"Structural change" literature analyzes the change the sectoral composition of GDP and implies changes in composition of labor which are inconsistent with data. I explore the power of having heterogeneous labor in a two-sector economy to explain the evolution of employment in the United States between 1956 and 2000. The key aspect of the model is that industries demand labor force "variety" differently. In particular, services employs a more diverse labor force, while manufacturing does not, as is well documented. The calibrated model shows that the impact of a growing labor force reflects on industries employing different amounts of workers to satisfy their needs, thus resulting in the observed data patterns.

Industry level data shows striking differences among sectors in their choices of export sales and foreign direct investment sales (i.e., export-sales intensities). Previous attempts to explain this phenomenon, such as in Helpman, Melitz, and Yeaple (2004), have relied primarily on product tradabilty, fixed setup costs, and firm productivity dispersion. In a standard Dixit-Stiglitz model of monopolistic competition, we show that these margins are not enough to account for the variation of export sales to FDI sales observed in the data. We calibrate two variations of this model: sector-specific FDI sales taxes and sector-specific home bias, both of which allow us to match the sector intensities as well as a measure of fixed operational costs of multinational firms.

[Extended Abstract]