Caplin and Leahy (2004) propose a major shift in how economists should think about welfare analysis. They demonstrate that the revealed preference welfare criterion is better suited to static analysis and is awkward in dynamic settings because it implies that policy makers would arbitrarily favor the most impatient decision rules from the Pareto set. To measure the economic significance of Caplin and Leahy's critique, we extend their conceptual framework to a quantitative setting of life-cycle consumption and saving. We find that using the revealed preference measure understates the socially optimal level of savings by a factor of 2 at our preferred parameterization. This is because the revealed preference criterion ignores the regret that rational individuals naturally feel about their past saving decisions. We relate our findings to the Fundamental Theorems of Welfare Economics, and we also argue that the standard rational framework is in fact consistent with phenomena (regret and undersaving) that are widely used as motivation to abandon it in favor of psychology-based theories like hyperbolic discounting.