Essays on international trade

This dissertation consists of three essays pertaining to the causes of the levels and composition of the international trade flows of nations, and the consequential implications for the levels of per capita income and welfare of their populations. The first of these documents a pattern of comparative advantage in product level, bilateral trade data that conventional quantitative trade models have difficulty explaining. It goes on to develop a theory of product level productivity differences based on endogenous differences in the allocation of research and development into product and process innovation across countries over time, and it shows that, when fitted to cross-country manufacturing wage data, the predicted product level technology distribution is consistent with the observed trade pattern.

The second essay shows that the distribution of technology levels inferred in the first essay can help explain the inability of both ad-hoc and theoretically based gravity models of trade to account for the observed positive correlation between the percentage of manufacturing output that is traded and countries' per capita income. It derives a modified gravity equation based on a Ricardian model of trade with deterministic product level technology differences across countries. It then uses estimates from a product level gravity estimation to compute the component of this equation that differs from a conventional gravity equation in order to determine the extent to which the observed concentration of comparative advantage in a common set of products for low-income countries explains the small percentage of their output that is exported.

The final essay shows that a simple model of firm profit maximization in the presence of sunk costs of entering the export market is broadly consistent with the observed persistence of exporting behavior in firm level data. It uses this simple model and moments from data on US manufacturing firms to estimate the value of the sunk export entry costs faced by these firms using an indirect inference strategy. These costs are shown to be substantial relative the revenue stream of a typical firm.