We develop a model that fleshes out, extends, and modifies existing models of reference dependent preferences and loss aversion while accommodating most of the evidence motivating these models. Our approach makes reference-dependent theory more broadly applicable by avoiding some of the ways...