Intergenerational Neighborhood-Type Mobility: Examining Differences between Blacks and Whites

Abstract

Using sibling data from the Panel Study of Income Dynamics linked with US Census data, this paper examines the intergenerational nature of neighborhood quality. It is hypothesized that the quality of where one resides as an adult is a function of one's childhood neighborhood through the conditioning and constraining of adult residential choice. Further, it is posited that this relationship varies by race and is stronger for those living in the most disadvantaged neighborhoods, especially blacks. Descriptively, the study finds that childhood neighborhood conditions of black and white children are vastly different. Few whites live in the most disadvantaged neighborhoods, and few blacks live in the most advantaged neighborhoods. The sibling fixed effect regression results confirm the hypothesis that childhood neighborhood disadvantage has negative effects on adult neighborhood quality for those living in the lowest quality, race-specific neighborhoods.

Keywords:

• Neighborhoods
• poverty
• mobility
• intergenerational
• epidemic theory

Acknowledgements

A version of this paper was presented at the Association for Public Policy Analysis and Management meetings in November 2005.

Notes

¹ Cutoff points have been set such that the spline function includes different linear segments for the top 10 per cent of neighborhoods, the 11th to the 25th percentiles, the 26th to the 50th percentiles, the 51st to 75th percentiles, the 76th to the 90th percentiles, and at the 91^st to 100th percentiles. There would be evidence for the epidemic theory if the poorest neighborhoods (i.e. 91st to 100th percentile) were to have the most detrimental effects on future neighborhood conditions. In estimating the spline specification, each section of the spline function is entered into the model separately. In this specification, the overall effect of the neighborhood index–that is, the overall contribution of childhood neighborhood conditions to adult neighborhood conditions–is cumulative, in that it depends on the coefficients associated with each of the previous pieces of the spline function. However, the marginal effect of a change in childhood neighborhood conditions depends only on the coefficient associated with the part of the spline function in which a particular sample member is located. For example, if the coefficient on the piece of the spline function representing the 91st to 100^th percentile is negative, then going from the 91^st to the 95th percentile will have a negative effect on adult neighborhood outcomes, regardless of the coefficient on the segment of the spline function representing the 76th to 90th percentile.

² The neighborhood index takes on a roughly equal number of positive and negative values across the full sample. However, among white sample members the index is negative for most sample members, indicating better neighborhood conditions. Conversely, most black samples have positive values of the neighborhood index, indicating more disadvantaged conditions.

³ Childhood neighborhood values are exchanged by percentile. For example, if the childhood neighborhood index value at the 50th percentile is − 2 for whites and +1 for blacks, white sample members at that percentile would be given a childhood neighborhood value of +1 and black sample members a value of − 2, keeping all other characteristics the same.

⁴ The purpose of this simulation is to provide a baseline from where it is possible to compare the results of the simulation in which black and white childhood neighborhood conditions are exchanged. If the baseline simulations worked perfectly, it would be found that 5 per cent of white sample members were simulated to live in the top 5 per cent of neighborhoods (using the race-specific neighborhood index) as adults. However, the actual baseline simulation results do not quite achieve this goal, particularly at the extremes of the neighborhood distribution. These discrepancies between the actual baseline simulation value and its expected value arise because of differences between the true value of a given sample member's error term and its simulated value. The simulated value is based on a random draw from a normal distribution, and could be different from the actual value either because of the random nature of its selection or from the assumption that the error term has a normal distribution across sample members.

⁵ An important caveat in interpreting this last set of simulations is that they are based on out-of-sample inferences. In particular, because a large proportion of black children live in neighborhoods worse than almost all of the neighborhoods in which white sample members live, there is no direct empirical evidence as to how white children would react to living in those very bad neighborhoods. Instead, this effect is inferred based on the relationship between childhood and adult neighborhood conditions that are observed.