The Nonprofit Paradox

When tax exemption becomes privilege.

A mutual aid organization in a Rust Belt city runs on three paid staff and perpetual grant anxiety. When funding dries up, programs get cut. Its Form 990 is a filing obligation nobody reads. Forty miles away, a major metropolitan symphony orchestra sits on tens of millions in endowment assets and hosts galas attended by the city’s donor class. It is also a 501(c)(3). It also files a Form 990. Under the tax code, these two organizations are legally identical.

The American nonprofit sector reported $2.62 trillion in revenue and $5.99 trillion in assets as of 2016, growing faster than GDP over the prior decade. It is the third-largest workforce in the country. The regulatory architecture governing it was designed for a world in which none of those numbers were imaginable.

The 501(c)(3) bargain is straightforward: exemption from federal income tax, donor deductibility, property tax exemptions at the state and local level, and in exchange, a mission serving the public good. The enforcement mechanism for that exchange is, in practice, self-reporting. Approximately 35 percent of nonprofits registered with the IRS in 2016 were required to file a Form 990 or its variants. Nearly two-thirds filed nothing comprehensive at all. The form is transparency theater for the minority of organizations large enough to trigger it, and the IRS — structurally underfunded and politically constrained — has by and large done little to enforce the limits the code nominally imposes.

The system is one of disclosed self-governance. Oversight arrives, if at all, after something breaks visibly enough to attract an attorney general or a journalist.

In a for-profit corporation, shareholders can fire boards, boards can fire CEOs, and markets can punish underperformance. In a nonprofit, there are no owners, no shareholders, and no market signal. Mission is the mandate. But nobody has built a reliable instrument for measuring whether the mission is being met, and the people responsible for that measurement — the boards — are frequently constituted by people in the executive’s professional orbit.

Research published in Compensation & Benefits Review in 2025 confirmed what practitioners already knew: board governance activities significantly shape nonprofit CEO compensation outcomes, and interlocking directorates — board members who sit on multiple nonprofit boards simultaneously — create peer networks that normalize upward pay benchmarking. The result is a compensation ratchet with no countervailing force. Large nonprofit CEOs are now routinely paid at levels that would have been unthinkable two decades ago, justified by the argument that organizations must compete for talent with the for-profit sector. That logic is self-reinforcing: each organization benchmarks against the others’ highest compensation, and the ceiling rises with every cycle.

At the other end — the mutual aid org, the neighborhood clinic — extreme undercompensation is the norm. The sector’s pay disparity mirrors and reinforces the class inequities nonprofits often claim to address.

The Nonprofit Starvation Cycle, Stanford Social Innovation Review’s landmark framing, describes how donor pressure to minimize overhead starves organizations of the administrative capacity they need to function. A 2004 Urban Institute study characterized administrative costs as infrastructure and found that organizations spending too little on it had more limited effectiveness. The finding should have changed how donors evaluate nonprofits. It mostly didn’t.

At large, sophisticated nonprofits, the opposite pressure has taken hold. Fundraising has become so professionalized, so technologically intensive, and so donor-acquisition-driven that it consumes the mission rather than serving it. Charity Navigator, GuideStar, and the BBB Wise Giving Alliance spent decades promoting overhead ratios as performance metrics before collectively recanting, calling the focus counterproductive, and arguing that donors should evaluate transparency, governance, and results instead. The recantation was welcome. Charity Navigator’s ratings still shape donor behavior in ways that reward organizations skilled at reporting low overhead rather than achieving low overhead. No broadly adopted, standardized measure of actual impact exists in the nonprofit sector. Donors navigate by instruments that measure the cockpit’s appearance, not the flight path.

A major symphony, art museum, or private university receives full federal income tax exemption, property tax exemption often worth tens of millions annually in urban real estate markets, donor deductibility that flows disproportionately to high-income taxpayers since only itemizers benefit and the value of the deduction scales with marginal tax rate, access to tax-exempt bond financing, and public grants from the NEA and state arts councils. A $200 opera ticket subsidized by taxpayers who cannot afford it is not a public good in any democratically defensible sense of the term.

