This paper is about a useful way of taking account of frictions in asset pricing and macroeconomics. I start by noting that complete frictionless markets models have a number of empirical deficiencies. Then I suggest an alternative class of models with incomplete markets and heterogenous agents which can also accommodate a variety of other frictions. These models are quantitatively attractive and computationally feasible and have the potential to overcome many or all of the empirical deficiencies of complete frictionless markets models. The incomplete markets model can also differ significantly from the complete frictionless markets model on some important policy questions.