Historical Structural Racism in the Built Environment and Contemporary Children’s Opportunities Free

We chose Allegheny County, Pennsylvania, as the context for this work because it is one of the most inequitable regions in the United States, particularly when it comes to race.^40,41  For example, Black residents of Pittsburgh have worse educational, socioeconomic, and health outcomes compared with white residents of Pittsburgh and also compared with Black residents of comparable US cities.⁴⁰  Allegheny County also has a history of industrial pollution and urban development during an era when housing discrimination was ubiquitous across the United States.⁴²  Historical redlining in Allegheny County is linked to current levels of concentrated poverty and home vacancies,⁴³  as well as heightened risk of exposure to pollution and higher rates of asthma.⁴⁴  However, redlining is just one of a host of historical policies and practices that likely diminished contemporary education, socioeconomic, and health opportunities for children living in Allegheny County. This study was deemed exempt by the University of Pittsburgh institutional review board (#2303015).

We operationalized SRBE through census tract-level measures of redlining, blockbusting, freeway displacement, and urban renewal in Allegheny County created in ArcGIS, a geographic information system software developed by the Environmental Systems Research Institute that is used to make maps and analyze geospatial data.

Redlining refers to the Home Owners’ Loan Corporation’s practice in the 1930s to 1960s of rating neighborhoods from most desirable (Grade A = green), still desirable (Grade B = blue), definitely declining (Grade C = yellow), to hazardous (Grade D = red).^28 –31  The Home Owners’ Loan Corporation’s grades were used to justify favorable home lending practices in more affluent, mostly white Grade A and B neighborhoods, while undermining lending in redlined neighborhoods.^10,22,45  This practice made it challenging for Black populations and other racially and ethnically minoritized and/or low-income populations to become homeowners and build equity, which disrupted the transfer of wealth to subsequent generations.^{10,21,22,45 –47}  Redlining also exacerbated inequities in government investments into neighborhoods along dimensions of race and SES.⁴⁵  Our redlining data come from the Mapping Inequality Project.⁴⁸  We created a weighted measure of the percentage of areas within census tracts given C and D grades with the formula “C + 2D” (creating double-weight for D areas).

Blockbusting refers to shifts in neighborhood racial composition because of panic selling induced by real estate companies, resulting in white flight (white populations fleeing from cities to suburbs to maintain racial homogeneity), which caused a decrease in the proportion of the white population in previously majority white areas.^5,35,49,50  Our blockbusting measure comes from historical US Census data and captures the percentage change in the white population in each census tract from 1950 to 1980 (with higher percentage change reflecting areas more likely to have been deliberately blockbusted by real estate companies and agents). We selected 1950 to 1980 because this is when blockbusting affected the largest number of residents, although it did persist beyond 1980.^49,50 

Freeway displacement reflects the supplanted populations and community devastation that accompanied the construction of US highways and interstates, which disproportionately impacted low-income and/or racially and ethnically minoritized populations.^38,39  Our freeway displacement measure captures the percentage of land area within each census tract that is within 100 meters of a freeway. We only include freeways inside the city of Pittsburgh and freeways that are not in a tunnel to capture the areas where communities were displaced (eg, areas that were urbanized as of the date of freeway construction versus communities that emerged because of freeway construction).

Finally, urban renewal refers to local government and, later, private developers using eminent domain to demolish existing neighborhoods and redevelop them, typically displacing economically vulnerable residents.^36,37  Our urban renewal variable measures the percentage of a census tract that falls within an urban renewal zone. Urban renewal zones were coded in ArcGIS according to information on historical urban renewal projects from the 1940s through 1970s in Allegheny County from archival sources including newspapers, libraries, historical societies, and museums.^51 –57 

For all SRBE measures, we used a weighted kernel density analysis with street centerlines as the unit of analysis (from which kernel density was calculated) aggregated up to the census tract-level to capture maximal variability. We then standardized and mean-centered SRBE measures for easier interpretation of results.

We examined broad standardized composite measures of children’s opportunities for positive development across domains of education, SES, and health as of 2015 from the Child Opportunity Index 2.0 (COI).⁵⁸  The COI was created by diversitydatakids.org to measure a comprehensive set of neighborhood and environmental features that influence children’s development at the census tract level. The COI pulls data from a variety of third-party sources including the US Census, Department of Education, Centers for Disease Control and Prevention, Department of Agriculture, and Environmental Protection Agency.

