This CAREER proposal outlines a four-year plan with the primary objective of establishing and refining a research technique for policy evaluation that combines powerful econometrics with institutional knowledge and qualitative fieldwork. This technique is intended to make methodological and substantive contributions to economic knowledge which are directly useful to social scientists and public policy makers: the goal is to uncover the behavioral mechanisms through which policies operate, in order to assess the effectiveness of current policy and to intelligently create new policies. The researcher will use the methods developed to assess the impact of particular policies affecting employment, crime and the general well being of families in poor urban neighborhoods, and also to train future researchers and public policy makers. There is an overwhelming association between residence in a very poor area and negative outcomes for individuals, such as low education attainment. However, the apparent pathology of these urban ghettos could simply be the result of low skills and low motivation of individuals that would not be affected by changes in the surrounding neighborhood. Very little empirical evidence convincingly distinguishes between individual and neighborhood effects, or identifies the social forces within neighborhoods that influence individuals; social scientists do not have the double-blind randomized trials that a biostatistician might use to evaluate the causal effect of a new drug. However, social scientists can use quasi-experiments in which otherwise comparable groups of individuals face different rules or incentives that affect their behavior. The proposed research uses institutional features of housing mobility programs, which select participating families by a type of lottery, to distinguish distinct effects of residential location on well being. Preliminary empirical work will estimate the direction and magnitude of the average effect of moving out of the ghetto to a new neighborhood by comparing program participants to families who would like to have moved if they had been selected to participate. Various outcomes, such as mortality, criminal victimization, health, education, welfare use, and employment, will be measured to assess well-being, and the researcher will concentrate on identifying the social forces that influence those outcomes that appear to be affected by where a family lives. Observing how the programs work, and directly interviewing participants, will help identify the aspects of neighborhoods that exert this influence. For instance, survey work will allow the researcher to ask detailed questions about some types of activities, such as drug use, crime, and sexual activity, that are vitally important but only crudely recorded in administrative data (e.g., arrests, births). The fieldwork will provide a context within which to interpret the average program effects. In addition, the qualitative work will generate specific hypotheses about the mechanisms through which neighborhoods influence outcomes, and suggest implications of these hypotheses that may be tested through further data collection and statistical analysis. Development of econometric techniques tailored to these new projects is a another element of this plan, such as formulation of a technique to measure sentencing disparity between judges that has general applicability as a nonlinear model for correlated random effects with count data. Educational activities will involve teaching and advising graduate students in Economics and Public Administration programs. Students will work with a government agency in New York City to help them evaluate the effectiveness of their anti-crime initiatives in public housing, using the combination of qualitative and quantitative methods developed and refined in the research component of this project.
