A Theory of Structural Change That Can Fit the Data

We study structural change in historical consumption expenditure of theUnited States, the United Kingdom, Canada, and Australia over morethan a century. We characterize the most general class of preferencesin a time-additive setting that admits aggregation of the saving deci-sion and allows to identify preference parameters from aggregate data.We parametrize and estimate such intertemporally aggregable (IA) pref-erences and discuss their properties in a dynamic general equilibriumframework with sustained growth. Our preference class is considerablymore flexible than the Gorman form or PIGL, giving rise to a good fit ofthe non-monotonic pattern of structural change.