The Overhead Myth, again: why "fund programs, not people" is backwards

How the overhead myth worked

For most of the 2000s and 2010s, the fastest way to look like a good charity was to report that almost none of your money went to administration or fundraising. Charity Navigator's star ratings leaned heavily on overhead ratios; donors learned to hunt for the magic "90 cents of every dollar goes to programs" line. The result was a sector-wide arms race in creative accounting and, worse, genuine under-investment: chronically underpaid program staff, no training budgets, duct-taped databases, no evaluation.

The watchdogs eventually noticed what they had built. In June 2013, the CEOs of GuideStar, Charity Navigator and the BBB Wise Giving Alliance published an open letter, "The Overhead Myth", asking donors to "pay attention to what matters: the true measure of impact." They wrote that overhead ratios "do not say anything about the true impact of the organization" and that judging charities on them "can be devastating to the people and communities we are all trying to help." It is hard to overstate how remarkable that was: the three organizations that built the ratings industry publicly admitted the ratings had been rewarding the wrong thing.

The damage was already done, though. A generation of donors still equates low salaries and cheap offices with virtue, and a generation of nonprofit executives still apologizes for the cost of doing the work well.

The same myth, new costume

Listen to how new money talks about giving today — AI-lab philanthropy, crypto wealth, effective-altruist funds — and you hear the overhead myth again, translated: "We fund research, not individuals." The reasoning is familiar. Research is a program; a person is overhead. Institutions convert dollars into knowledge; individuals just consume dollars. It sounds rigorous. It is the same accounting error with a lab coat on.

The error is treating the delivery mechanism as the outcome. Overhead ratios measured the cost of the machinery, not the results. "Fund research, not people" measures the prestige of the vehicle, not the results. Both substitute a proxy you can easily count — a ratio, a citation, an institution — for the thing you actually care about: did someone's life get better, and by how much per dollar?

What the cash-transfer evidence says

This is not a hypothetical. Direct cash transfers are among the best-studied interventions in modern development economics, and the studies keep finding the same thing: when you give poor people money, they use it well.

• Cost efficiency: GiveDirectly, the best-known direct-giving charity, reports that roughly 85–90% of each donated dollar reaches recipients directly. That is a better "program ratio" than most traditional nonprofits achieve after their program-side costs.
• Results: randomized evaluations of unconditional cash transfers find sustained increases in consumption, business assets and food security, and measurable improvements in mental health — with no meaningful increase in spending on alcohol or tobacco, the outcome skeptics always predicted.
• Comparison: when researchers have compared cash to in-kind programs, cash usually holds its own or wins, because the people receiving it know their own needs better than a distant program designer does.

In other words, the "overhead" in a cash transfer is almost entirely the cost of identifying recipients and moving money. Everything else is the program. The recipient's judgment is the program.

Paying people IS the program

Here is the reframe the 2013 letter started and the new funders haven't finished: there is no clean line between "people" and "programs." A program is people. A researcher is a person whose salary you pay. A case manager is a person whose salary you pay. The only honest question is which people, paid how, produce how much good per dollar.

Sometimes the answer is a lab. Basic research has enormous long-run returns, and some of it should be funded at scale. But sometimes the answer is a poor household, and the evidence for that is at least as strong as the evidence for most institutional funding — and the cost ratio is often better. Refusing to fund individuals on principle, while funding institutions by default, isn't rigor. It's the overhead myth with the sign flipped.

The sector already learned this lesson once, expensively. Donors spent two decades forcing nonprofits to pretend that their staff, their systems and their measurement were waste. The watchdogs recanted. Let's not make the new money learn it the same slow way: judge giving by outcomes per dollar, and let individuals be a legitimate line item — not as overhead, but as the program itself.

FAQ