The standardized education composite measure includes items like early childhood education availability and enrollment, percentage of elementary-aged students reaching proficiency on standardized math and reading tests, high school Advanced Placement enrollment and graduate rates, school poverty, and teachers’ experience levels. The standardized SES measure includes items like the rates of poverty, employment, and homeownership, as well as residents’ commuting distances. Finally, the standardized health measure includes items like access to healthy food and green space; walkability; toxic exposures including hazardous waste dump sites, industrial pollutants, and extreme heat; and health insurance coverage. See Noelke et al (2020)⁵⁸  for full descriptions of items within each composite measure and how each measure was constructed.

We conducted spatial regression analyses in Stata 18.0⁵⁹  to examine individual (ie, redlining, blockbusting, freeway displacement, and urban renewal modeled separately) and total (all 4 SRBE measures modeled together) associations between SRBE and children’s opportunities for education, socioeconomic attainment, and healthy development. Models included spatially autoregressive errors to adjust for the spatial autocorrelation of census tracts (as indicated by significant Moran’s tests)⁵⁹  and quadratic terms for redlining and blockbusting to adjust for curvilinear trends (there were no observed curvilinear trends for freeway displacement or urban renewal). Each child opportunity outcome (education, SES, and health) was modeled separately. We examined effect sizes using standardized regression coefficients (β) to understand how an SD change in SRBE related to an SD change in children’s opportunities, and R² values to understand the percentage of variance in each child opportunity outcome explained by SRBE. We also examined influence statistics (DFBETAs) to ensure that our results were not driven by outliers.

According to the 2010 Decennial Census, Allegheny County, Pennsylvania, consisted of 402 census tracts encompassing Pittsburgh and adjacent cities and suburbs. Our analyses included 394 census tracts for which we had complete outcome data. Eight census tracts were excluded either because they did not have land area (ie, tracts that are fully covered by water) or they had missing data on >50% of the items in each COI composite variable.⁵⁸  We used tract designations from the 2010 census to be consistent with the census tract identification numbers used in the 2015 COI (because census tracts are generally updated every 10 years).⁵⁸ Table 1 includes demographic information on Allegheny County (and the United States for comparison) from the COI. Figure 1 and 2 include maps of SRBE and children’s educational opportunities in Allegheny County overall and in Pittsburgh and its immediate surrounding areas (the most population-dense area of the County), respectively. Figure 3 and 4 include maps of SRBE and children’s socioeconomic opportunities in Allegheny County and Pittsburgh, respectively. Finally, Figs 5 and 6 include maps of SRBE and children’s health opportunities in Allegheny County and Pittsburgh, respectively. In Figs 1–6, the lighter green dots reflect fewer or worse opportunities and the darker green dots reflect more or better opportunities.

Table 2 includes the standardized regression coefficients, 95% confidence intervals, P values, and R² values from the spatial regression models where each SRBE measure was modeled separately. Table 2 shows that redlining and blockbusting were significantly negatively associated with children’s opportunities for education, SES, and health (with positive quadratic terms indicating that associations with children’s opportunities slightly diminish at higher levels of these SRBE measures). Urban renewal was significantly negatively associated with children’s education and socioeconomic opportunities. Urban renewal was not associated with children’s opportunities for healthy development. Similarly, freeway displacement was not associated with children’s education, socioeconomic, or health opportunities.

Results of spatial regression analyses examining all 4 SRBE measures modeled together are displayed in Table 3. Table 3 shows that redlining and blockbusting were each significantly negatively associated with children’s opportunities for education, SES, and health even when modeled together and with the other SRBE measures. Additionally, urban renewal was significantly negatively associated with children’s socioeconomic opportunities even when modeled with the other SRBE measures. Modeling all 4 SRBE measures together (Table 3) resulted in R² values that were larger by 0.018, 0.165, and 0.111 for children’s education, socioeconomic, and health opportunities, respectively, compared with the models that only included redlining as an independent variable (Table 2). For example, the R² for the model with only redlining predicting children’s socioeconomic opportunities was 0.308 (Table 2) and the R² for the model with all 4 SRBE measures predicting children’s socioeconomic opportunities was 0.473 (Table 3), resulting in an R² that is larger by 0.165, or 16.5 percentage points.

