We propose a general framework to study whether and how common trends and common cycles are still present when the original variables are linearly aggregated or only a subset of them is analysed. This is particularly important because of the adoption in empirical analysis of aggregated data on a limited number of variables.