Nonprofit tax returns and registration forms are the public’s (and the government’s) primary window into the workings of America’s enormous and economically impactful nonprofit sector, which all together pays $670 billion annually in wages and benefits. Every year in the United States, approximately 1.5 million registered tax-exempt organizations file a version of the federal “Form 990” with the IRS and state tax authorities. These forms — whose questions vary a bit depending on the type of organization — collect details on the financial, governance, and organizational structure of America’s universities, hospitals, foundations, and charities, to the end of ensuring that they are deserving of their tax exempt status. All but ten states also require that nonprofits operating in their states file state-specific registration forms. The information these filings contain about executive compensation, fundraising expenses, and donation activities, for example, can help regulators spot potential bad actors and alert each other to targets for further investigation.
Yet despite the richness and utility of the information contained in these filings, major barriers prevent state regulators from efficiently sharing and analyzing the data. Although every Form 990 is required by law to be open and available for public inspection, the IRS makes it a practice to print out the returns and scan them back in so that the resulting file is an image file, rather than in machine-readable electronic format (MeF). The IRS has followed this antiquated practice even for the large percentage of returns already filed digitally. State regulators, moreover, can in theory share and compare their registration data. But practically, the widespread use of hard-copy paper filings — along with differences in the formats used by states with electronic filing systems — makes pooling registration data extremely challenging. Where they exist, paper and analog systems also impose serious, needless delays on investigations.
What results from these barriers is a regulatory system unable to harness the full power of computable open data. In the current cancer charities case, good old-fashioned sleuthing and teamwork allowed state regulators to make do with the data they have. Between 2008 and 2012, it is asserted that the Cancer Fund group of nonprofits raised more than $187 million from donors throughout the U.S., but spent a pittance on actual aid to recipients, instead channeling the monies to themselves. But what if fully open, machine-readable 990 data had revealed suspicious patterns earlier? How many bad actors could be stopped if state regulators could more easily pool and analyze their registration data?