State and local governments invested 13% of their revenues on infrastructure in 2002, but existing literature provides only mixed evidence that infrastructure contributes to economic activity. To estimate the effect of infrastructure on the economy, Chapter 1 analyzes how the construction of the Interstate Highway System (IHS) contributed to regional development in the United States by expanding intercity trade, using data on the construction of the IHS, intercity trade and regional economic activities. Empirical results provide evidence that the IHS reduced driving times among cities and subsequently increased intercity trade in the following two ways. First, it increased the volume of trade among existing trading partners. Second, it increased the probability of trade among cities that previously did not trade. Moreover, trade expanded more for nationally traded goods more than for locally traded goods, because the former relied more on the IHS than the latter. By expanding trade, the IHS increased regional output, employment and firm entry.
Existing literature provides mixed evidence on whether infrastructure contributes to economic growth, because of a problem of reverse causality--better infrastructure may not lead to higher growth, but regions with higher growth may invest more in infrastructure. Chapter 2 uses an instrumental variable to identify the impact of the construction of infrastructure, focusing on the construction of the Interstate Highway System (IHS). A close link exists between the 1956 Interstate Highway (IH) plan and pre-existing economic prosperity. But both historical evidence and regression analysis show that construction priority among highway segments partly depended on how easy it was to build those segments; and ease of construction is plausibly exogenous to economic growth. We instrument the open-to-traffic time for each segment of highway using construction costs from the 1958 Interstate Cost Estimates. We apply the instrument to estimate the effect of new IH on driving times, and find that an OLS regression under-estimates the effect of IH on driving times.