This paper provides and implements a method to estimate the expected marginal utility of income across observable traits like income. Good policy analysis tries to match people to the policy that best fits their normative beliefs. In a welfarist framework this requires knowing something about the marginal utility of income, but despite over a century of effort, economics has not agreed upon a method for estimating utility. The key assumption for my method is that the marginal utility of pain relief is independent from a person's income. While the utility I estimate may not actually be marginal well-being or happiness, it is the marginal utility implied by standard economic assumptions that justify using willingness to pay as a measure of welfare. Surprisingly, I find the implied marginal utility of income does not change across income. This result suggests that more research is needed to properly interpret policy analysis using willingness to pay.