We derive asset-pricing and portfolio-choice implicationsof a dynamic incomplete-markets model in which consumers areheterogeneous in several respects: labor income, asset wealth, andpreferences. In contrast to earlier papers, we insist on at leastroughly matching the model's implications forheterogeneity — notably, the equilibrium distributions of income andwealth — with those in U.S. data. This approach seems natural:Models that rely critically on heterogeneity for explaining assetprices are not convincing unless the heterogeneity is quantitativelyreasonable. We find that the class of models we consider here isvery far from success in explaining the equity premium whenparameters are restricted to produce reasonable equilibriumheterogeneity. We express the equity premium as a product of twofactors: the standard deviation of the excess return and the marketprice of risk. The first factor, as expected, is much too low in the model. The size of the market price of risk depends crucially on theconstraints on borrowing. If substantial borrowing is allowed, themarket price of risk is about one one-hundredth of what it is in the data(and about 15% higher than in the representative-agent model).However, under the most severe borrowing constraints that we consider, the market price of risk is quite close to the observedvalue.