Thesis (Ph.D.)--Massachusetts Institute of Technology, Dept. of Urban Studies and Planning, 2000. / "February 2000." / Includes bibliographical references. / Three studies were conducted to analyze the relationship between public infrastructure investment and regional economic performance. The first study examines the literature on economic development and productivity growth. I show that conflicting results from studies by other analysts are the likely result of poor public capital data spanning to short an interval, and an inadequate modeling framework. Public investment may generate small improvements in productivity, but models understate economic impacts owing to the public goods character of some forms of public capital. The second study explores the relationship between economic distress and public infrastructure investment. I use a sample of U.S. counties to analyze public investment according to level of economic distress. With simple investment models, I estimated infrastructure needs for counties with apparent shortfalls. I analyzed the needs-estimates in a series of case studies in which jurisdiction planning and budget personnel were consulted about the accuracy of the estimates. I show that short-run economic distress is not to be linked to public infrastructure investment. Over the long-run, investment varies by level of distress, but as a consequence of private residential investment. The needs-estimating models were reasonably accurate, but missing investment data proved troublesome. Counties proved to be a poor unit of analysis for infrastructure needs, as since significant variation was observed among jurisdictions within counties. The third study demonstrates the need for better estimates of public infrastructure capital stock. I prepared new capital stock estimates for two regions using local investment data and survey-based public capital service lives. I surveyed one thousand jurisdictions in the New England region and the state of Texas. Survey-based service-lives seem to differ significantly from estimated lives. Stock estimates using local investment data and survey-based service-lives produce dramatic differences compared to estimated stocks at the state and regional level. The new data, however, performed just as poorly as other series when used to estimate aggregate production functions. Prior analysts' understanding the relationship between economic performance and public infrastructure investment has been limited because of poor data, and inadequate appreciation of infrastructure's inherent complexity. The research presented here demonstrates that significant improvements are possible and worth undertaking. / by Nicholas O. Rockler. / Ph.D.