About 30,000 state and local governments publish audited financial statements. But because these financial reports are in PDF format, the data is hard to analyze, compare and aggregate. To make government finance more transparent and provide more accessible data to municipal bond investors, we have advocated transitioning from PDFs to a more data-friendly format known as XBRL.
The SEC has required public companies to file their financial statements in XBRL for almost a decade now. XBRL is also used for local government financial reporting in Spain, Brazil and Italy. But, in the US municipal market, progress toward disclosure modernization stalled after the Association of Government Accountants worked with the State of Oregon on a pilot XBRL filing in 2008.
Now, due to work by Reason and others, this idea is starting to attract interest. Perhaps as a result, panelists at an SEC conference on Municipal Disclosure were asked about their views about the technology. Unfortunately, several comments made at the conference and subsequent online commentary reveal serious misconceptions about the cost and benefits of XBRL for municipal disclosure that merit closer examination.
Exaggerated Implementation Costs
In a recent report on XBRL, the firm Municipal Market Analytics (MMA) says that “a transition [to XBRL] is highly unlikely absent a multi-billion-dollar Congressional appropriation to pay for its costs (emphasis added).”
MMA’s cost estimate is exaggerated by at least two orders of magnitude. To understand why this is the case, we need to consider the two major costs to implementing XBRL: (1) developing a reporting taxonomy (the list of standard reporting items and their interrelationships), and (2) building software that can produce XBRL documents conforming to the taxonomy.
The Spanish government paid €200,000 for the development of its local government XBRL taxonomy a decade ago. In Florida, House Bill 1073 (2018) appropriates $500,000 for taxonomy development. A number of draft taxonomies for the US municipal market have already been published, so there are precedents for Florida to leverage.
Further, the XBRL US State and Local Government Disclosure Working Group (chaired by this author) is publishing its first discussion draft taxonomy later this month with more than 100 elements. A second discussion draft reflecting broad stakeholder input will appear later in 2019 – at no cost to taxpayers. It will likely contain more than 200 elements.
A more ambitious taxonomy like the one developed for US corporate filers can cost several million dollars to build. But the SEC US GAAP taxonomy peaked at about 19,000 elements (before being trimmed back in recent years) and addresses all aspects of corporate 10-k filing. Such an extensive taxonomy seems like overkill for the municipal market, which may be adequately served by 1,000-2,000 elements limited to the major financial statements and key footnotes.
An adequate municipal taxonomy should be expected to cost in the low seven figures at most, with some of that money coming from sources other than the federal government. So we have yet to find corroborating support for MMA’s contention of the need for a multi-billion dollar appropriation.
What about software upgrades needed to comply with the taxonomy? Right now, local governments often prepare their Comprehensive Annual Financial Report (CAFRs) using some combination of Microsoft Word and Excel. They then “print” the documents and spreadsheets to PDFs which are then combined into the final CAFR.
Both Word and Excel have macro languages – based on Microsoft’s Visual Basic for Applications. As this author has demonstrated, it is possible to create Word and Excel macros that generate XBRL tagged documents. Software vendors could create commercial quality XBRL generation tools for use within Microsoft Office and sell them profitably for a few hundred dollars. Alternatively, a state or federal agency could provide macro-based Office tools to local government financial statement filers.
Governments seeking more sophisticated CAFR production solutions are licensing subscription products like CAFR Online, CaseWare or Workiva WDesk. These tools store CAFR data in an internal database. It is relatively easy to upgrade such products to produce XBRL-tagged output. Indeed, CaseWare and WDesk are both used by corporate 10-k filers and already have XBRL capabilities built in.
Discounting the Value of Financial Reports
MMA also does not see much point in analyzing state and local government financial statements. At the SEC Municipal Disclosure Conference, MMA’s Matt Fabian said “we don’t care so much about what’s in the financials; it’s more about whether there are financials.” (SEC video at 5:12:30) In their report, MMA cites the city of Harvey, Illinois which defaulted on its General Obligation bonds after failing to post financials thirty months prior.
It is true that failing to post financial statements is a red flag, but recent experience shows that the contents of the financial statements also have predictive value.
In June 2015, financial blogger Michael Shedlock (Mish) asked this author to research Harvey. As I told him at the time, Harvey’s general fund balance of negative $19.3 million was a warning sign of distress. Mish reached out after seeing my research identifying five Cook County municipalities that would be bankruptcy candidates in the event that Illinois gave its local governments access to Chapter 9 of the federal bankruptcy law. One of the communities listed, Dolton Village, subsequently defaulted on its municipal bonds as well. It would have been ideal to research all Illinois local governments rather than a sample, but there are too many and their financial statistics are too costly to gather, which brings us to a third misconception.
PDF Scraping is Not a Substitute for Clean Source Documents
MMA suggests that automatically extracting data from PDFs is a more “state of the art” technology than XBRL. While PDF scraping tools can somewhat reduce the burden of manual data capture, they are not sufficiently reliable to replace human data collectors, and the prospects of this changing any time soon are dim.
Mark Kim, COO of the Municipal Securities Rulemaking Board—the self-regulatory agency that operates the municipal market’s principal disclosure portal—told attendees at the SEC Conference that PDFs created by local governments are “not all of uniform quality and they’re not all equally able to have their data extracted.” (SEC video at 4:07:00).
MMA is critical of what it calls the “decade-old XBRL regime.” And there is some truth in this concern: XBRL is based on XML, a standard developed in the 1990s that is declining in popularity. Earlier this year, however, the SEC began transitioning filers to a newer standard called Inline XBRL, which are HTML web pages with XBRL tags. It is this newer standard that Reason and other advocates recommend applying to municipal disclosure.
Conclusion: Rather than Debate Filing Mandates, the Community Should Agree on Data Standards
MMA worries about a costly XBRL filing mandate being foisted on government financial statement preparers. But, right now, a federal mandate is not on the table. Instead, advocates are asking MSRB to collaborate with the XBRL working group on a taxonomy and then facilitate voluntary filing of Inline XBRL documents that conform to the taxonomy.
This approach has several benefits. First, by collaborating on taxonomy development, we can achieve widespread agreement on which data points in a government’s CAFR are worth standardizing and capturing. This is valuable regardless of whether a technological solution is built with XBRL or some other language. A voluntary filing period would facilitate a process of testing and debugging, which will lead to lower document preparation costs and higher quality. It would also allow regulators to gather market feedback on the usefulness of the reported data. Then and only then would it make sense to discuss a mandatory XBRL filing regime.