In Chicago there are 1,492 separate local government jurisdictions, including 366 school districts. In Miami there are only 36 jurisdictions, and 2 school districts.
The fragmentation of local government has real political consequences: smaller districts can compete for the advantaged by offering tax breaks. More school systems reduce the opportunity to mix students across racial and class lines. Public transportation and job opportunities stop at the borders of wealthy neighborhoods, exacerbating the spatial mismatch between where the poor live and where good jobs are to be found. How does it affect health for different groups?
In an article in the Milbank Quarterly, Malo Hutson and coauthors examine the association between metropolitan fragmentation, measured by the number of local government jurisdictions, special-service districts, and school districts in a metro area, and the all-cause mortality rate in the corresponding counties stratified by age, race, and sex. As covariates they include size of the metro, population density, proportion black, age distribution, and the poverty rate. Although they display a measure of racial segregation in their descriptive tables they don’t include it as a control in their regressions, perhaps because that variable is downstream on the causal pathway to metropolitan fragmentation.
The key study figures with age-adjusted mortality are reproduced below:
On average, more fragmented metro areas have higher black mortality rates, but white mortality is invariant to metro areas. The age band where the disparity was widest for fragmented districts was in middle adulthood: ages 25 to 44. The only age band where the disparity was not impacted by fragmentation was for older adults. The study authors argue that this is a finding that would be consistent with literature that black elders have more access to health services because of Medicare, and may also be positively selected (leading to “cross-over” effects).
This is an important study, but I imagine that it will receive mixed reactions from epidemiologists and economists. The authors admirably lay out a conceptual story tying metropolitan fragmentation to mortality disparities, and their measure is plausible. However, all ecological studies using cross-sectional data are vulnerable to severe confounding from omitted variables. We have touched upon this theme before, for example, with Wilkinson and Pickett’s cross-country inferences in the “Spirit Level.” The main omitted variables that concern me are the structure of the local economy (e.g., manufacturing versus services) and the spatial distribution of environmental toxins.
American cities are not randomly assigned fragmentation, but rather such fragmentation arises from social, political, and economic polarization. Still, the universe is not entirely deterministic. Through odd quirks of American administrative law, different areas have different policies about metropolitan incorporation (sprawling cities in the American southwest have greater annexation powers, for example). I imagine one extension of this kind of research will be to identify some plausible policies that cause fragmentation. This is not just a problem of causal inference. As public health officials have come to understand how zoning and housing laws make people’s health worse, they have become more sophisticated advocates for reforming those laws. Perhaps health will improve if cities in the future re-consolidate authorities.