The National Committee for Responsive Philanthropy has consistently pressed the question: who benefits from philanthropy, and whether it is transparent and accountable to those with the least wealth, power, and opportunity. Their finding — that traditional philanthropy has been falling short of addressing critical public needs not because of fraud but because the system’s incentives direct resources toward prestige, stability, and donor preference rather than toward the communities most in need — is sharpest here. Unlimited tax privilege for institutions whose primary beneficiaries are wealthy patrons constitutes an upward wealth transfer dressed in the language of public good. That transfer should at minimum be named.

Brookings research on nonprofit-government entanglement identifies a parallel distortion: large nonprofits that have become service-delivery arms of the state, funded primarily through government contracts and grants. The arrangement creates a specific perverse incentive. Mission is subordinated to contract compliance, and advocacy capacity — the thing that would hold government accountable for the conditions the nonprofit exists to address — is quietly surrendered in exchange for continued funding. The organizations best positioned to critique the system are the ones the system has learned to feed.

Where the 501(c)(3) ends and political power begins is a line the tax code drew, the courts blurred, and the IRS stopped policing. OpenSecrets has tracked more than $2.8 billion in dark money spending reported to the FEC since Citizens United in 2010. The 2024 cycle was the worst yet: dark money groups, nonprofits, and shell companies moved more than $1.9 billion into the election, a near-doubling of the prior record set in 2020.

Politically active 501(c)(4) nonprofits keep reported political spending carefully below the de facto limit of 49 percent of total expenditures, describing tangential political activities as educational or membership-building. A 501(c)(4) that sponsors ads in federal races faces no FEC reporting obligation if the ads air outside narrow electioneering-communication windows. The windows are narrow. The calendar is long. The result is a system in which political actors enjoy the tax advantages of nonprofit status while facing none of the disclosure requirements attached to declared political organizations. This is a bipartisan architecture: outside spending on 2024 federal elections hit a record $4.5 billion, with more than half coming from groups that do not fully disclose the source of their funding. The problem is the design, not the ideology.

The reforms that would matter are not mysterious. Stronger board independence standards — mandatory independent-director supermajorities above a revenue threshold. Compensation ratios — CEO pay as a multiple of median organizational pay, not a ceiling but a disclosed metric with a public face. Mandatory impact reporting — standardized, auditable measures of mission performance, not just financial compliance. Greater IRS enforcement capacity, which requires Congress to fund the agency, a politically fraught proposition that reveals exactly whose interests current underfunding serves.

The structural rethinking that would matter more: tiered tax exemption, so that organizations whose beneficiaries are predominantly wealthy receive a different subsidy level than those serving low-income populations. Re-evaluation of cultural institution exemptions — a museum sitting on a billion-dollar endowment and charging $300 galas should not receive the same structural treatment as a food bank. Foundation payout reform, because foundation giving has declined as a percentage of assets despite fifteen years of booming asset growth, and the 5 percent minimum payout requirement is rarely exceeded. Dark money disclosure — requiring 501(c)(4)s to identify donors when political spending crosses a defined threshold.

Some of these reforms require Congressional action in a legislative environment where the largest nonprofits are significant political donors and lobbying forces. The sector has captured, at least partially, the oversight apparatus that would constrain it.

The nonprofit sector is not a conspiracy. It contains hundreds of thousands of organizations doing genuine, underfunded, undervalued work. But it has also become a parallel economy — $2.62 trillion in reported revenue, sheltered from taxation, largely self-governed, answerable primarily to donors rather than to the public whose subsidy makes it possible. Within that economy, a prestige tier has emerged: elite universities, hospitals, cultural institutions, and foundations that are, in any functional sense, major economic actors. They hold real estate, employ thousands, shape policy, and influence elections — all while retaining the regulatory treatment designed for a church bake sale.

The gap between the sector’s economic weight and its accountability infrastructure is a policy choice, made repeatedly by legislatures that benefit from the current arrangement.

The public subsidy is real. The question is whom it serves. And the answer, for the prestige tier, has not been the public for a long time.