Essays on productivity and sectoral differences

This dissertation uses sectoral data in production and patents to provide explanations for previous findings and improves the measurement of technology diffusion and productivity differences across countries. Chapters 1 and 2 deal with measuring technology diffusion. In Chapter 1 I document stylized facts regarding technology diffusion across pairs of countries in different types of goods. I focus primarily on the composition of the patents that are requested by researchers from one country in any given destination. In the data, for a given origin, some types of ideas are more likely to be patented in richer, more technologically advanced countries, while other types of ideas are more likely to be patented in poorer, less advanced destinations. In chapter 2 I tackle the challenge of measuring international technology diffusion by proposing a model that allows for more precise estimation of bilateral diffusion rates between countries compared to previous literature using the data presented in Chapter 1. Due to the abstract nature of technology and, consequently, the limited amount of data that can be used this has been a particularly difficult task. In most cases it has been possible to measure how foreign research affects a country's productivity or even how good one country is at importing or exporting knowledge from abroad. Nevertheless, measurement of how much each partner contributes to the technological development of each country has not been performed successfully. My approach is to generalize an Eaton-Kortum model for diffusion of ideas to match empirical facts regarding patent data. In the model proposed the heterogeneity in the predicted data arises from the relative differences in technological progress of the pair of countries involved. I argue that this extension contributes to measuring international technology diffusion more precisely. This is possible since the generalization of the model requires the introduction of a few parameters, while allowing for a much larger data set to be used in the estimation.;Chapter 3 deals with a productivity question. Traditionally, variance decomposition exercises across countries result in total factor productivity (TFP) accounting for about two thirds of the variance in GDP per capita. In most cases a country is modeled as a one-sector economy. Recent work by Ferreira et al. (2008) that applies this methodology to 31 years of data shows that this result does not hold in the early seventies: disparities in physical and human capital were more relevant in explaining income differences. I expand the model to include agriculture, a sector assumed to be different in terms of the production function where land in particular plays an important role. I show that ignoring the existence of this sector can cause the importance of TFP to be overestimated due to the productivity of land being implicitly assigned to it, and that this overestimation fades away as the share of agriculture in production decreases. I show that part of the increase in the importance of TFP in explaining the variance of GDP per capita can be explained by the structural transformation that most economies have experienced recently, especially poorer ones.