Consistent with our hypotheses, we found that SRBE indices, with the exception of freeway displacement, were largely associated with fewer educational, socioeconomic, and health opportunities for Allegheny County’s children in 2015. Additionally, the models that included all 4 SRBE indices as independent variables explained more variance in children’s opportunities than the models that only included redlining as an independent variable. Although the effect sizes of associations between SRBE and children’s opportunities are small (between −0.006 and −0.219 SD), the large amount of variance explained (up to 47.4%) by the SRBE indicators is striking. Our findings suggest that historical racist policies and practices of redlining, blockbusting, and urban renewal likely have pervasive and enduring consequences for the neighborhood contexts that shape children’s development up to 85 years later. For example, SRBE may be driving present-day inequities in children’s physical health through factors such as disproportionately worse access to health care, with minoritized families receiving inferior and discriminatory treatment by the health care system, and heightened exposure to toxic stress for minoritized children.¹²  SRBE’s enduring impacts on health are especially important to consider because recent research has found that fewer opportunities for healthy development are associated with greater cardiometabolic risk and increased mortality among children and adolescents.^60,61 

Past work has been limited in scope in considering only the impacts of redlining, but we have demonstrated that blockbusting in particular may be equally, if not more, related to children’s opportunities. Urban renewal also appears to be a significant component of SRBE; however, effect sizes were smaller and more inconsistent. Additionally, our analyses did not find support for freeway displacement as a significant component SRBE related to children’s opportunities. The small effect sizes for urban renewal and insignificant effects for freeway displacement may be because of specific conditions in Allegheny County. For example, most census tracts in Allegheny County do not overlap with former urban renewal sites or places where communities were displaced by freeway construction (see Figs 1, 3, and 5). Thus, a larger county with a history of more urban renewal and freeway displacement may show more robust results when linking these measures to children’s present-day opportunities. Hence, more research on these 2 SRBE measures is warranted.

Furthermore, our 4 dimensions of SRBE demonstrated unique associations with children’s present-day opportunities. Had we combined these measures into 1 composite index, which is a common practice in the structural racism literature,^23 –25  this might have weakened our understanding of each measure’s individual associations. This study thereby lays the foundation for future work investigating pathways through which these 4 historical SRBE measures come to be uniquely associated with present-day inequities in children’s opportunities.

We acknowledge several limitations. Our study does not examine causal associations. For example, we cannot state that blockbusting expressly caused fewer contemporary opportunities for children’s healthy development. Descriptive analyses, however, are key to understanding how historical practices are associated with present-day conditions. Second, our COI outcome data are from 2015, and associations could look different today, especially considering impacts from the coronavirus disease 2019 pandemic.^62 –66  A study linking COI data to pandemic conditions found that children in the lowest-opportunity neighborhoods received significantly less acute health care during the pandemic compared with before.⁶⁵  Thus, if recent events and policy contexts exacerbated the effects of SRBE, links between SRBE and children’s opportunities may be even stronger with more recent outcome data. Alternatively, we might hypothesize that more recent outcome data would yield weaker links between SRBE and children’s opportunities if SRBE’s impacts attenuate over time. Despite these limitations, we believe linking a more comprehensive set of historical structural racism indicators to several facets of children’s opportunities makes an important contribution to the expanding literature on structural racism and child development.

An important future direction of our work is to examine how more recent policies and practices in the built environment (ie, gentrification⁴  and changes to school attendance boundaries⁶⁷ ) may exacerbate or attenuate effects of SRBE on children’s opportunities. Furthermore, although we consider children’s opportunities as our outcome, there is a robust literature using the COI as a pathway through which contemporary forms of structural racism relate to outcomes including physical fitness,⁶⁸  hospitalization rates,⁶⁹  cardiometabolic risk,⁶⁰  exacerbation-prone asthma,⁷⁰  mortality risk,⁶¹  and life expectancy.⁷¹  Thus, examining children’s educational, socioeconomic, and heath opportunities as mediators of links between historical forms of structural racism and other child development outcomes is an important future extension of this study. Additionally, it is important to expand this work to consider additional operationalizations of opportunity, such as the “Opportunity Atlas,” which examines long-term socioeconomic mobility from childhood through adulthood.⁷² 

Our study explicitly tested multiple dimensions of how historically racist actions of government agencies, policies, and practices are related to current opportunities in the neighborhoods children are living in today. In doing so, we demonstrated that redlining, blockbusting, and urban renewal in Allegheny County are largely negatively associated with children’s present-day education, socioeconomic, and health opportunities. This should inform policies for equitable, community-led redevelopment that may improve neighborhood quality and promote opportunities for healthy development for all children, particularly in municipalities like Allegheny County where racism has been declared a public health crisis.^41,73  Overall, historical US policies and practices deliberately promoted disproportionate disinvestment from neighborhoods, and this study contributes valuable information for designing equitable solutions to combat the enduring consequences of structural racism.

COI

Child Opportunity Index

SES

socioeconomic status

SRBE

historical structural racism in the built environment