Following SEC’s announcement that filers should pay closer attention to Calculation Relationship errors, we made XBRLAnalyst smart enough to analyze and show in Excel different types of Calculation Relationship errors in XBRL filings (screenshot). It can also help filers understand how to fix such errors in following reports and review draft reports with XBRLAnalyst before submitting them to SEC.
We also looked at all XBRL filings submitted to SEC and understand why SEC decided to take a closer look at the Calculation Relationships. The number of Calculation Relationship errors continues increasing, probably due to detail tagging that small filers had to start in 2013.
On the left plot, we show the number of filings containing errors and calculation warnings in percent of total number of reports filed in that year. Warnings are generated when the filer forgets to provide data for all lineItems involved in Calculation Relationship for some additional reported periods. For example, annual reports often contain Q4 data for the Income Statement but not for every lineItem on the statement, which results in incomplete Calculation sets. Also, the warnings are generated when the filer forgets to include dimensional information for all lineItems involved in a Calculation Relationship if some of them already have dimension members reported.
The right plot shows the number of errors/warnings per filing. It seems that the number of errors remains fairly stable at an average of 4.5 whereas the number of warning per filing grows really fast, which is probably related to the fact that the number of Tagged Note Details grows fast since 2011 when filers had to start detailed tagging phase.
The graph on the top shows the number of filings with some issues by software vendor (ordered by market share). Clearly, filers do not select the software based on its Quality Assurance capabilities. Also, there is just a handful of software types that can prevent Calculation Relationship errors well enough (highlighted in green). So, we ask an obvious question if some regulator should start certifying XBRL software vendors if the “efficient” market fails to select the best ones due to all kinds of reasons.