Most simulated micro-founded macro models use solely consumer-demand aggregates in order to estimate preference parameters of a representative consumer, for use in policy evaluation. Focusing on dynamic models with time-separable preferences, we show that aggregation holds if, and only if, momentary utility functions fall in the Identical-Shape Harmonic Absolute-Risk Aversion (ISHARA) utility class, identifying which parameters of ISHARA utility functions are allowed to vary over time. Given this theoretical result, it should be easy to empirically reject the aggregation properties that the macroeconomic representative-consumer identification approach requires: it suffices to show that permanent incomes guaranteeing the same living standard across households of different size violate an affine relationship. In order to test the validity of this affine equation, we develop a vignette survey that produces appropriate data without demand-estimation restrictions imposed by models. Surprisingly, in six countries, this equation is not rejected, lending support to using consumer-demand aggregates.