Can Marginal Rates of Substitution Be Inferred from Happiness Data? Evidence from Residency Choices