Life-cycle consumption: can single agent models get it right?

In the quantitative macroeconomics literature, single agent models are widely used to explain ``per-adult equivalent'' data, which are obtained at the household level. In this paper we suggest a simple framework to understand the sources of bias when these models are used to make predictions for aggregate consumption. In both a theoretical and a quantitative exercise, we find that economies of scale in consumption inside the household are positively related to the bias introduced by the single agent approach in predicted consumption profiles over the life-cycle. We also do an external validation exercise, which suggests that economies of scale inside the household are rather large, pointing out the need to approach life-cycle consumption with models that consider households rather than single agents.