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