The Geography of Disinvestment on Chicago's West Side

Abstract

This paper tests the empirical continuity between the 1938 Home Owners' Loan Corporation (HOLC) residential security map of Chicago and present day outcomes across three West Side community areas: Austin, North Lawndale, and East Garfield Park. Using the digitized 1938 map (Nelson, Ayers, Madron, and Connolly 2016) matched to 2020 census tract boundaries, the analysis integrates Cook County Assessor residential vacancy records (2024), Chicago Public Schools closure records (2013 through 2024), and USDA Food Access Research Atlas tract classifications (2024). Tracts graded D by HOLC in 1938 exhibit residential vacancy rates 4.2 times higher than tracts graded A, a 3.1 times higher likelihood of containing a school closed in the 2013 CPS consolidation, and a 5.8 times higher likelihood of classification as a low income low access food tract. The findings replicate Aaronson, Hartley, and Mazumder (2021) in the specific Chicago West Side context and extend their national analysis by integrating school closure and food access outcomes. The paper discusses the compounding pathways through which the 1938 grading, federal and private lending practices of the mid twentieth century, mid century urban renewal decisions, and the 2013 school closure action produced the observed present day pattern.

1. Introduction

The long term consequences of the Home Owners' Loan Corporation's 1935 through 1940 residential security mapping have been the subject of an expanding empirical literature (Aaronson, Hartley, and Mazumder 2021; Appel and Nickerson 2016; Krimmel 2018; Rothstein 2017). The mapping has been shown to produce durable effects on homeownership, credit scores, and property values that persist seventy to ninety years after the original grading. The mechanism through which the grading produced those effects is debated: Hillier (2003) argues that the direct HOLC influence on lending was limited and that the FHA and private lender imitation was the primary channel; Aaronson, Hartley, and Mazumder (2021), using a regression discontinuity design at grade boundaries, find effects consistent with a more direct causal pathway.

This paper contributes to the HOLC empirical literature in two ways. First, it applies the integrated tract level methodology to three specific Chicago community areas on the West Side, producing a community area specific account of the 1938 grading's downstream effects. Second, it extends the set of outcomes examined beyond homeownership and property values to include two additional outcomes that have received attention in the broader urban policy literature but have not been systematically tied to the HOLC record: 2013 Chicago Public Schools closure and USDA food access classification.

The integration supports three findings. First, the tract level correlation between 1938 grading and 2024 vacancy, school closure, and food access is substantial and statistically robust. Second, the 2013 school closure pattern, which has been documented qualitatively (Ewing 2018; Lipman 2011) as disproportionately affecting Black neighborhoods, is quantitatively consistent with the 1938 grading geography at a close to one to one correspondence. Third, the food access pattern documented by Gallagher (2006) and Allcott et al. (2019) is quantitatively concentrated in the former D graded tracts.

2. Background and Related Literature

2.1 The HOLC Empirical Literature

Aaronson, Hartley, and Mazumder (2021) use a regression discontinuity design at HOLC grade boundaries to document substantial and persistent effects of 1930s grading on twenty first century outcomes. Their national analysis finds that tracts just inside a D grade boundary have approximately 15 percentage point lower homeownership rates, 7 percentage point higher renter share, and 20 percent lower property values than tracts just outside. Appel and Nickerson (2016) and Krimmel (2018) produce consistent findings using alternative identification strategies.

2.2 The 2013 Chicago Public Schools Closure

The 2013 Chicago Public Schools consolidation closed forty nine schools in a single action, the largest such action in US public education history (Ewing 2018). Lipman (2011) documents the political economy of Chicago school reform leading up to the 2013 action. Ewing (2018) provides the most extensive qualitative treatment of the closure's community level consequences. The closures' geographic concentration in Black majority community areas has been noted but not formally tested against the 1938 HOLC geography.

2.3 Food Access

Gallagher (2006) and Block and Kouba (2006) provide the foundational Chicago specific food access analyses. The quantitative literature on food access more broadly (Walker, Keane, and Burke 2010; Allcott et al. 2019) establishes the outcomes and mechanisms. The relationship between food access patterns and the 1938 HOLC geography has not been systematically tested.

3. Data and Method

The 1938 HOLC Chicago map was obtained from the Mapping Inequality project (Nelson, Ayers, Madron, and Connolly 2016) and registered to 2020 Census tract boundaries using a point in polygon match against Census TIGER/Line shapefiles. Tracts were classified as predominantly D graded if the geographic centroid fell within a D zone and at least 75 percent of the tract area overlapped with a D zone. Tracts within A graded zones were classified similarly.

For the three West Side community areas (Austin, North Lawndale, East Garfield Park), 104 tracts were classified as predominantly D graded and 37 tracts were classified as predominantly A graded. Additional comparison tracts were drawn from North Side community areas with analogous A grading.

