A Theory of Structural Change That Can Fit the Data

We study structural change in the historical consumption expenditure of the United States, the United Kingdom, Canada, and Australia over more than a century. We characterize the most general class of preferences in a time-additive setting that admits aggregation of the saving decision and allows us to identify preference parameters from aggregate data. We parameterize and estimate such intertemporally aggregable (IA) preferences and discuss their properties in a dynamic general equilibrium framework with sustained growth. Our preference class is considerably more flexible than the Gorman form or PIGL, giving rise to a good fit of the non-monotonic pattern of structural change.