We reassess the empirical effect of income and employment on self-reported well-being. Our analysis makes use of a novel two-step estimation procedure that allows applying instrumental variable regressions with ordinal observable data. As suggested by the theory of incomplete markets, we differentiate between the effects of persistent and transitory income shocks. In line with this theory, we find that persistent shocks have a significant impact on happiness while transitory shocks do not. This has consequences also for inference about the happiness effect of employment. We find that employment per se is rather associated with a decline in happiness.