This paper provides an explicit welfare basis for evaluating economic mobility. Our social welfare function can be seen as a natural dynamic extension of the static social welfare function presented in Atkinson and Bourguignon (1982). Unlike Atkinson and Bourguignon, we use social preferences a la Kreps-Porteus, for which the timing of resolution of uncertainty may matter. Within this generalized framework, we show that welfare evaluation of mobility depends on the interplay between aversion to inequality, risk aversion, and aversion to intertemporal fluctuations. This framework allows us to provide a welfare analysis not only of "reversal" (which has been the focus of much of the literature) but also of "origin independence" (which has not received an explicit welfare foundation in the literature). We use our framework to develop welfare measures of mobility, and apply these measures to intergenerational mobility in the United States using PSID data. We show that the value of origin independence is quantitatively important. We also show that different subpopulations experience different mobility patterns: reversal is more important than origin independence for blacks but the opposite is true for non-blacks.
