Section (2) presents a model to infer changes in MWP for non-wage characteristics from data on job leaving, adapting Gronberg and Reed (1994) to our context. Section (3) details our estimation of the determinants of job leaving with a focus on the estimators' treatment of heterogeneity and duration dependence. Section (4) presents the data. Section (5) provides evidence of changing job preferences by tracing the evolution of wages and job satisfaction for individuals who receive a windfall and subsequently change jobs. Section (6) estimates changes in MWP using all job leavers. Focusing on this larger set allows us to include the effect of wealth on transitions to non-participation in the analysis. Large changes in wealth, especially relative to income, are found to significantly increase MWP to pay for non-wage characteristics. Section (7) puts these findings into perspective and concludes.