Abstracted from: The Quality Of Interactive Data: XBRL Versus Compustat, Yahoo Finance, And Google Finance
By: Prof. J. Efrim Boritz and Prof. Won Gyun Nof
University of Waterloo (JEB); Iowa State University (WGN)
Published on Social Science Research Network (www.ssrn.com), 54 pages
Data is not always accurate. The introduction of interactive data for reporting puts investors, analysts, and regulators on the same page when analyzing company financial data. Using XBRL allows multiple users access to the same data. The SEC now posts corporate filings in XBRL, as coded by the issuer. In 2006, Christopher Cox, the SEC chair at the time, estimated that analysts had a 28% error rate when valuing companies. Errors were occurring when filers were incorrectly entering data into their own proprietary systems or trying to fit XBRL into their own longstanding systems. Accounting scholars Efrim Boritz and Won Gyun No wondered whether the data from analysts and aggregators is now as accurate and complete as the interactive data found on the SEC's EDGAR website.
Analyst and aggregator data is inaccurate and incomplete. The authors examined the degree to which the data published by aggregators and redistributors matched the XBRL data filed with the SEC by 75 companies and posted online in 2009-2012. They found a series of problems. Over half of the data provided in the XBRL filings was simply absent from the information available from the three major aggregators/redistributors considered by the authors. On average, up to 4.8% of the data provided by the aggregators/redistributors did not match the data in the XBRL filing on which it was based (when starting with the XBRL filing and then seeking the same data entry at the aggregator/redistributor). Beginning at the aggregator/redistributor site and trying to find the matching term in the XBRL report on the SEC's EDGAR website, the mismatch rate was up to 8%.
XBRL filings offer the better data. Data from the statements of cash flow showed the highest mismatch rate, although the rate varied somewhat by data provider. Over half of these mismatched items were material. Where there were mismatches, those made by the least accurate aggregator/redistributor were material 66% of the time. Of the middle performer's errors, 58.4% were material, and the best performer still made material errors in 45.9% of the mismatches. Generally, 61.5% of the errors in the statements of cash flows were material; in balance sheets, 55.3% of the errors were material; and in income statement, 54.9%. Analysts and aggregators may add perspective to the data, the authors conclude, but the data they provide omits about half of the available information filed by the company, and it is riddled with errors, over half of which are material.
Abstracted from The Quality Of Interactive Data: XBRL Versus Compustat, Yahoo Finance, And Google Finance, published on Social Science Research Network (www.ssrn.com), Social Science Electronic Publishing, 2171 Monroe Avenue, Suite 3, Rochester NY 14618. This working paper is available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2253638.