Abstract
The first chapter in this dissertation studies the importance of distinguishing between intermediate and final use for the gains from trade. The domestic expenditure share, which is central to calculating the gains from trade across a wide class of models, varies by end use. Failure to account for this heterogeneity in a one-sector, one-factor model systematically understates the gains from trade and conceals differences in the returns to openness across end use. I construct a multi-sector, multi-factor model with input-output linkages that incorporates variation in end use to investigate the full extent of these discrepancies, and to explore the relationship between relative income and intermediate relative to final use estimates of Ricardian comparative advantage, trade costs, and prices. I estimate the parameters of the model for 38 countries and 32 manufacturing and service industries using the World Input-Output Database. Lower income countries have a comparative disadvantage in producing and importing intermediate relative to final goods, which results in a higher relative price of intermediates for these countries. Including end-use variation raises the gains from trade by 14.4 percent on average.
The second chapter examines the relationship between distance and bilateral trade at the industry level, where zero trade flows are prevalent. I expand upon the work of Berthelon and Freund (2008) by incorporating zero trade flows into the estimation using Tobit and Poisson pseudo-maximum-likelihood methods. Methods that incor- porate zeros reveal that distance sensitivity is either decreasing (Tobit) or increasing only slightly (PPML) over time. This stands in contrast to the OLS approach, which shows a significant increase in distance sensitivity. I find that more substitutable goods and those with higher trade costs are likely to exhibit higher sensitivity to distance, but that these qualities cannot explain the small and often insignificant changes in Poisson-estimated distance elasticity over time.
The last chapter decomposes U.S. water use, which has followed a remarkable pattern since 1950, not mimicking the almost uninterrupted 110 percent increase in the size of the U.S. population, the relatively steady 570 percent growth in real GDP, and the 220 percent improvement in per capita GDP. After doubling between 1950 and 1980, the total volume of water withdrawn has stabilized and even decreased in recent years. Our decomposition shows that between 35 and 50 percent of the productivity gains that allowed the U.S. to produce each dollar of its GDP with increasingly less water stems from long-term structural changes of the U.S. economy since 1950 (growing service economy, declining manufacturing and agricultural sectors). The remaining 50 to 65 percent is due to improved production techniques, and in particular due to water productivity improvements in the electricity-generating sector, especially since the mid to late 1970s. We argue that while globalization has helped reduce U.S. water use particularly since 1980, the U.S. ability to import more water-intensive goods is not the main reason U.S. water use plateaued.
