers below and compare them to the prior results you will see that pretty much every generator is improving. I pester software vendors to include this testing in their products to help them understand that their software can help external financial reporting managers avoid errors like these. But more importantly, I am trying to get software vendors to realize that this is only the tip of a much, much bigger iceberg. This is a digital disclosure checklist summary which I created several months ago. This shows you where things are headed.
These rules may seem annoying to software vendors. I can tell you two things. First, external financial reporting managers already have to comply with these rules. They are just doing so using time consuming and expensive manual processes. Second, automate these tasks by leveraging the XBRL-based structured information and you will provide something very useful: reducing the risk of noncompliance.
It is these rules which will make the quality of public company financial information submitted to the SEC to be at a level to make it safely, reliably, predictably usable by machine-based automated processes. If you don't understand digital financial reporting, the risk of becoming a relic of the past is increasing every day.
If you don't want to become a relic of the past era of paper-based financial reporting, understand how digital financial reporting really works. Digital financial reporting is not only inevitable, it is imminent.
If you want to understand the moving parts and you can walk through an Excel macro, the prototype Excel application on a link at the bottom of this page can help you understand the important moving pieces.
Those who study financial reporting might find this information rather interesting, both accountants who prepare financial statements and analysts who interpret those statements. You can find additional information in this analysis here. While this analysis focuses on "Income (Loss) from Equity Method Investments", these same ideas apply to other financial statement line items.
Here are the facts as I see and understand them. Not all economic entities have investments which are accounted for using the equity method, but if an entity does then they most likely report "Income (Loss) from Equity Method Investments". (This is another reference for equity method accounting.) The accounting standards state that this line item must be "reported as a single line item on the income statement." The SEC has additional information as to how this line item is reported in Reg S-X. The FASB Accounting Standards Codification (ASC) summarizes what they think Reg S-X says.
So, I challenge you to sort through all those reference materials to figure out exactly how the line item "Income (loss) from equity method investments" should be reported. In addition to reading through all that material, here is other information which I gathered:
- Multiple sources confirm that while the accounting literature states that this line item must be reported as a single and separate line item, WHERE it must be disclosed is not as clear.
- Some reference material states that reporting this line item with the other "separately reported items" such as discontinued operations and extraordinary items is the best practice.
- The US GAAP XBRL Taxonomy represents this line item in two ways: before taxes as a separate line item and after taxes as a separate line item.
- One highly regarded accountant that I talked to says that this should ONLY be reported and ALWAYS be reported after taxes because the line item can be used to manipulate the effective tax rate.
So, now let us see exactly how public companies report this information. Per an analysis of 9,679 XBRL-based financial filings, 1,048 or about 11% of economic entities reported this line item. Of that 1,048 this is where they reported that line item in their income statement:
- 624 entities (60%) reported the line item before tax directly as part of income (loss) from continuing operations before tax
- 110 entities (10%) reported the line item after tax
- 128 entities (12%) reported the line item as part of nonoperating income (expense)
- 20 entities (2%) reported the line item as part of revenues
- 10 entities (less than 1%) reported the line item as part of costs and expenses
- 8 entities (less than 1%) reported the line item as part of operating expenses
- 2 entities (less than 1%) reported the line item as a sibling to income tax expense (benefit)
- 60 entities (6%) created an extension concept and the line item rolls up to that extension concept
- 86 entities (8%) did something else which was not directly analyzed so exact placement is unknown
The first thing that I learned when I did this analysis is that I did not really understand that this variability was even allowed. Intuitively as a professional accountant, I was surprised. Other accountants I spoke with were likewise surprised that income (loss) from equity method investments could be included within revenues. And I am not saying that any of these reporting entities did anything wrong. I am simply making an observation.
These observations raise the following questions in my mind.
- What is the purpose of this line item bouncing around the income statement so much? Is there a legitimate reason why entities which use US GAAP have so much flexibility with this line item and not nearly the flexibility with other line items? (I have not analyzed 100% of all the fundamental accounting concepts yet, but this line item has the most variability that I have seen thus far.)
- Why exactly does this variability exist for this line item, but other line items do not have so much variability? Are the accounting standards ambiguous? Was it a conscious choice to allow this variability, or was it caused by a sloppily written accounting standard?
- What would happen if someone like the SEC or FASB would say, "This line item always goes after tax with other special reporting items, similar to discontinued operations and extraordinary items." Could the FASB or SEC do this? Should the FASB or SEC do this? Would analysts be happy about this or would they not like this to be forced into one slot on the income statement? I really don't know the appropriate answer to this question, but it seems to be a reasonable question.
- Chevron includes income (loss) from equity method investments within revenues, you can see this on their income statement. Chevron did not feel the need to create an extension concept for us-gaap:Revenues. Chevron clearly changed the definition of that concept by including that line item within revenues. Should Chevron have created an extension concept?
- Exxon did pretty much exactly the same thing Chevron did, including income (loss) from equity method investments as part of revenues, but then the DID create an extension concept to express revenues which included that concept. Who is right, Chevron or Exxon? Are both approaches allowable? Should both approaches be allowable, extend or not extend?
