Transit investments can affect the clustering of economic activity within a region due to the changes in accessibility that transit can provide, either by increasing firm-based access to the central business district or increasing effective labor market size. This clustering can lead to what are known as agglomeration benefits that increase overall economic productivity and are external to the decisions taken by individual firms. Cost-benefit analysis of transit investments rarely account for such external benefits. Agglomeration benefits work through several mechanisms. Two mechanisms most likely relevant to transit are knowledge spillovers enabled by firm clustering near rail stops and better labor matching due to higher labor market access caused by expansions of transit networks. The actual linkages are complex and are not well captured by simple econometric models. We propose to examine these linkages using a structural equations modeling framework that can account for the direct and indirect effects of transit investment on external productivity benefits within a region.

This work will build upon a TCRP-funded study that the research team has been engaged in. As part of this work we have developed two large data sets to examine agglomeration impacts. One is a nationwide data set of metropolitan areas with measures of GDP, average wages, city size, central city and urbanized area employment density, transit capacity, road capacity, and human capital. We analyzed this dataset using a two-step path analysis relying on standard crosssectional econometric techniques to account for endogeneity, and found potentially large and significant effects of transit capacity upon agglomeration and hence productivity. The main shortcoming of this work to date is the lack of control for various other potentially confounding variables that could influence population growth and employment density and may cause inaccurate coefficient estimates for transit capacity. Our theoretical framework could alternatively be implemented using structural equation modeling to account for both direct and indirect effects, and to provide stronger evidence of a causal relationship from transit capacity to agglomeration and from agglomeration to productivity.

Our second data set consists of firm-level data for two metropolitan regions with large and growing rail transit systems: Portland, Oregon, and Dallas, Texas. Analysis of these data has found interesting differences in how transit investments in the two regions are correlated with the distribution and density of employment by industrial classification. Our current analysis has aggregated firm data to the block-level rather than analyzing specific firm-level data. We will extend this by developing models at the firm level to investigate firm births and deaths, withinfirm growth, and other more detailed phenomena that enable a clearer understanding of dynamics associated with agglomeration near rail stops in the two regions. Although our dataset does not include explicit productivity measures such as revenues or wages, this analysis will enable a ocus on industry structure and the process by which agglomeration can lead to new firm formation, a key indicator of potential increased productivity.

We expect both strands of the research to provide informative results that will greatly increase the value of the work we have already completed. Assessing the costs and benefits of proposed transit infrastructure investments depends on a better understanding of how regional and firmlevel productivity might be affected. The research will also have implications for decisions about funding high-speed rail service.

For the New York metropolitan region, in particular, investigating how regional and firm-level productivity is affected by major transit improvements is a critical missing piece for quantifying the benefits of key links such as the ARC/Gateway project.