One of the many benefits of the FFIEC's XBRL-based call report information reporting system was the reduction of mathematical errors in the bank call reports. In the last reporting period of the old system there were 18,000 mathematical errors. In the first reporting period of the new XBRL-based system there were 0 mathematical errors.
How did the FFIEC achieve this result? If you go to the PDF referenced above and compare the old data collection and management process on page 9 to the new data collection and management process on page 10 you can see exactly what the FFIEC did. Basically, the FFIEC moved the validation of information from a manual, after call reports receipt validation process, to a process where the inbound reports were AUTOMATICALLY validated prior to a report being accepted by the FFIEC.
This graphic shows where inbound call report validation exists in the FFIEC's new system which went live in the third quarter of 2005:
That is right, the FFIEC does not accept call reports from banks if the mathematical relations are not what they are expected to be.
So, why is it that the FFIEC can validate the mathematical relations within reports but the SEC cannot verify the high-level relations of reported financial information (i.e. there are mathematical errors in the high-level fundamental accounting concept relations)?
One would thinkthat the answer to this question would be, "Because the FFIEC call reports are basically forms with known relations and validation rules but the financial reports submitted to the SEC are not forms, the information is more variable."
But, that is NOT THE REASON. Yes, it is true that bank call reports are forms and that it is an order of magnitude easier for the FFIEC to verify their reports because they are forms. And yes, it is true that financial reports have significantly more variability.
And so, basically what the SEC needs to do is take that variability into consideration and apply different inbound validation rules to different reports; but fundamentally, the SEC can achieve the same result as the FFIEC. This blog post explains a technique for achieving exactly that result.
Basically, rather than have ONE set of business rules like the FFIEC can because call report information is a form; the SEC needs to have separate sets of validation rules for each reporting style used by public companies. As I have shown for both US GAAP and IFRS, about 80% of reports fit into less than 20 different reproting styles.
What if someone like the SEC and/or FASB and/or IFRS Foundation made these reporting rules, the fundamental accounting concept relations of high-level information available in machine-readable form? What if a public company submitting a report had to "declare" their reporting style or "configure" the high-level relations to indicate what relations they were using in their report?
Or, what if the FASB/IASB published these rules in machine-readable form in addition to the information that they make available in books for humans to read. What if software vendors could allow the accountants creating reports to configure these rules to help the creator of the report verify that the report is being created correctly?
Basically, the variability of reports needs to be managed in some manner. The variability is not random. The term generally accepted accounting principles(GAAP) is used for a reason. There are standards for reporting financial information. Again, yes, variability is allowed. I would point out that financial institutions that report to the SEC consistently use the same interest based revenues reporting style.
There are many technical approaches that could be used to manage this variability such as the Zachman framework. I am sure there are plenty of other approaches.
Bottom line: So here is the bottom line. There are a lot of different ways that this could be done, but the financial reports submitted to the SEC can be verified for consistency with expected high-level financial accounting relationships in reported information. It is an order of magnitute harder than what the FFIEC's situation, but it is very achievable. The question is what is the best approach for achieving this objective and improving information quality.