By: John Besl
One of the truisms of demography is that there are only three components of population change: birth, death, and migration. We have little choice in the timing of our birth or death, but as adults we typically have a great deal of control in migration decisions, that is, the choice to move from one residence to another. I first became interested in the field of population studies back in the 1980s upon reading stories in the popular press about individuals and families moving from the “Rust Belt” to the “Sun Belt” in the face of plant closings. Northern cities like Detroit and Youngstown, Ohio became symbols of the economic dislocation of that era, losing huge shares of their resident populations. Meanwhile, southern cities like Houston and Atlanta attracted northern migrants and grew rapidly.
The American Community Survey (ACS) includes a questionnaire item on residential mobility, asking where the respondent lived one year ago. Responses are tabulated into the following categories: (1) same house or apartment, (2) moved within the same county, (3) moved from a different county within the same state, (4) moved from a different state, or (5) moved from abroad. Anyone living in a different house or apartment from one year ago can be counted as a mover; the share of movers serves as a measure of residential mobility or its flip side, stability.
All of us probably sense intuitively that small areas like neighborhoods can have widely varying mobility rates. We expect that an area saturated with rental apartments will have high mobility. Other areas are dotted liberally with relatively inexpensive starter homes that are affordable to a wide spectrum of home buyers. Entry and exit in these areas is relatively easy. Map 1 below covers the seven largest counties in the Cincinnati Metropolitan Statistical Area (MSA), displaying the percent share of population, by census tract, that moved within the past 12 months. Keep this caveat in mind - some movers may stay within the same census tract; we can’t infer from these ACS data that all movers are new residents of the tract.
Two clusters in particular should surprise no one since they’re dominated by highly mobile college students attending large state universities: Miami University in Oxford (northwest Butler County) and the University of Cincinnati. Tracts in the Cincinnati neighborhoods of Lower Price Hill, Camp Washington, Roll Hill, Over-The-Rhine, and Downtown also demonstrate high residential mobility, with at least one third of residents having lived in a different house or apartment a year earlier. In general, older urban centers such as Cincinnati, Covington, Newport, Hamilton, and Middletown show higher mobility rates than suburbs. The lowest rates of mobility are seen in areas with more of a rural character on the outer fringes of the region, such as the northern parts of Butler, Warren, and Clermont counties and the southern reaches of the three Kentucky counties. FYI –the unshaded tract in Boone County is the Greater Cincinnati-Northern Kentucky Airport, which has no resident population and thus no mobility rate.
Map 1. Current residents who moved within the past year, by census tract
Wanting to zero in on a few specific areas, I chose the five Place Matters neighborhoods. Place Matters is a place-based model for community investment in which local funders and neighborhood nonprofit organizations work collaboratively to revitalize distressed neighborhoods and improve quality-of-life for residents. In one recent success spurred in part by the Place Matters initiative, Avondale was chosen two years ago for a $29.5 million Choice Neighborhoods Implementation grant from the U.S. Department of Housing and Urban Development to renovate five multifamily structures into mixed-income housing. Figure 1 below shows mobility rates for the five Place Matters neighborhoods along with two middle-to-upper-income comparison areas: Hebron in Boone County and Montgomery in Hamilton County. The thin line straddling the tip of each bar represents the 90-percent confidence interval around each estimate. Since the confidence intervals are overlapping, we can’t conclude that there’s any real difference among the five Place Matters neighborhoods, but the ACS results certainly indicate high levels of mobility in these disadvantaged areas. Montgomery and Hebron, on the other hand, are much more stable with fewer than one in 10 residents having moved in the past year.
Figure 1. Percent of current residents ages 1 and over who moved within the past year, selected places
My employer, Cincinnati Children’s Hospital Medical Center, is actively engaged with local public health departments and a range of community partners to improve child health in our region. Several improvement activities are place-based (e.g., in Avondale and Price Hill), and high mobility in the child population could pose challenges for this work. Map 2 below displays child mobility rates by census tract for the seven-county region. At least one third of children had moved within the previous 12 months in 31 Hamilton County tracts, mostly in Cincinnati. Nine Butler County tracts surpassed the same 33.3% threshold, including one in West Chester, an upper-income suburb. Across the three northern Kentucky counties, six tracts covering parts of Florence, Elsmere, Covington, and Newport also witnessed child mobility rates of one third or higher. Note that the additional unshaded tract, in western Warren County, is the Lebanon Correctional Institution, which has no child residents.
Map 2. Child residents (ages 1 to 17) who moved within the past year, by census tract
Child mobility rates for the Place Matters neighborhoods are presented in Figure 2. Mobility is higher in the child population than the general population in Price Hill, Avondale, and Walnut Hills, where approximately one third of children had moved in the prior 12 months. In Covington and Madisonville, proportionally fewer children, one in five, were movers. In contrast, only one in 20 child residents of Montgomery or Hebron had lived in a different house 12 months earlier.
Figure 2. Percent of current residents ages 1 to 17 who moved within the past year, selected places
Sharp-eyed readers might notice that the 90-percent confidence intervals in Figure 2 are wider than the comparable intervals in Figure 1, ranging from one percentage point wider for Madisonville to seven points for Walnut Hills. This is attributable to the narrower age band (1 to 17) in this chart compared to all ages 1 and over referenced in Figure 1. Recall that these are estimates derived from a sample survey, and there are inevitably fewer sample cases in the 1-to-17 age range. For further illustration of this point, see Figure 3 below, which further limits the population of interest to children ages 1 to 4. When dealing with a very narrowly defined sub-population (e.g., ages 1-4) and small areas like neighborhoods, ACS estimates become almost unusable, as evidenced in Figure 3. Confidence intervals overlap for all areas, including Montgomery, where a 3% estimate of movers in the preschool population is not statistically different, at the 90% confidence level, from Hebron’s 12% estimate or Walnut Hills’ 25% estimate. The lower limits of confidence intervals surrounding the Montgomery and Hebron point estimates extend to zero, while the Walnut Hills estimate has a range of 7% to 44%. Since ACS is the only source for small-area data on residential mobility as well as a range of other topics, I stop short of saying that the estimates are unusable. My point about the width of confidence intervals is raised merely to warn data users to proceed with caution when using ACS small-area estimates, especially for restricted sub-populations.
Figure 3. Percent of current residents ages 1 to 4 who moved within the past year, selected places
High mobility among children is a challenge not only for K-12 educators, but potentially for anyone designing community improvement projects using a place-based model.