Solving the Puzzle: A New Measure of Trade Distance In The Gravity
Fecha
2014-06-11Metadatos
Mostrar el registro completo del ítemResumen
The gravity model is a workhorse tool that has been widely used in international trade. However, one empirical question that frequently arises is related to the conceptualization and measurement of distance. To overcome this limitation, our study proposes an index of distance based on multivariate statistical analysis. Specifically, we build our index using Factorial Analysis for Mixed Data. For robustness check, we use Principal Component Analysis. Both techniques summarize in one factor information related to geographical, cultural, political and economic variables that might affect international trade between countries. We use this index as proxy of distance, and Gross Domestic Product as proxy of mass, and we run some panel data exercises between 1995 and 2000 for 10 Latin American economies. Estimations indicate that the sign of the load factors in Factor Analysis for Mixed Data are intuitively plausible, and that panel data exercises give sensible robust outcomes.
Resumen
The gravity model is a workhorse tool that has been widely used in international trade. However, one empirical question that frequently arises is related to the conceptualization and measurement of distance. To overcome this limitation, our study proposes an index of distance based on multivariate statistical analysis. Specifically, we build our index using Factorial Analysis for Mixed Data. For robustness check, we use Principal Component Analysis. Both techniques summarize in one factor information related to geographical, cultural, political and economic variables that might affect international trade between countries. We use this index as proxy of distance, and Gross Domestic Product as proxy of mass, and we run some panel data exercises between 1995 and 2000 for 10 Latin American economies. Estimations indicate that the sign of the load factors in Factor Analysis for Mixed Data are intuitively plausible, and that panel data exercises give sensible robust outcomes.
