As many as one in eight XBRL filings contains at least one problem in scaling, a simple mistake that could threaten the quality of investment recommendations and increase liability, according to a recent analysis.

Calcbench, a technology firm promoting XBRL, says 8 to 12 percent of all filings provide the wrong scale for a number -- such as reporting a number as 15 when the correct figure is 15,000. That mistake was most common in the fourth quarter of 2012, when one in eight filings contained a scaling problem, the firm says.

Scaling most often occurs in tags associated with shares, but “a non-trivial number of errors” also occur in areas that are watched closely by analysts and investors, like revenue, net income, and assets. “Errors in these accounts may cause potentially wrong investment recommendations and decisions which may lead to increased liability by filing firms,” the report says. That makes them a high priority for correction, according to Calcbench.

Even more common, says the report, are sign switches, a problem in 40 to 60 percent of filings over the period analyzed by the firm. Sign switches are not as right or wrong as scaling errors, the firm says, but they can be confusing. As an example, cost of goods sold might be presented as a negative number that is added to revenue, or as a positive number that is subtracted from revenue. The analysis also finds a correlation between the presence of sign switches and the average number of tags in a filing. “The more tags you use, the more likely you are to have a sign switch,” the report says.

Calcbench initiated the analysis to address concerns that have persisted over the quality of data that is presented through XBRL filings, says CEO and co-founder Pranav Ghai. Business advocates have lobbied Congress to exempt smaller companies for the XBRL filing requirement, a measure that has been both hot and cold at different times. The Securities and Exchange Commission has called on companies to correct errors but has not taken an aggressive stand on the issue.

“The fact is some issues with data quality are in very visible places,” he says. “It's just carelessness. But when an investor or an analysts sees that carelessness with this particular number, they don't care if the company got the product warranty accruals right. Once people realize this and start fixing the really easy stuff, they're going to realize how valuable this data really is.”

      

Topics