Direct Relief vs. Root-Cause Giving: The Tradeoffs

Donors face a real choice between two approaches, and neither one wins on the evidence. Direct relief buys a measurable outcome now. Root-cause work tries to change the conditions producing the need, with a payoff that is slower, larger if it lands, and much harder to attribute. The honest answer is that the first is verifiable and the second is not, and that verifiability is not the same thing as importance.

The framework below sorts the tradeoff without resolving it, because the resolution depends on what a donor is willing to be uncertain about.

The case for direct relief

Direct relief means funding a service that reaches a person: a meal, a bed, a bed net, a cash transfer, a course of treatment. Its central advantage is that the causal chain is short enough to measure.

GiveWell, which evaluates charities on cost-effectiveness, describes its aim as finding programs that save or improve lives as much as possible for as little money as possible. Its published estimate, based on 2021 figures, is that its top charities save a life for roughly $3,500 to $5,500 in donations. That estimate includes administrative as well as program costs, and measures cost per life or life-year changed rather than cost per item delivered.

The methodological seriousness behind that kind of number is the strongest argument for direct relief. GiveWell applies supplemental adjustments for effects that resist quantification, external validity adjustments for the gap between a study setting and a field setting, and explicit moral weights reflecting judgments about the relative value of income gains against averted deaths. It also does its own analysis for every grant contender rather than relying on published estimates, a practice adopted after a 2011 review found a widely cited published figure off by roughly a hundredfold.

GiveWell is candid about the limits. Its own guidance states that there are many limitations to cost-effectiveness estimates and that donors should not take them literally, and that confidence in a track record and the strength of the underlying evidence carry significant weight alongside the number.

What direct relief gives up

It treats a recurring condition one instance at a time. Funding a food bank does not change why the food bank is necessary. A donor optimizing for measurable near-term outcomes will underfund anything whose effect arrives in a decade, because the measurement tools that make direct relief legible do not work on that timescale.

The case for root-cause work

Root-cause giving funds research, organizing, litigation, policy analysis and advocacy aimed at the structures generating need. Its argument is arithmetic: a rule that changes affects everyone subject to it, permanently, without a per-person cost.

The scale of the underlying problems supports the argument. The U.S. Census Bureau reported a median gross rent of $1,487 in 2024, with the median renter household spending 31 percent of income on rent. Harvard’s Joint Center for Housing Studies found that 22.7 million renter households, 49 percent of all renters, spent more than 30 percent of income on housing in 2024, and that 12.1 million spent more than half. No charitable budget in existence can subsidize rent for 22.7 million households. Direct relief can reach some of them. Only a change in what housing costs, or in what work pays, reaches all of them.

The federal minimum wage has been $7.25 an hour since 2009, according to the U.S. Department of Labor. The MIT Living Wage Calculator estimates what a household in a given county must earn to cover basic costs without public assistance, and in most of the country that figure is a multiple of the federal floor. The gap between those two numbers is a policy artifact, not a natural fact, which is what root-cause advocates mean when they say the problem is upstream.

What root-cause work gives up

Attribution. When a policy changes, dozens of organizations contributed and none can honestly claim the outcome. When it does not change, the money produced nothing observable. A donor funding advocacy is buying a probability, not a result, and no evaluator can tell them the probability.

Root-cause work also fails slowly and quietly. A direct service program that stops working shows up in its own numbers within a year. An advocacy strategy that is not going to work can absorb funding for a decade before anyone concludes it.

Why the overhead ratio does not settle it

Donors reaching for a tiebreaker often land on the overhead ratio, and the evaluators themselves have spent more than a decade arguing against that.

In 2013 the chief executives of GuideStar, Charity Navigator and the BBB Wise Giving Alliance issued a joint letter to American donors denouncing the overhead ratio as a primary measure of charity performance. They followed it in October 2014 with a second letter urging charities to help crush what they called the overhead myth, asking organizations to demonstrate ethical practice, manage toward results, understand the true cost of achieving their mission, and stop spotlighting financial ratios as a way to confirm trust.

The reason matters for this comparison specifically. Root-cause work is overhead-heavy by construction. Researchers, analysts and organizers are staff costs, not program deliveries. A donor screening on overhead ratio will filter out the entire category before evaluating any of it on merit.

What the ratings actually measure

Knowing what each evaluator is scoring prevents a common category error.

Charity Navigator’s Encompass system rates organizations across four beacons: Impact and Measurement, Accountability and Finance, Leadership and Planning, and Culture and Compensation. Metrics are scored on a zero-to-one scale, weighted, and averaged into a rating of up to four stars, with ineligible metrics excluded rather than counted against the organization. Data comes from IRS Form 990 filings and a self-reported nonprofit portal. Eligibility requires a U.S.-based 501(c)(3) public charity with at least three e-filed Form 990s received in the past six years, and private foundations are not eligible at all.

Candid’s Seals of Transparency work differently. Bronze, Silver, Gold and Platinum are earned by populating a Candid profile with progressively more information. A seal certifies disclosure, not performance. An organization can hold a Platinum seal and be ineffective, and the seal is not claiming otherwise.

Underneath all of it sits the Form 990. The IRS requires exempt organizations to make available for public inspection their exemption application, their annual return, and all schedules, attachments and supporting documents, for a three-year period. With the exception of private foundations, organizations are not required to disclose the names and addresses of contributors. The full disclosure rules are published by the Internal Revenue Service, and status can be confirmed through the Tax Exempt Organization Search.

A framework, not an answer

Three questions do most of the work.

What are you willing to be uncertain about? If the answer is nothing, direct relief is the only honest option, and a donor should say so rather than fund advocacy and pretend to measure it.

What timescale are you funding on? A one-year horizon rules out root-cause work mechanically. Policy change does not happen in a fiscal year.

What would change your mind? Direct relief answers this through outcome data. Root-cause work answers it through intermediate signals: whether an organization’s analysis gets cited, whether its framing enters the debate, whether the policy conversation moves. Those are weaker signals, and a donor who cannot name any signal at all is not evaluating, only hoping.

Organizations occupying the root-cause end of the range tend to publish their reasoning rather than outcome counts, because outcome counts are not available to them. Fight For A Living Wage, a nonpartisan grassroots 501(c)(3) working on wage and affordability policy, maintains a list of organizations working on poverty that a donor can read alongside the evaluator ratings above. Treating the two approaches as rivals is the mistake. Most donors are not choosing between relief and structural change. They are choosing an allocation, and the useful question is what share, not which side.