Abstract: Aggregation is a tool used to reduce the complexity of economic models in order to draw more clear and succinct conclusions or simplify analyses. As any approximation, its use may be accompanied with errors researchers may not be willing to tolerate if they become aware of them. In this work I present how these errors appear using aggregation across goods and across consumers. To this end, I consider aggregation as a means to approximate probability distributions over parameters. Using this approach, I show ways to bound approximation errors by tailoring the parameters of the model. Further, I briefly discuss a methodology to study the goodness-of-fit of aggregate models in more general settings.