In this paper we analyse the determinants of structural change based on historical simulations. We calibrate a recursive dynamic Eaton-Kortum quantitative trade model featuring multiple sectors, intermediate linkages and non-homothetic CES preferences to WIOD data from 1995 until 2011. We conduct historical simulations imposing actual growth rates of GDP, employment by skill type, population, changes in trade balance, savings rates and trade costs, estimated sectoral differences in productivity growth, and changes in intermediate demand. Changes in trade costs are based on actual changes in both tariffs and NTMs, instead of inferring trade cost changes based on the rise in international relative to intra-national trade. We use the discrepancy between simulation outcomes and actual data to estimate historical changes in the structure of intermediate demand of firms. We use the model simulations to identify the most important determinants of two important changes in the structure of the global economy between the 1990s and the 2010s: (i) structural change, measured by the falling share of manufacturing and the rising share of services in production; and (ii) the increasing import competition from emerging economies. The analysis compares the following determinants: (i) shifting preferences because of rising incomes (non-homothetic preferences); (ii) differential productivity growth; (iii) changes in the intermediate demand structure; (iv) economic growth in developing countries; (v) falling trade costs; (vi) changes in the savings rate.