The Same Overhead, Twice the Donations

People do not hate overhead. They hate paying for it themselves. When another donor covered the identical cost, giving doubled: 855 donors against 475.

Ask people why they distrust a charity and most of them will land on the same word within a minute. Overhead. The rent, the salaries, the accountant, the mailing that asked you for money in the first place. Nobody wants their twenty dollars going into a photocopier.

That instinct feels like efficiency. It turns out to be something else entirely, and a study published in Science in October 2014 caught it doing something a purely efficiency-minded instinct could never do.

Uri Gneezy, Elizabeth Keenan and Ayelet Gneezy started in a lab with 449 undergraduates in southern California. Participants chose between two real charities, and the researchers varied how much of a donation would be eaten by administrative costs. As overhead rose, fewer people picked the charity. Nothing surprising there.

Then they changed one thing. The overhead stayed exactly the same size — someone else just paid for it. A separate donor absorbed the administrative cost, and the participant's own money went straight to the work.

The aversion vanished. Same charity, same total overhead, same money leaving the same building. People behaved as though there were no overhead at all.

That is not efficiency. If overhead were a signal of a badly run organisation, it would still be a bad signal when a third party covered it. What people actually want is for their contribution to land on the cause. Whose accountant gets paid is somebody else's problem, as long as it isn't paid out of the twenty dollars they just wrote a cheque for.

Forty thousand letters

The lab result is neat but it is undergraduates in a room. So the researchers went and tested it on real donors with real money.

In the spring of 2013 a foundation mailed 40,000 solicitation letters on a single day for an education project costing $20,000 per site. Ten thousand letters each went out in four versions, and recipients could tick a box for $20, $50 or $100.

The control letter said only what the project was and what it cost. The second told recipients that a private donor had already put in $10,000 as seed money — a technique fundraisers have used and studied for decades, and one that reliably works. The third offered a $1:$1 match up to $10,000, which also reliably works. The fourth said that a private donor's $10,000 was covering the campaign's administrative and fundraising costs, so every dollar sent in would go directly to the project.

Same foundation. Same project. Same $10,000 sitting behind the appeal in all three of the non-control letters. The only difference was what the letter said that money was doing.

Three hundred and thirty-six people donated to the control letter. Four hundred and seventy-five to the seed letter, four hundred and forty-one to the match. Eight hundred and fifty-five to the overhead-free one.

Eighty percent more donors than seed money produced. Ninety-four percent more than matching. In dollars, $23,120 against $13,220 and $12,210 — and $8,040 for the plain letter.

Look at the largest gifts and it gets stranger. In the control arm, out of ten thousand letters, exactly three people gave $100. In the seed and match arms, twenty-seven each. In the overhead-free arm, forty-three.

Why this is worth taking seriously

The finding has an obvious operational reading, which is that if you have a lead gift you should spend it on the boring things and say so. That reading is correct and slightly beside the point.

The more interesting reading is about what a donation is for. If people were buying units of impact, seed money and matching should beat overhead-free by a wide margin — a matching grant literally doubles the impact of every dollar, and it still lost by ninety-four percent on participation. Instead people paid for a clean line between their own money and the thing they cared about.

I find this uncomfortable and true at the same time. It means a charity can be punished for honest bookkeeping while a competitor with a wealthy patron quietly paying the rent looks like the efficient one. It also means the fix is cheap, and it doesn't require lying to anybody.

There is a caveat worth stating plainly. This is one field experiment with one foundation and one cause, run in the United States thirteen years ago, and the mailing was to people already on a donor list. A single 40,000-person trial is strong evidence that the effect is real and large in that setting, not proof that the same multiple shows up everywhere.

But the direction of it fits everything else we know about how giving actually works. People are not optimising a spreadsheet. They are trying to feel that something they did reached someone. Take away the sense that their money got skimmed before it arrived, and twice as many of them will do it.

Source: Uri Gneezy, Elizabeth A. Keenan and Ayelet Gneezy, "Avoiding overhead aversion in charity", Science, vol. 346, issue 6209, 31 October 2014, pp. 632–635 (DOI 10.1126/science.1253932). The laboratory study with 449 participants, the field experiment with 40,000 letters in four arms of 10,000, the per-arm donor counts in the paper's Table 1, and the headline increases of 80% and 94% in donation rate and 75% and 89% in total donations all come from that paper. The dollar totals per arm are calculated from Table 1. Read 30 August 2026.