Spatial dimensions of economic growth: Technological leadership and club convergence
Abstract
This dissertation investigates the spatial dimensions of regional economic growth processes, with particular attention for technological leadership and convergence clubs, and contains empirical applications as well as a new modeling and estimation method. The first essay of this dissertation investigates the extent to which technology and technological leadership affects regional employment growth, concurrently accounting for differences among sectors and the existence of spatial spillover effects. Results indicate a strong and positive role of the regional stock of human capital in all sectors. The effect of technological distance and geographical distance on employment growth, however, varies across sectors but seems to be particularly strong in the manufacturing and the service sectors. The general effect of agglomeration economies varies across sectors as well, but evidence of the positive impact of diversity seems to be consistent across sectors. In the second essay of the dissertation a statistical method to investigate nonlinearity and spatial heterogeneity in spatial processes is developed. A spatial version of the Time series STAR is developed that incorporates standard spatial autoregressive processes in the STAR framework, and effective maximum likelihood estimation and testing procedures are introduced. The finite sample properties of the new tests for nonlinearity and/or spatial autoregressive processes are investigated through Monte Carlo simulations. All statistical tests show relatively good size and power characteristics in small samples. The new spatial STAR method is applied to U.S. county growth data for the period 1990–2007 in the third essay. The goal is to determine the “membership” of the different convergence clubs endogenously within the model. Three convergence clubs were detected: a rather broad club of counties with a large convergence rate, a very small group of counties with a relatively low convergence rate, and a rather extensive group of counties with in between values of their convergence rate. A comparison of the spatial STAR model results with estimates obtained using Geographically Weighted Regression or an a priori, exogenously defined, distinction between two groups reveals that the spatial STAR model provides a more realistic and accurate representation of the economic growth process in U.S. counties.
Degree
Ph.D.
Advisors
Florax, Purdue University.
Subject Area
Agricultural economics
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.