The Puzzling Income Elasticity of US Imports

Existing estimates of the income elasticity suggest that, in the absence of price increases, US imports will eventually exceed US income. That the United States will change from being a largely self-sufficient economy to one that cannot cover its imports has received a great deal of attention in developing models for imports: allowance for simultaneity in the import market, recognition of dynamic adjustments and optimization, disaggregation of imports across product and countries, removal of measurement errors in official data, and differentiation between cyclical and secular forces. I show that previous methods to resolve the puzzle, though insightful, are not successful: three decades of methodological improvements in modeling and estimation yield estimated income elasticities for imports much greater than one. I resolve this puzzling estimate by removing the representative-agent assumption and addressing the substitution bias embodied in official import prices stemming from their exclusion of new products. I follow Feenstra to address the substitution bias induced by the omission of new-products' prices. To remove the representative-agent assumption, I start with an individual's demand for imports depending on income and relative prices. Then, consistent aggregation of the micro relations yields a macro equation with income, relative prices, and the variances of the distributions of imports, income, and relative prices. These variances embody individuals' heterogeneity which I model using the share of foreign-born individuals in the United States. For parameter estimation, I use the cointegration method of Johansen (1988). Recognizing either the substitution biases in import prices or the heterogeneity induced by immigration point to unitary income elasticities. Excluding these factors and using the conventional model yields estimated income elasticities much greater than one.