Proximity and coresidence of adult children and their parents in the United States : description and correlates

In this paper we provide an overview of the patterns of intergenerational proximity and coresidence of adult children and their mothers in the U.S., using data from the National Survey of Families and Households (NSFH) and the U.S. Census. We highlight the importance of three specification and sample choices in the analysis. First, most previous studies consider coresidence to be the limiting case of proximity, using Tobit, ordered logit, or ordered probit specifications. We argue that proximity and coresidence are qualitatively different, and show that the multinomial logit provides a better representation of the patterns in the data. Second, we argue that substantial differences in the correlates of proximity by gender and marital status indicate the importance of modeling these categories separately. Third, the NSFH allows us to consider the proximity of couples to both his mother and her mother. This information is rarely available in survey data but is important for complete analyses. Our results show that education and age are the most robust predictors of proximity: college graduates are less likely to live near their mothers and older children live further from their mothers. Other demographic variables such as race, ethnicity and only child status also affect the probability of close proximity and coresidence. However, characteristics indicating adult children's current need for transfers (e.g. grandchildren) are not correlated with either close proximity or coresidence, while characteristics indicating mothers' current needs for transfers (e.g., disability) are correlated with coresidence but not close proximity.