| A debate on the "infrastructure crisis" in the United States has emerged in the last decade, spurred by perceived underinvestment in infrastructure facilities and its possible adverse consequences. Several observers have called for large increases in national infrastructure spending, while others have questioned these proposals. Much of the recent debate has been generated by econometric studies that have provided evidence of the importance of infrastructure. These empirical assessments, however, have encountered much criticism from various perspectives. This dissertation addresses the importance of criticisms stemming from the treatment of space in the statistical analysis of the relationship between infrastructure and economic activity.;Spatial effects (e.g., spatial dependence and spatial heterogeneity) are relevant in two ways, both from the "spatiality" of the impact of much infrastructure (especially transportation) on the economy, and also more generally from the nature of spatial data. Such concerns have been largely ignored in recent studies of the relationship between infrastructure and economic activity, in violation of the assumptions of the statistical techniques often used by researchers. The possible existence of spatial effects may lead to incomplete or erroneous interpretations of the role of infrastructure in local, regional, and national economies, and may result in flawed policy prescriptions. This dissertation represents an attempt to remedy this situation by extending a previous statistical study, conducted at the Federal Reserve Bank of Boston, through an explicit analysis of spatial effects. The results indicate that spatial effects indeed belong in statistical estimates, but that their inclusion does not, in general, alter basic interpretations of the importance of infrastructure in the economy. |