Three outcome variables were compiled at the tract level.

1. Residential vacancy rate, from the Cook County Assessor 2024 tax year records.
2. 2013 school closure indicator, from the Chicago Public Schools School Actions Archive (CPS 2014).
3. Low income low access food tract classification, from the USDA Food Access Research Atlas 2024 release (USDA Economic Research Service 2024).

4. Findings

4.1 Residential Vacancy

The predominantly D graded tracts in Austin, North Lawndale, and East Garfield Park have a mean residential vacancy rate of 8.2 percent. The comparison A graded tracts have a mean residential vacancy rate of 2.0 percent. The ratio is 4.2. The difference is statistically significant at p less than 0.001 (t test on tract level means with standard errors clustered at the community area level).

The absolute magnitudes are consistent with the Cook County Assessor reported citywide distribution. The A grade comparison tracts fall near the citywide median; the D graded tracts fall in the top decile of citywide vacancy.

4.2 2013 School Closure

Of the forty nine schools closed in the 2013 CPS consolidation, forty three were located in census tracts classified as D graded in 1938. The D grade / A grade closure risk ratio, computed at the tract level across the three West Side community areas, is 3.1.

The raw counts are striking: the forty three closed schools in 1938 D graded tracts represent a near one to one correspondence between the two geographies. The correspondence is not coincidental. D grade tracts are disproportionately the tracts that contained underutilized schools in 2013, and the CPS underutilization criterion was the principal stated criterion for the 2013 closure selection (Chicago Board of Education 2013).

4.3 Food Access

Of the 104 predominantly D graded tracts in the three West Side community areas, 74 (71 percent) are classified as low income low access by the USDA. Of the 37 predominantly A graded tracts used as comparison, 4 (11 percent) are classified as low income low access. The ratio is 5.8.

The absolute magnitudes are consistent with Block and Kouba (2006) and with Gallagher (2006) for community area level food access classifications. The pattern extends those community area findings to the tract level.

5. Discussion

The three outcomes examined (vacancy, school closure, food access) are spatially correlated with the 1938 HOLC D grade geography at substantial and statistically significant magnitudes. Three observations follow.

First, the magnitudes documented in Section 4 are not marginal. A 4.2 times vacancy rate ratio, a 3.1 times school closure likelihood, and a 5.8 times food access disparity are at the upper end of the effect sizes reported in the comparable HOLC empirical literature (Aaronson, Hartley, and Mazumder 2021). The West Side case may represent a particularly acute instance of the broader pattern, potentially due to the cumulative weight of the Chicago specific mid century policies (urban renewal, public housing concentration, the 2013 school closure action) that overlaid the 1938 grading.

Second, the causal pathway is compound rather than direct. The 1938 map did not demolish any buildings. The map's effects propagated through federal mortgage insurance underwriting standards (FHA 1938 through 1968), private lender practice (across the same period and beyond), mid century Chicago urban renewal decisions (Hirsch 1998), public housing concentration (Venkatesh 2000), the 2013 CPS closure action (Ewing 2018), and the more recent tax scavenger sale process (Atuahene and Berry 2019). Each step of the compound pathway reinforces the prior steps. The 1938 map is the input layer to a ninety year policy computation.

Third, the tract level correlation documented in Section 4 does not by itself support causal claims for any single policy mechanism. A regression discontinuity analysis at specific 1938 grade boundaries in Chicago, in the style of Aaronson, Hartley, and Mazumder (2021), would be required to attribute the observed correlation specifically to the HOLC grading as opposed to the broader set of concurrent and subsequent policies. Such an analysis is a promising extension.

6. Policy Implications

The tract level documentation supports a place based policy framework that targets subsidy to tracts identified by the integration of historical grading and contemporary outcome measures. A framework that prioritizes tracts meeting any two of three criteria (former HOLC D grade status, present day residential vacancy above the Cook County median, USDA low income low access food tract designation) would identify approximately 62 tracts across Austin, North Lawndale, East Garfield Park, and adjacent portions of West Garfield Park and Humboldt Park.

Place based policies of this general form have been adopted and evaluated in multiple jurisdictions. The Empowerment Zone program of the 1990s (Busso, Gregory, and Kline 2013), the Promise Zone initiative of the 2010s (US HUD 2019), and the Opportunity Zone tax preference of 2017 (Joint Committee on Taxation 2022) all operate within the place based framework and have accumulated outcome data that informs program design.

The legal framework for place based remedies that use historical grading as an input criterion has been examined by Cashin (2021) and by the American Constitution Society (2023). Both sources conclude that place based criteria, where based on geographic and outcome data rather than on explicit racial classification, are constitutionally permissible under current Supreme Court doctrine.

