My dissertation studies the dynamics of specialization patterns in relation to economic growth. The first chapter employs a quantitative framework to study whether and how knowledge diffusion could explain the cross-country convergence in industry-level productivity and its implications on specialization dynamics. The second chapter focuses on the network aspects of economic diversification in an open-economy context. The third chapter studies the path dependence of evolving comparative advantage in an endogenous growth model. The fourth chapter conducts numerical experiments for models of endogenous growth cycles in both closed- and open-economy settings.The first chapter studies the dynamic evolution of the patternsof Ricardian comparative advantage from the perspective of knowledge diffusion. The theoretical analysis builds knowledge diffusion into a quantifiable model of trade by allowing for industry-level productivity to evolve through a spatial flow of ideas. This may take place through four channels: Firms could upgrade their technology via meetings with domestic producers and foreign sellers, and meetings are both intra- and interindustry. Thistheoretical framework yields a law of motion of industry-level productivity across countries, capturing strong interdependence in theevolution of Ricardian comparative advantage. I calibrate the model to a large sample of countries. My quantitative results capture important patterns in the data: There is strongconvergence in industry-level productivity and substantial mobility in specialization patterns. A decompositionexercise based on the theoretical law of motion suggests that international and interindustry channels play a major role in knowledge diffusion. The framework yields additional quantitative implications. Analysis of the knowledge-diffusion network facilitates the identification of the countries or country--industry pairs that contribute most to global productivity growth. The calibrated model also suggests that dynamic gains from trade through knowledge diffusion are economically significant, amounting to at least one third of static gains from trade.The second chapter studies how specialization patterns evolve through dynamic formation of global input--output networks to provide a network-theoretic explanation for empirical findings reported in the literature concerning the network structure of economic output. Production is modeled as fragmented across borders, so countries specialize in different parts of the global production network. As in Carvalho and Voigtl;;ander (2015), which is extended into a multicountry setting in this chapter, a new product is invented each period by recombiningexisting products through two steps. First, the new product is given its defining features by drawingprimary inputs from the set of intermediate inputs of its predecessor(s). Second,additional inputs are randomly sampled from the rest of the global input--output network.A country is more likely to adopt a new product if it currently produces more of its inputs. The centrality of a product is measured by its importance as an intermediate input in the production network. My theoretical and simulation results suggest that there is a negative correlation between diversity and ubiquity of a country;;s product mix, provided that the agglomeration forces are sufficiently strong, echoing the empirical findings in the literature. The simulation exercise demonstrates that a countryinitially specializing in more central products tends to diversify its production structure more quickly, consistent with what the literature of the product space documents.The model also provides implications on the evolution of the global production network and unconditional convergence across countries.The third chapter offers a novel channel through which patterns of specialization affect economic growth. I build into a two-country, quality-ladder model of growth the concept of the technologicalrelatedness of products. In the model, firms in the North innovate, and firms in the South imitate. The product space is modeled as a circle on which the distance between two productsmeasurestechnological relatedness.For southern firms, the ``closer;;;;a product is to the domestic product mix in the South, the higher theefficiency of imitation. Similarly, efficiency of innovation in the North decreases with the distance betweenthe northern product mix and the product to be innovated. This gives rise to multiple balanced growth paths. In a separating equilibrium, there is no overlap between the ranges of the products in which two countries specialize. Due to low efficiency of innovation and imitation, firms in each country have little incentive to enter the part of the product space that is occupied by the other country, dampening global economic growth. A pooling equilibrium follows the symmetric balanced growth path in Grossman and Helpman (1991).The range of products each country produces overlaps with each other, so the efficiency of imitation and innovation is at an intermediate level. Firms actively engage in innovation and imitation, thus bringing about constant shifts of production location between the North and South, namely, product cycles. The model demonstrates the importance of the product space in understanding the nexus between trade and growth.The fourth chapter provides numerical analysis of two maps that arise from models of endogenous growth cycles in closed- and open- economy settings. The numerical results shed light on the existing technical work on Matsuyama (1999). Several interesting patterns emerge, calling for future analytical investigation.
PDF
download
878987797
028-85220240
