The evolution of regional macroeconomic dynamics in the United States

Nations like the United States are made up of several diverse geographic regions and it is well known that economic activity is not uniformly distributed across these diverse regions. This study examines several issues related to regional macroeconomic dynamics over short and Ions run time horizons. ^ In Chapter I, I reexamine the issue of cointegration in regional economies and present evidence in support of a long-run relationship between BEA regions. I propose an alternative explanation for the apparent lack of bivariate cointegration between U.S. and regional economies found in earlier studies. ^ In Chapter II I review the theoretical and empirical evidence for convergence, discuss the criticisms of convergence, and show how the time-series evidence for cointegration can be applied to the issue of convergence among U.S. regions. I extend Bernard and Durlaf (1995, 1996) and apply their framework to the U.S. BEA data to show the importance of cointegration in the regional data. I also find empirical evidence for structural convergence using 1-digit SIC code employment data for the BEA regions. ^ In Chapter III I examine the evolution of regional business cycles using non-parametric techniques developed by Pagan and Harding (2001a, 2001b). I apply the analysis to BEA regional employment series over several historical business cycle episodes. The results of the non-parametric analysis indicate the presence and strength of a common component in the regional business cycles. ^ In Chapter IV I employ a variety of time-series techniques in order to examine the role of region-specific and aggregate forces in regional economies across the United States. I begin with an examination of the sensitivity of regional employment to changes aggregate employment. I perform an analysis of bivariate VAR models of employment growth to make inference about the role that shocks to aggregate employment growth have in explaining regional fluctuations. Finally, I modify the Clark and Shin\u27s (2000) disaggregate VAR factor model, which allows me to take advantage of the cointegration between the regions and to test for dynamic stability by breaking the sample into sub-periods.