EMMA: Time to Grow Up and Be Like Your Big Brother, EDGAR

In 2009, the Municipal Securities Rulemaking Board (MSRB) launched its Electronic Municipal Market Access (EMMA) system: the place to go for all things muni. EMMA contains information about all publicly traded municipal bonds and their issuers including offering documents, trade activity, ratings, issuer financial statements and event notices (such as those required when an issuer misses a payment or calls its bonds).

As a frequent user, I’m impressed not only by the wealth of information available on EMMA, but also with the system’s usability, reliability and ongoing feature improvements. That said, EMMA has very serious limitations that are inconsistent with both open government and a liquid municipal bond market.

In a 2014 open letter to the MSRB, the Sunlight Foundation pointed out that restrictions on downloading and the fact that much of EMMA’s data is still in PDF form greatly limit the system’s transparency. In these respects, it is worth comparing EMMA with the SEC’s system for collecting and presenting company financial filings, which is known as EDGAR (Electronic Data Gathering, Analysis and Retrieval).

Unlike EMMA, EDGAR provides free FTP and RSS access, allowing users to consume as much content as they wish. EMMA only offers bulk downloads as a high cost subscription option and specifically forbids using automated techniques to quickly capture (or “scrape”) site content. It also limits the number of records that can be returned in “Advanced Searches”, hampering the ability of market participants and academic researchers to gather and analyze the big data EMMA contains.

EDGAR further facilitates analysis by providing key company disclosures – most notably quarterly (10-Q) and annual financial statements (10-k) – in machine readable format. Municipal financial statements on EMMA typically appear only in PDF form, requiring laborious parsing or re-keying to obtain usable data.

Recent legislation proposed by Congressman Darrell Issa (R-CA) and co-sponsored by 26 other representatives from both parties would require MSRB to implement machine-readable disclosures on EMMA. The Financial Transparency Act of 2015 (HR 2477) mandates the use of standards based, machine readable disclosures by all financial regulatory agencies and self-regulatory bodies deriving their powers from federal regulators. This includes the MSRB whose power to oversee the municipal securities market is delegated by the Securities and Exchange Commission (SEC).

The SEC also operates EDGAR, which – as we have seen – is far more open than EMMA. But the SEC has not always been an exemplar of open data. It took a combination of outside pressure and bureaucratic innovation to make corporate financial disclosure fully open.

As late as the early-1990s, the primary method of reporting corporate financial results to the public was through printed annual reports and paper regulatory filings. Even after the SEC received company filings electronically, it proved unable to share this machine readable data with the general public.

This situation changed by virtue of work done by Carl Malamud, a northern California open government advocate. Malamud obtained SEC disclosures and began posting them on a web site he built with a National Science Foundation grant. Seeing the success of Malamud’s efforts, the SEC was shamed into providing this service itself. More recently, Malamud, through this work at Public.Resource.Org, has made a similar breakthrough with not-for-profit organization disclosures submitted to the IRS –Form 990. Malamud’s group began putting these forms on line at no charge a few years ago, and recently won a court judgment against the IRS requiring the agency to provide the Form 990 disclosures in machine readable format.

Meanwhile, the SEC has continued to improve EDGAR data. When it began publishing corporate disclosures in the late 1990s, the data appeared in SGML format (SGML is a close relative of HTML). SGML is more easily parsed than PDFs, so the SEC was way ahead of the MSRB and the IRS from the start. But the SGML disclosures were not self-describing: the data files were not tagged in such a way as to provide consistency across files. In the mid-2000s, the SEC began to embrace eXtensible Business Reporting Language (XBRL) which is self-describing. Beginning in 2009, the SEC began to mandate that corporate filers use XBRL – starting with the largest companies and working down to smaller ones. Now EDGAR users can click an “Interactive Data” button next to each disclosure to see the XBRL rendered as an interactive web page.

To this author, it seems odd that private companies and now private, not-for-profit entities have more accessible financial filings than do state and local governments. Many private organizations affect relatively small number of stakeholders – perhaps just a few hundred customers, employees and shareholders. But governments large enough to issue bonds touch the lives of thousands of taxpayers, service users, beneficiaries and other parties: their financial affairs are much more a matter of public interest.

EMMA could serve that public interest if its content were more open – but a number of factors prevent this. For example, MSRB’s board contains members employed by firms in the municipal bond industry whose revenue might be reduced by greater industry transparency.

Some of the content on EMMA is proprietary. This restricted data includes CUSIP numbers that identify each bond, as well as credit ratings. CUSIPs are owned by the American Bankers Association and administered by McGraw Hill Financial; they normally cannot be displayed on a web page without a costly CUSIP license. Although individual bond ratings may be freely reproduced, rating agencies take measures to prevent the bulk redistribution of credit ratings, because they sell ratings feeds to large financial industry customers. Finally, the MSRB also realizes revenue from selling EMMA content in bulk:  users are offered subscriptions to EMMA data feeds that includes various portions of the primary market and continuing disclosures available on the system.  If this material could be bulk downloaded at no charge, MSRB would lose subscription revenue.

While these institutional factors may preclude free bulk access to EMMA content, it is less clear why MSRB has not mandated filings in XBRL or some other open, standardized format – rather than PDFs. This idea appeared on an MSRB road map in 2012, but there does not seem to be momentum toward implementing it.

