Geography, non-homotheticity, and industrialization: a quantitative analysis

We propose a quantitative framework for the analysis of industrialization in which specialization in manufacturing or agriculture is driven by comparative advantage and non-homothetic preferences. Countries are integrated through trade but trade is not costless and geographic position matters. We use a number of analytical examples and a multi-country calibration to explain two important empirical regularities: (i) there is a strong positive correlation between proximity to large markets and levels of manufacturing activity; (ii) there is a positive correlation between the ratio of agricultural to manufacturing productivity and shares of manufacturing in GDP. Our calibrated model replicates these facts and also provides a better fit to cross-sectional data on manufacturing shares than frameworks which ignore the role of trade costs or non-homotheticity. We use the calibrated model to quantitatively analyze the effect of increases in agricultural productivity and a further lowering of trade barriers.