Written by Bob Schneider Posted on July 30, 2010
A few weeks ago, Haaretz, Israel’s oldest daily newspaper, published an article about the country's adoption of XBRL with the chilling headline A third of Israeli companies get int'l financial reports wrong. Summarizing the findings of a recently published study by a trio of researchers, the piece states:
The ISA [Israel Securities Authority] had hoped that Israel would be a light unto the nations. Barely a flicker, really: In practice the project has been more reminiscent of the Tower of Babel…34% of the 2008 corporate reports showed inconsistencies between the XBRL reports and the Hebrew versions. In 11% data was missing, and in 28%, there was a mismatch in the content.
The article includes this eyebrow-raising quote from Ariel Markelevich, a member of the research team:
...There were cases in which companies reported losses per share as profits per share in XBRL, and a case in which a company put its data for 2008 in its balance sheet, but in the middle of the collection of numbers, it switched to figures from 2007. There were 15 companies that simply didn't report a profit and loss statement in XBRL.
In the paper, which is available online, the authors describe how XBRL statements are prepared in Israel. Filers neither map or tag their financials; instead, they enter line-item data (footnotes are largely excluded) into a form at the ISA's TOFES website. The ISA converts the company data to XBRL and uploads it to its MAGNA reporting site. Importantly, the authors note that “...a similar data entry form has existed on the ISA’s website for a while. The only difference seems to be that the ISA is now converting this data to XBRL format.”
After reading the paper a couple of times, I’m uncertain how things went wrong. Some of the mistakes described -- such as a reversal of signs that would, for example, turn an EPS loss into a profit -- have been well documented in XBRL conversions. The SEC (among others) has cautioned filers about the potential for error, and these mistakes are unlikely to reappear in any number in future XBRL exhibits. Wholesale mistakes such as the complete absence of any P&L numbers at all, however, are just baffling.
But for argument’s sake, let’s assume the errors in the XBRL statements are as serious and as extensive as the authors contend. What does the Israeli experience tell us about the utility of XBRL adoption, and what lessons does it hold for regulatory agencies throughout the world that are implementing XBRL reporting?
Very little. As the authors make clear, the “unique process” by which XBRL statements are prepared in Israel -- with filers doing no tagging, and the ISA itself performing the XBRL conversion -- is poles apart from that of SEC and many other national registrars. Because Israel’s method of implementing XBRL bears no resemblance to the way other regulators are pursuing XBRL adoption, drawing any conclusions about the viability of XBRL statements from the Israeli experience is impossible.
Nevertheless, the authors occasionally imply such a connection. Although stating the uniqueness of Israeli adoption, they spend much time discussing the SEC experience, focusing almost exclusively on the Voluntary Filing Program (VFP) and the mistakes researchers have found in participants’ XBRL exhibits. Based on this evidence, the authors state that “Due to the amount of human intervention (manual or software tagging), the reliability of XBRL filings currently made available is highly worrisome.”
That’s like a baseball writer in mid-May forecasting that, given its lousy spring training record, a team will finish last in their division -- when they’re already 26 and 14 in the regular season. The largest companies have now been filing XBRL exhibits for a year, and by most accounts adoption has gone relatively smoothly with low data error rates. (Footnote data may prove more challenging, but, as noted, the ISA adoption did not include footnote disclosures.)
I understand the needs of researchers, who are eager to analyze whatever large sets of data are available. And it’s both reasonable and useful to discuss mistakes in XBRL financials prepared under such programs.
But the VFP was specifically set up so that companies could gain experience with XBRL without worrying about getting everything right. Assessing the overall viability of XBRL adoption for financial reporting on this data is not only wrongheaded: it can also discourage regulators from inaugurating such programs, and companies from participating in them.