"The Importance of Being Different: Product Differentiation and the Duration of Trade,"
            joint with Thomas J. Prusa, (November 2002)
This paper studies how product differentiation affects the dynamics of US import trade. Using nonparametric and semiparametric techniques we find that trade in homogenous and reference priced products is of much shorter duration than differentiated products. Differentiated products have a median survival time of five years; reference priced goods and homogenous products have a median survival time of only two years. As compared with differentiated products the hazard rate for reference priced goods is 16 percent higher and for homogenous goods is 20--24 percent higher. Importantly, we find that these results are not driven by small trade relationships. To the contrary, we find that the differences in survival rates among the product types increases when we drop the smallest trade relationships. If we limit our analysis to those observations whose trade value in the first year is greater than $100,000 we find that the hazard rate for reference priced goods is 50 percent higher and for homogenous goods is 60--85 percent higher. These results provide compelling evidence that trade in differentiated goods is different than in homogenous goods.

"Ins, Outs, and Duration of Trade,"
            joint with Thomas J. Prusa, (November 2002)
This paper uses a new approach to study the dynamics of international trade. I employ survival analysis to study the duration of US imports. Several important conclusions are reached. The average duration of exporting a product to the US is very short, anywhere between 2 and 4 years. However, this result varies significantly across countries and industries. The data imply a threshold model of trade duration. If a country is able to export a product for the first few years it will face a very small probability of failure and will export the product for a long period of time. This result holds across countries and industries. Aggregating the data causes the duration to worsen with decreases in the median duration. The relative survival experience, however remains unchanged. Countries are ranked by their survival experience. The rankings are strongly positively correlated with the rankings in Feenstra and Rose (2000), indicating that product cycle models can explain duration of trade.

  "Decomposing the Growth of Trade 1884-1992" (August 2001)
This paper uses a Ricardian model of trade to study the growth of international trade from 1884 to 1992. The Dornbusch, Fischer, Samuelson (1977) Ricardian model with a continuum of goods is used following Evenett and Yeung's (2000) derivation of a relationship between trade and trade impediments. A key supply side parameter is estimated using weighted least squares and correcting for autocorrelation in the residual. This parameter is then used to decompose the growth of trade for five countries, Canada, Denmark, Sweden, the UK, and US. The main result is that trade impediments themselves cannot alone account for the growth of trade. Supply side changes via technology play an important role in the evolution of trade.