The euro area represents a case-study of great institutional relevance for the econometric problem of aggregation bias. The available data can be used to analyze the area either with aggregate or with country-specific models. The choice should be the result of a statistical comparison between the two options, with respect to the specific model. In this paper we suggest a representation of the aggregation error based on unobservable components and explicitly conceived for aggregations over a small number of economies. In the empirical application two alternative models are estimated: the first specifies the main euro countries while the other refers to the whole area. We then evaluate the aggregation error either from the viewpoint of a comparison of the two models with standard methods, or looking at the components of the representation suggested here. Both categories of results indicate non-negligible aggregation errors for the euro area.
Published in 2007 in: Economic Modelling, 24, pp. 236-261