Modelling smooth and uneven cross-sectoral growth patterns: an identification problem

This paper shows that the available stylized facts on productivity dynamics, such as persistent cross-sectoral heterogeneity, do not allow to solve an identification problem regarding the impact of common drivers - such as General Purpose Technologies (GPTs) - on economic growth. The evidence of persistently heterogeneous productivity performances is consistent both with a GPT-driven model, and with a model characterized by purely independent and idiosyncratic sectoral dynamics. These results are obtained within a simple theoretical framework, and illustrated with reference to measures of concentration of the sectoral contributions to aggregate total factor productivity growth.