That situation would change if the Financial Transparency Act of 2015 (HR 2477) becomes law. Once PDFs are replaced by structured data, the cost of creating municipal finance data sets will greatly decline and their availability will greatly increase. The ultimate results should be better value for municipal bond investors and substantial cost savings for cities, counties, school districts and other issuers.

Good Intentions are Not Enough: The Problem of SEC Mandated XBRL Reporting

Public companies have been required to supply financial reports since the Depression, but gathering and analyzing this disclosure has had its challenges. In the 1990s, the SEC began uploading 10-K’s and 10-Q’s to the internet, greatly simplifying the data collection task. These electronic reports were not standardized, creating the need for downstream users to write complex parsing algorithms and/or use manual processes to harvest the financial statements.

In the late 1990s, accounting and technology firms devised a standard called XBRL – eXtensible Business Reporting Language – to streamline the data acquisition process. XBRL disclosures rely on a common system of tags that consistently identify financial statement elements. The universe of elements differ amongst accounting standards, such as US Generally Accepted Accounting Principles (US-GAAP) and International Financial Reporting Standard (IFRS). An XBRL taxonomy lists all the acceptable financial statement elements for a given accounting standards.

Beginning in 2009, the SEC started requiring public companies to file 10-K and 10-Q disclosures in XBRL using a US-GAAP taxonomy – maintained by the accounting community and approved each year by the SEC.

Recently, I worked with UK-based OpenCorporates to gather SEC XBRL disclosures and harvest data from them. The goal was fairly simple: walk through all the XBRL documents and gather some basic parent company data points (like total assets, total liabilities, total revenue and net income) for the latest fiscal year from these disclosures.

This task proved surprisingly difficult because of a lack of standardization between XBRL documents from different companies. For example, many companies did not report a value for Total Liabilities. One might “back into” this value by subtracting Shareholders’ Equity from Total Assets, but this doesn’t always work. A small percentage of XBRL reports even lacked a Total Assets field. On the income statement side, the dispersion was even greater, with Total Revenue, Operating Income and Net Income often unavailable.

Finding data for the latest period also proved challenging. XBRL files can contain numerous contexts. Each context refers to a reporting period (e.g., a particular quarter or year) and a scope – which may be the parent company or a particular segment of the corporation (e.g., a subsidiary). Contexts contain period elements and an optional segment element indicating which timeframe and what scope the context covers. To find the latest year’s parent company data, it is necessary to develop a program to walk through each context.

These examples suggest that processing SEC mandated XBRL disclosures is less than straightforward. Indeed, the industry group XBRL.US reports finding 1.4 million errors in the universe of XBRL documents filed thus far.

A recent letter from Darrel Issa (R-CA) to the SEC notes that the agency itself is not using the XBRL files it requires corporations to file. Instead, it continues to rely on commercial data aggregators. Electronic disclosure won’t improve unless numerous eyes are scrutinizing it and reporting issues. Data sets need to be exercised; otherwise they remain unfit.

The lack of XBRL utilization represents a major threat for transparency advocates. If we ask for more accessible disclosures and then don’t use them, filers can be expected to push back. In the case of SEC XBRL, the filings are sufficiently complex to require the use of third party XBRL submission firms. In other words, it is too difficult for most companies to prepare XBRL submissions themselves – they need to use an independent preparer, just as individuals often need to hire professionals to file their annual tax returns. Corporations would undoubtedly like to economize on this cost, and can be expected to resist the XBRL reporting requirement if the filings are not used.

From my perspective, a big problem with the XBRL rollout is that it started with large public companies. By 2009, many data aggregators already had mature processes for assimilating the traditional SEC disclosure. As a result, fielded public company financial data has become a commodity; individuals can access these data for free at Yahoo Finance and many other portals. The incentive for aggregators to use XBRL is thus limited because the problem has already been solved at some level, and because the data are widely available, there is little benefit to potential new entrants.

XBRL can provide much greater benefits for data sets that have not received as much attention. I became interested in XBRL back in 2001 because I was hoping to get a standard source of private company data at my bank. The idea was to provide unlisted corporate borrowers with an XBRL template to provide quarterly disclosure.

Another high impact application for XBRL is state and local government financial reporting. When XBRL was growing up in the 1990s and 2000s, US municipal bonds were generally perceived to be safe. That perception started to change in 2008 with Vallejo’s bankruptcy filing and the collapse of the municipal bond insurance industry. Subsequent municipal bankruptcies culminating with that of Detroit in 2013, have reinforced the perception that municipal securities are risky. Government financial statements, which may have been ignored previously, now have significance as investors search for the next bankruptcy candidates.

However, the rollout of XBRL to other areas – such as local government – may now depend on its successful implementation in existing areas: especially the high profile SEC US public company application. If the SEC is unwilling or unable to engage, the community would be well served by collaborating to implement its own improvements. Although XBRL filing companies compete with one another, they all have an interest in the success of the XBRL standard. Thus, as an industry group, these companies can propose and implement improvements to SEC XBRL filings that will make them easier to use. For example, they can develop an enhanced XML Schema Definition which provides additional checks over and above those legally mandated. Such a schema should ensure that filers always include common financial statements such as Total Assets and Total Liabilities. It should also ensure that, within any given XBRL file, the latest period’s parent company filing is easily identified.

XBRL was and remains a good idea. Transparency advocates need to ensure that it does not become an idea whose time has come and gone. To keep XBRL on track, its public company instance needs to be refined so that implementation costs are reduced. Further, it needs to be applied to other areas – such as US local governments – which stand to gain greater benefits from its adoption.