- Caterpillar reported income (loss) from equity method investments after tax, they created an extension concept cat:ProfitOfConsolidatedCompanies with the definition "Income (Loss) from Continuing Operations less Income Taxes and before Income (Loss) from Equity Method Investments" because they feel they moved income (loss) from equity method investments.
- Most of the other 110 who reported this information after tax did NOT create an extension concept.
What is the most interesting about this analysis is the fact that you can even perform this analysis and do so very effeciently and effectively. How? XBRL-enabled structured information. Each financial report alone and all financial reports collectively are data in a database which can be queried, sorted, sliced, diced. That has huge value.
Accountants work hard to represent information within a financial report. Sometimes the reference materials they have to work with are more ambiguous than one would prefer. One thing that is not ambiguous is what reporting entities actually do. External financial reporting managers pay people like Big 4 consultants a lot of money to sort through reporting issues. Smaller CPA firms have to likewise understand how to create financial reports. Actual financial reports are, in my view, the best resource for understanding how to create a financial report.
Certified public accountants have a duty and responsibility to improve the art of accounting. Improving the art of accounting enhances the profession of accounting. Certified public accountants have a duty to serve the public interest. The needs of both the reporting entities who create reports and the investors/analysts who use the reports must be balanced.
To employ the tools of XBRL-based digital financial reporting correctly, professional accountants need to learn the tool of digital financial reporting correctly. The rules are slightly different for human and machine-readable information than they are for human-only readable information the profession has been dealing with for 100 years. Making the change is not hard, it is just a little different. Machines are not smart, they are dumb. Machines only act smart if smart people make the machined serve their needs.
Personally, I don't understand all the answers, but I think I have a lot of very, very good questions.
I don't know, perhaps I am missing something. But, how long do you think that public companies are going to tolerate spending millions of dollars going through an exercise to structure information using XBRL and the information is not of a quality that makes it generally useful?
What do you think the probability is of expanding the market for digital financial reports beyond the 10,000 or so US public companies who are required to use it to the some 30 million private companies in the US and as I understand it 100 million private companies globally if digital financial reports don't work? That probability would be ZERO.
Now, I really don't blame anyone for not believing me that the market for digital financial reporting could be ever be in the realm of 30 million to 100 million entities. Given the fact that pretty much zero SEC filers like their software for creating XBRL and given the poor quality of reported financial information available, why would you believe me? Who would ever think that digital financial reporting could provide benefits?
But what if I am right? What if the quality of public company XBRL-based financial filings does improve? What if using structured information actually did provide value? What if XBRL-based digital financial reporting could be made to work?
If you don't think a market of 30 to 100 million businesses is achievable, stop reading this, I am sure there are some good cat videos on YouTube that you can watch. If you are inspired enough to want to know how to make digital financial reporting work and maybe provide software which is useful, works, and provides value to those creating and consuming financial information, here is some good information for you.
On the page of my blog Understanding SEC XBRL Financial Filings I have summarized issues which cause each of the fundamental accounting concept conformance rules to fail. I have also provided information about why filings do conform to those rules. With the information I am providing, you can basically walk through the entire process of making use of the reported information of XBRL-based financial filings and see for yourself what works, what does not work, and why it does or does not work.
Probably the most useful thing is the overview document which helps you interpret what you are looking at.
I pointed out that the quality of SEC XBRL financial filings in terms of conformance to the fundamental accounting concepts has moved from 25.6% to 53.1%. I predict that by March 1, 2015 (after the year end 10-K filings are in) the conformance rate will be 75% or higher. The rate of conformance will continue to gradually but consistently and with an increasing pace improve. These are not pie in the sky guesses. These predictions are based on good information about things that are already happening.
While this improvement is only one small battle in terms of getting the total quality where it needs to be, it is an important battle. The reason is that everyone and anyone will be able to see exactly what impacts the quality of XBRL-based financial filings in positive ways and what impacts the quality XBRL-based financial filings in negative ways.
And this is not just SEC XBRL-based financial filings. Those are only a means to an end. Because all of this information is available publically everything about this discussion is transparent. The truth is easy to understand, all you need to know is where to look.
Remember, XBRL is a means to an end. None of this is about XBRL really, it is about the benefits of structured information. It is about taking what can only be human-based manual processes today and making some of those processes that can be performed by machines. It is about increasing compliance. It is about reducing risk of noncompliance. It is about making something complicated just a little less complicated. And, should someone want to press the "Save as XBRL..." button in their software, fine. There are zero mandates necessary to expand the market of digital financial reporting. But what is necessary is software that provides real value and that really works.
What really baffles me is why the members of XBRL International and others don't understand that it is in their interest to collaborate to create the necessary inroads into the financial reporting market. Step 1: collaborate and create an XBRL-based structured approach to creating a digital financial report and MAKE SURE IT WORKS CORRECTLY. Step 2: compete. The enemy right now is not each other. The enemy is the status quo.