The Concentration of Urban Poverty

National Bureau Of Economic Research Inc, Cambridge MA

Investigators

Abstract

What determines the extent of concentration in urban areas and why have those areas been so attractive to minorities? This project pursues research aimed at improving our understanding of the roots of urban agglomeration and the concentration of poverty within cities. First, it examines the determinants of industrial agglomeration within the United States, which should ultimately help us to understand why cities exist. Urban policy needs to recognize the forces that attract both industry and the poor to urban areas. Second, it studies the causes and consequences of segregation by ethnicity within cities. Third, it determines why there is more redistribution to the poor in big cities. As this redistribution may attract the poor to large cities, it may play a role in creating pockets of urban poverty. Theories of urban concentration emphasize the role of cities in eliminating transport costs for goods, people and ideas. This project quantifies the relative importance of these forces using the Census Bureau's Longitudinal Research Database. In particular, it looks at the determinants of new plant birth and employment levels. It explains the level of new plant growth (and the steady-state level of employment) in a location-industry cell, as a function of the presence of other industries in that location. In particular, the project constructs industry-location specific variables indicating the presence of industries that produce inputs, consume outputs, share ideas and hire the same types of workers. Preliminary work suggests that the most important variable in explaining the location of industries is that the presence of other industries that use the same labor force. To address the endogeneity of the explanatory variables, the interaction of state and industry characteristics are used to predict the level of employment that is used in constructing the explanatory variables. The project examines the formation and effects of immigrant ghettos. It builds a database on ethnic segregation for a large set of ethnicities and cities dating back to 1910 (and back to 1890 for the foreign born as a whole). This data can be used as a dependent variable to test different theories explaining the extent of immigrant segregation. The data may also help us to understand whether the negative connection between segregation and outcomes for African-Americans in 1990 was true in earlier time periods among different ethnicities. A particular focus of this research proposal is to use within metropolitan area government fragmentation, the location of employment for natives and other variables as sources of identification. Preliminary research suggests that ghettos are not bad for their first generation of inhabitants, but that among later generations there is a much stronger correlation between segregation and negative outcomes for later generations (both for blacks and for immigrants). If segregation only does harm over time, then this fact would support the idea that ghettos are harmful because they discourage human capital accumulation among their younger residents. The third section of this proposal examines why there is more redistribution in larger cities. The project considers three hypotheses: First, cities have more renters (because of the nature of the housing stock). Second, cities' fixed resources lessen the tendency of the rich to flee high taxes. Third, greater mixing between rich and poor in cities may leads to more altruism or to more of a desire to limit crime through redistribution. The research focus is on finding exogenous sources of variation that determine the level of homeownership, mixing between rich and poor and fixed resources in the city.