Transparency Activist Wants Machine-Readable Form 990s

Print Share on LinkedIn More



June 18, 2013; Washington Post, “WonkBlog”

As NPQ readers know all too well, 990s are public information and can be accessed through websites like ProPublica and Guidestar. (Guidestar displays a few years for free, and an organization’s entire history of 990s for a fee.) Both sites also provide a summary of some of the pertinent data from the 990s; ProPublica opens it all to the…well, public, while Guidestar saves them for premium subscribers, though Guidestar’s summaries contain more data and more fields.

But beyond that, digging into a nonprofit’s 990 requires scanning the data in a PDF version and hoping to find what you’re looking for, like that nugget on payments to a vendor or an investment in this or that corporate stock. It’s possible to do some in-depth data dives on campaign finance and lobbying at, our favorite source, but on nonprofits, both 501(c)(4)s and (c)(3)s…not so much, to put it mildly.

Earlier this year, Carl Malamud, who runs Public.Resource.Org, filed a Freedom of Information Act request for all 990s in machine-readable format. The IRS was less than enthusiastic about Malamud’s FOIA request, saying that the information was already available in PDF format, so nothing else needed to be done.

Based on getting stiffed by the IRS, Malamud this week filed a lawsuit in the U.S. District Court in Northern California to challenge the IRS denial. We find it’s always interesting to read the court documents for insights into what is going on. A few items from Malamud’s complaint are reproduced below:

  • “There is no dispute that the IRS receives Form 990 tax records electronically from NPOs and maintains them in the same machine-readable manner in which they are originally filed by the NPOs with the IRS as MeF [Modernized e-File] files…There is no technical or similar reason that the IRS cannot produce…in MeF format copies of the Form 990s for the NPOs specified. There is also no dispute that the non-MeF versions of Form 990s that the IRS has offered…as publicly available records, which the IRS claims remove all of the MeF versions from FOIA, are far less useful than the MeF versions, and accordingly, that the Form 990s in their disparate formats are not equivalents.”
  • “In 2007, Public.Resource.Org began a program to make Form 990 filings of exempt organizations more accessible to the public. Since then, Public.Resource.Org has obtained and processed 98.7 pounds worth of DVDs from the IRS. The IRS DVDs contain Form 990s in TIFF image file format, regardless whether or not the organization filed its 990 in electronic form or paper form…[T]his is the current method that the IRS uses to make Form 990 information available in bulk…Because Form 990s are received in TIFF format, it means the Form 990 files are in an image format that is the same as a low-resolution photograph….As a result of the conversion to TIFF, it is no longer possible—as it would be with the native MeF data—to have a computer scan the text and extract information systematically. Instead, human interaction is required to gather data from each return.”

The complaint recounts meetings with IRS personnel who confirmed to Malamud that there were no technical reasons why the 990s could not be released in machine-readable format; “only policy considerations were preventing release.” What policies, and what consideration? The complaint recounts a meeting Malamud had with Lois Lerner, the now-former head of the IRS tax-exempt unit, who explained that nonprofits with less than $10 million in assets had the option of e-filing, but were not required to do so. According to Malamud’s complaint, Lerner explained that, “if these nonprofits knew their e-file data were to be released in a ‘more transparent’ fashion, they would elect to file paper copies instead to avoid the problem of ‘too much transparency.’”

So, in Lerner’s analysis, although the 990s are public information, making them actually usable and machine-readable could result in “too much transparency.”

In its denial of the FOIA request, the IRS contended that, “Form 990 data do not constitute a recognizable record, but rather a continuous string of numbers that includes confidential return information. Our existing process for providing releasable copies of Form 990 is to convert the MeF data into a PDF format and withhold confidential return information from the resulting form.” Whether or not one agrees that the MeF-formatted 990s contain confidential information (Malamud says that the Schedule B information is removed from the MeF 990s), the IRS’s contention that the scrubbed returns couldn’t be converted back to MeF or some other machine-readable format is hard to believe.

Our thoughts? You go, Carl Malamud! We have heard from plenty of experts that the technology is there, and so the task could be undertaken. If Malamud wins, the cause of transparency will be significantly advanced. Sign us up!—Rick Cohen

  • Michael Wyland

    Interested parties may download data files on nonprofits from the IRS web site at:

    The data files are available in MS Excel format for each state’s nonprofits. In addition, the data is available in fixed-width ASCII format by state or by region.

    These data files are based on the Publicaiton 78 listing of all nonprofits recognized by the IRS, but they also include the most recent total asset, total revenue, and total expense figures from the latest Form 990 filed by each charity. The IRS updates the files avaiulable for download each month.

    I, too, hope Malamud wins, since it will make more information available in an easier-to-analyze format. Maybe the IRS could even begin with a more modest innovation – packaging the currently available data into a single file for download in MS Excel, or if that’s too large a file, make it available in comma-separated ASCII (CSV) format. That way, PC-based programs like Excel can read the large files without expensive conversion routines.

  • The Data Guy

    For real transparency, Schedule B is indispensable information. So any electronic release without that is not really useful. Also, you need a list of the grantees. Unfortunately the IRS with its dinosaur IT practices using flat record files with fixed length fields has a hard time to conceptualize a more useful data release.