Working Paper: NBER ID: w10319
Authors: Tibor Besedes; Thomas J. Prusa
Abstract: We examine the extent that product differentiation affects the duration of US import trade relationships. Applying nonparametric and semiparametric techniques to highly disaggregated product-level data we estimate that the hazard rate is at least 18 percent higher for homogenous goods than for differentiated products. Put another way, the median survival time for trade relationships involving differentiated products is five years as compared to two years for homogenous products. We find that our results are not only highly robust but often are strengthened under alternative specifications. For instance, if we define trade relationships using industry-level rather than product-level data we find that the hazard rate is 30-35 percent higher for homogenous goods than for differentiated products. We also find that the survival ranking across product types holds across individual industries. We show that dropping the smallest trade relationships further accentuates the differences among product types. We also control for the possible measurement error in measuring spell lengths and the role of multiple spell relationships and find that in all cases the differences among products types are greater than in our benchmark analysis.
Keywords: Product Differentiation; Trade Relationships; Duration; Hazard Rate
JEL Codes: F1
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Product differentiation (L15) | Duration of U.S. import trade relationships (F10) |
| Homogeneous goods (H41) | Hazard rate for trade relationships (C41) |
| Differentiated products (L15) | Duration of U.S. import trade relationships (F10) |
| Industry-level data (L69) | Hazard rate for homogeneous goods (C41) |
| Larger trade relationships (F10) | Hazard rate for homogeneous goods (C41) |
| Larger trade relationships (F10) | Hazard rate for reference priced goods (D11) |