Pre-print, Working paper: There is now a great deal of micro-econometric evidence, both cross-section and panel, showing that income is positively correlated with well-being. Yet the famous Easterlin paradox shows essentially no change in average happiness at the country level, despite spectacular rises in per capita GDP. We argue that survey well-being questions are indeed good proxy measures of utility, and resolve the Easterlin paradox by appealing to income comparisons: these can be to others (social comparisons) or to oneself in the past (habituation). We review a substantial amount of econometric, experimental and neurological literature consistent with comparisons, and then spell out the implications for a wide range of economic issues.