The Benefits of Tagged Data: A Short Case Study

Most data filed with the SEC is not "tagged" or structured. As a result, the data is not machine readable and any analysis must take place one filing at a time.

Financial statements are a significant exception. In 2009, the SEC put in place rules that required companies to file financial statements using XBRL. Companies were phased in, with the last tranche subject to the requirements in 2012.

Tagged data has many advantages. In particular it facilitates transparency by making data cheaper and easier to read and analyze. Nonetheless, there is resistance. Some in Congress are seeking to legislatively prohibit the SEC from requiring smaller companies to use XBRL. The approach is designed to provide smaller issuers (below $250 million) with a vague and potentially illusory short term savings (whatever the net costs, if any, associated with the tagging process) at a long term cost of reducing transparency and ultimately making investment more difficult.

With that in mind, we turn to the article in the WSJ on the practice by public companies of leaving taxable income overseas. According to the article, companies do so to avoid taxes in the United States. One consequence, however, is that companies have an incentive to borrow rather than bring the cash home. See Id.  ("Few corporations publicly say they are borrowing to avoid a tax hit, but analysts and economists say the dynamic is clear."). As a result, companies can simultaneously have record amounts of debt and record amounts of cash.   

The point is an interesting one but it was made more interesting by observations that appear to have been developed through an analysis of tagged data. According to the article: "Among more than 240 companies disclosing increases in unremitted foreign earnings in 2013, those with bigger increases tended to also see bigger increases in corporate debt, according to data from research firm Calcbench." 

The information, therefore, came from Calcbench. Calcbench operates an interactive data platform that permits analysis of the XBRL data provided to the SEC. The platform was able to identify the 240 companies filing financial statements with the SEC in 2013 that increased their unremitted foreign earnings.  

That straightforward sentence would have been impossible to make (at least in a cost effective manner) in an analog (that is, HTML) universe. Someone would have needed to pull every financial statement filed with the SEC in 2013 and every financial statement filed in 2012. Given the 10,000 or so SEC filers, this would have required the examination of some 20,000 sets of financial statements. They would then need to be compared to determine the companies that saw an increase in unremitted foreign earnings. 

This task, something that would probably take hundreds of hours, would produce a universe of 240 companies.  The next step would be to examine the filings to determine the relationship between foreign earnings and corporate debt. 

So a task that would potentially take hundreds of hours was likely done in minutes (possibly seconds). An investor in Japan might have undertaken this analysis to determine where to invest. Companies keeping cash abroad but taking on more debt may well have different risk profiles than those that bring the money home and avoid the debt (or at least it could be a factor in the risk profile). 

Yet if some in Congress have their way, the Japanese investor will not be able to easily conduct this analysis for emerging growth companies. With less information, the investor will presumably invest in other places.    

The SEC's Investor Advisory Committee has recommended that the SEC increase efforts to require that filed data be "tagged" or structured. Particularly with the ongoing evaluation of corporate disclosure, tagging should become the default. The benefits are becoming increasingly obvious. 

The ABA Journal is again accepting nominations for their Blawg 100. Please consider nominating the Race to the Bottom.  Instructions for doing so are here.