Auditing inequity in resource allocation under unobserved confounding is possible with informative bounds, as demonstrated in a real-world study.