The portfolio approach to giving
Rees Calder · 3 May 2026 · 6 min read
Most people who give to charity give to one. Maybe two. They pick their cause, they back it, they stick. The giving equivalent of putting everything into a single stock.
In investing, this would be considered reckless. In giving, it's considered loyal.
The effective giving community has been quietly developing a different model: the portfolio approach. Spread your giving across cause areas, risk profiles, and evidence levels. Not because you're indecisive, but because uncertainty is real and diversification is how rational agents handle it.
The case for concentration
The standard effective altruism argument runs the other way. If you've identified the single most effective charity, why would you dilute your impact by splitting funds? Every pound diverted from the top pick to a lower-ranked option reduces total expected impact.
This is logically sound and practically misleading, for three reasons.
You don't know which charity is best. GiveWell's cost-effectiveness estimates come with confidence intervals that span 4-10x. Their top charity might save a life for 3,000 dollars or 12,000 dollars. When the uncertainty is that wide, the "best" charity is not a point estimate. It's a probability distribution.
Cause areas have different risk profiles. Global health interventions (bed nets, deworming) have strong evidence but face diminishing marginal returns as coverage increases. Speculative cause areas (AI safety research, alternative proteins) have weaker evidence but potentially enormous payoffs. A portfolio captures both the reliable and the asymmetric.
Your values are not a single variable. You might care about human welfare, animal suffering, and existential risk. Forcing yourself to pick one is not rationality. It's arbitrary constraint.
How to build a giving portfolio
The framework borrows from investment theory but adapts it for charitable giving's unique features: no financial return, high uncertainty, moral rather than monetary weights.
Step 1: Allocate across risk tiers.
Divide your annual giving budget into three tiers.
The core tier (50-70% of budget) goes to organisations with strong evidence of impact. GiveWell's top charities, Evidence Action's programmes, the Against Malaria Foundation. These are your index funds: reliable, well-evidenced, low variance. You know roughly what a dollar does here.
The growth tier (20-35%) goes to organisations with plausible theories of change but less rigorous evidence. Corporate animal welfare campaigns, policy advocacy organisations, climate adaptation programmes. The evidence is emerging, the potential is high, the uncertainty is real. These are your growth stocks.
The frontier tier (5-15%) goes to speculative, high-expected-value bets. Research organisations, early-stage nonprofits, cause areas that might matter enormously but lack the evidence base to prove it yet. Existential risk reduction, wild animal suffering research, governance reform in neglected countries. These are your venture bets. Most won't pay off. The ones that do might matter more than everything else combined.
Step 2: Allocate across cause areas.
Within each tier, spread across at least two cause areas. If your core tier is 60% of budget, you might split it 40% global health and 20% animal welfare. If your growth tier is 25%, split it between climate adaptation and policy advocacy. The exact split reflects your values, not an optimisation formula.
The point is not to find the mathematically optimal allocation. The point is to ensure that if you're wrong about one cause area (and you probably are wrong about something), your entire giving portfolio doesn't go to zero impact.
Step 3: Rebalance annually.
Once a year, review three things. Has the evidence changed for any of your picks? Has a frontier-tier bet graduated to growth tier (or been abandoned)? Have your values or financial situation shifted?
GiveWell updates their models quarterly. Open Philanthropy publishes annual cause area reviews. Animal Charity Evaluators releases new recommendations each November. Use these as your rebalancing triggers.
What the research says
The portfolio approach is not just theoretical. Several frameworks have been developed and tested.
Giving What We Can's cause area allocation. GWWC's 2024 research suggests donors can capture roughly 80-90% of the value of "giving perfectly" by spreading across their top 3-5 charity recommendations, even if the exact ordering is wrong. The cost of diversification is small. The insurance against being wrong is large.
Open Philanthropy's worldview diversification. Open Phil explicitly allocates across "worldviews": a neartermist bucket (global health, animal welfare), a longtermist bucket (AI safety, biosecurity), and a reform bucket (criminal justice, policy). They weight these based on credence levels: how much they believe each worldview is correct. This is the most sophisticated version of the portfolio approach in practice.
Rethink Priorities' cause area prioritisation. RP's 2024 analysis provides quantitative comparisons across cause areas using a consistent framework. Their conclusion: reasonable people with different moral weights end up with very different optimal portfolios, but almost no reasonable set of weights produces a portfolio concentrated in a single cause area.
A starter portfolio
If you're giving 5% of income and want a simple starting allocation:
Core (60%): Split between AMF or Malaria Consortium (global health, 40%) and The Humane League (animal welfare, 20%).
Growth (30%): Split between GiveDirectly (cash transfers, 15%) and CATF or Carbon180 (climate, 15%).
Frontier (10%): Pick one research organisation in a cause area you find intellectually compelling. MIRI, Rethink Priorities, the Legal Priorities Project, the Centre for the Governance of AI. This is your bet on the future.
Total: five to six organisations. Review once a year. Adjust the percentages as evidence updates. Done.
The objection and the answer
"This is overcomplicating it. Just give to AMF."
Fair. If the alternative to a portfolio is analysis paralysis and giving nothing, then yes, just give to AMF. The best giving strategy is the one you actually execute.
But if you're already giving, and you're thinking about where to give, the portfolio approach is strictly better than concentration. It handles your uncertainty honestly. It covers your moral bases. And it means that when next year's evidence update reshuffles the rankings, you're not starting from scratch.
Diversification is not indecision. It's humility about what you don't know, structured into a plan you can actually follow.
One action
Write down your current giving. Calculate the percentage in each cause area. If 100% is in one bucket, move 20% to a different cause area this month. You've just built your first portfolio.
Sources used: GiveWell cost-effectiveness analysis uncertainty ranges (2024), Giving What We Can cause area allocation research (2024), Open Philanthropy worldview diversification framework (2024), Rethink Priorities cause area prioritisation analysis (2024), Animal Charity Evaluators top charity recommendations (2024), modern portfolio theory applied to philanthropy (Mogstad & Wiswall, "Giving Portfolios", working paper 2023). Full links in the planning doc.