This article uses survey data of workers in Japan to study the effects of own and self-reported reference wages on subjective well-being. Higher wages lead to higher life and job satisfaction. When workers perceive that their peers earn higher wages, they report lower well-being. We compare our results with relative utility tests in the literature and develop a generalized version of the classical measurement error model to show that the estimated bias of the reference wage effect can go in both directions. We propose an IV strategy when the self-reported reference wage is not available that does not eliminate the bias but delivers a lower bound of the "true" effect.