The US Trade Deficit: A Disaggregated Perspective

The paper prepares new estimates for the elasticity of US trade flows using bilateral, commodity-detailed trade data for 31 countries, using measures of expenditure and trade prices matched to commodity groups, and including a commodity-and-country specific proxy for global supply-cum-variety. Using the United Nations Commodity Trade Statistics Database (UN Comtrade) we construct bilateral trade flows for 31 countries in four categories of goods based on the Bureau of Economic Analysis’s “end-use” classification system—autos, industrial supplies and materials–excluding energy, consumer goods, and capital goods. We find that using expenditure matched to commodity category yields more plausible values for the demand elasticities than does using GDP as the measure of demand that drives trade flows. Controlling for country and commodity fixed effects, we find that industrial and developing countries have demand elasticities that are statistically significant and that generally differ between development groups and across product categories. Relative prices for the industrial countries have plausible parameter values, are statistically significant and differ across product groups, but the relative prices for developing countries are poorly estimated. We find that variety is an important variable for the behavior of capital goods trade. Because the commodity composition of trade and of trading partners has changed dramatically, particularly for imports, we find that the demand elasticity for imports is not constant. Comparing the in-sample performance of the disaggregated model against a benchmark that uses aggregated data and GDP as the expenditure variable, our disaggregated model predicts exports better in-sample but does not predict imports as well as the benchmark model.