This purpose of this research is to increase our understanding of the economics of the inner city. The common theme of this project is that the poverty of the urban underclass is intimately connected to, and even a result of, a dearth of positive social interactions and role models. This project has three sections. The first primarily involves data collection. The second presents a concrete attempt to answer a crucial question in America's cities: Why is there more crime in big cities? The third is primarily theoretical and methodological . The first section will help to explain why ghettos are so pervasive, despite their many problems. The primary goal is to build a century long data set on segregation by city, and when possible by SMSA, for ethnic and racial groups. Comprehensive data on ghettos is itself a serious contribution, but this data will also help determine why ghettos came about and why they persist so much longer for African-Americans than for other immigrant groups. Micro-evidence will be developed as to why ghettos are initially appealing, but end up being costly for their settlers' descendants. The second section will examine why crime is so much higher in big cities than in small cities or in the hinterland. Evidence gathered so far ensures that the city effect increasing crime is real and not merely a statistical artifact coming from more crime prone individuals living in cities. The urban crime effect is also not the result of the poor being close to the rich in crowded cities. Less than 10% of the effect comes from lower probabilities of arrest in big cities. Approximately 20% comes from higher returns to crime in cities. Of the remaining 70%, 45% can be connected to variables describing family structure in large cities. This work also documents urban anonymity. The last section addresses the empirical puzzle of the strikingly large range of urban crime rates-- the variance of actual crime rates is much larger than the varia nce of crime rates that would be predicted based on the differences in other attributes across cities. This high variance is incompatible with individuals making decisions about crime independently: there must be social interactions. A formal model connecting social interactions with the variance of urban crime rates is used to create a natural index of the degree of social interactions. This index shows that murder and rape have few interactions, but that larceny and auto theft are highly interactive. The research will use the time patterns in urban crime rates, drug use and migration to further examine the size and nature of social interactions.
