Industry structure and regional growth: A vector autoregression forecasting model of the Wisconsin regional economy

One of the oldest and most durable theories of regional growth is economic base theory. Despite its wide acceptance as a framework for applied regional economic analysis, economic base theory and its methodological techniques have long been criticized as inadequate for such purposes. In recent years, however, time-series modeling and econometric techniques have been employed to provide additional evidence in support of the economic base hypothesis. The purpose of this research was to confirm and extend the results of these time-series econometric models of regional economic activity.The results of the empirical tests conducted in this research fall into three distinct categories. The first category of results are those in which the traditional formulation of the economic base hypothesis was subjected to a set of four time-series econometric tests of the economic base hypothesis. The results of this category of tests were conclusive: They provided no evidence to support the validity of the traditional formulation of the economic base hypothesis.The second category of empirical results were those in which an alternative formulation of the economic base hypothesis was considered. An out-of-sample forecasting competition was established as a test of this alternative hypothesis. The results of this forecasting competition suggested that both economic base and input-output theoretic vector autoregressions captured significant information about the nature of the Wisconsin regional economy.The third and final category of empirical results provided information with respect to the robustness debate of the macroeconomics time-series literature. Two issues in particular, the time aggregation issue and the trend specification issue, were addressed within the context of the first set of tests of the economic base hypothesis. With respect to the time aggregation issue, the results of this research indicate that time aggregation of the data can have an effect upon inferences drawn within the context of any given hypothesis testing procedure. By contrast, the inclusion of a time trend does not appear to alter the results of such hypothesis tests.