Why doesn’t the Modigliani-Miller theorem apply to the relationship between endowments (“firms”) and donors (“shareholders”), rendering an endowment’s investment policy neutral? We show that donor altruism (even if impure) breaks neutrality. Under a condition—which quickly converges to standard DARA preferences in the number of donors—endowment risk-taking reduces donor free-riding, is Pareto improving, and required by endowment competition. Contrary to asset-liability matching, large endowments optimally take substantial risk even if donors are very averse to changes in endowment-supported spending, reversing the standard moral hazard interpretation. Total giving grows unbounded in the number of donors even with pure altruism.