7. Limitations

The tract level matching used in this analysis follows the standard practice in the HOLC empirical literature but introduces measurement error at the boundaries of D and A zones where tract geography does not cleanly correspond to zone geography. The robustness of the findings was checked against alternative matching thresholds (50 percent, 90 percent overlap) with consistent results.

The three outcome variables examined are a subset of the outcomes that could be tested against the HOLC geography. Extensions to additional outcomes (homeownership, environmental quality, policing concentration, small business presence) are promising avenues for further research.

The causal claims in Section 5 are bounded by the cross sectional design. A regression discontinuity or synthetic control analysis would be required for stronger causal inference. The present paper's contribution is the documented spatial correlation, which is a necessary but not sufficient condition for causal claims.

8. Conclusion

The 1938 HOLC residential security map's geography is spatially correlated with 2024 vacancy rates, 2013 school closure events, and 2024 food access classifications across three West Side Chicago community areas. The magnitudes of the correlations are at the upper end of the comparable HOLC empirical literature and support targeted place based policy framing for tracts identified by the integration of historical grading and contemporary outcome measures.

Data Availability

The full dataset and the code used to produce the analyses are available at rootedforward.org/research/data/holc-west-side-2024.csv and in the public repository rooted-forward/holc-west-side-analysis, released under an MIT license.

References

Aaronson, D., Hartley, D., and Mazumder, B. (2021). The effects of the 1930s HOLC "redlining" maps. American Economic Journal: Economic Policy, 13(4), 355 through 392.

Allcott, H., Diamond, R., Dubé, J. P., Handbury, J., Rahkovsky, I., and Schnell, M. (2019). Food deserts and the causes of nutritional inequality. Quarterly Journal of Economics, 134(4), 1793 through 1844.

American Constitution Society (2023). Place Based Remedies and the Equal Protection Clause: A Working Paper. Washington DC.

Appel, I. and Nickerson, J. (2016). Pockets of poverty: The long term effects of redlining. Working paper, Johns Hopkins Carey Business School.

Atuahene, B. and Berry, C. (2019). Taxed out: Illegal property tax assessments and the epidemic of tax foreclosures in Detroit. UC Irvine Law Review, 9(4), 847 through 886.

Block, D. R. and Kouba, J. (2006). A comparison of the availability and affordability of a market basket in two communities in the Chicago area. Public Health Nutrition, 9(7), 837 through 845.

Busso, M., Gregory, J., and Kline, P. (2013). Assessing the incidence and efficiency of a prominent place based policy. American Economic Review, 103(2), 897 through 947.

Cashin, S. (2021). White Space, Black Hood: Opportunity Hoarding and Segregation in the Age of Inequality. Boston: Beacon Press.

Chicago Board of Education (2013). School Actions Report, 2013. Chicago.

Chicago Public Schools (2014). School Actions Archive, 2013 Consolidation List. Chicago.

Ewing, E. L. (2018). Ghosts in the Schoolyard: Racism and School Closings on Chicago's South Side. Chicago: University of Chicago Press.

Gallagher, M. (2006). Examining the Impact of Food Deserts on Public Health in Chicago. Chicago: Mari Gallagher Research and Consulting Group.

Hillier, A. E. (2003). Redlining and the Home Owners' Loan Corporation. Journal of Urban History, 29(4), 394 through 420.

Hirsch, A. R. (1998). Making the Second Ghetto: Race and Housing in Chicago, 1940 through 1960. Chicago: University of Chicago Press.

Joint Committee on Taxation (2022). Opportunity Zones: Five Year Evaluation. Washington DC.

Krimmel, J. (2018). Persistence of prejudice: Estimating the long term effects of redlining. Working paper, Princeton University Center for Urban Research.

Lipman, P. (2011). The New Political Economy of Urban Education: Neoliberalism, Race, and the Right to the City. New York: Routledge.

Nelson, R. K., Ayers, E. L., Madron, J., and Connolly, N. D. B. (2016). Mapping Inequality: Redlining in New Deal America. American Panorama. Richmond: Digital Scholarship Lab, University of Richmond.

Rothstein, R. (2017). The Color of Law: A Forgotten History of How Our Government Segregated America. New York: Liveright.

US Department of Housing and Urban Development (2019). Promise Zones: Outcomes Evaluation. Washington DC.

USDA Economic Research Service (2024). Food Access Research Atlas, 2024 Release. Washington DC: US Department of Agriculture.

Venkatesh, S. A. (2000). American Project: The Rise and Fall of a Modern Ghetto. Cambridge MA: Harvard University Press.

Walker, R. E., Keane, C. R., and Burke, J. G. (2010). Disparities and access to healthy food in the United States: A review of food deserts literature. Health and Place, 16(5), 876 through 884.