Where Did Redlining Matter? Regional Heterogeneity and the Uneven Distribution of Advantage

Abstract

This article analyzes the regional variation in outcomes of a seemingly standardized federal neighborhood valuation principle used in home mortgage insurance grading. The objective is to highlight the contingent discriminatory and economic conditions that mediated heterogeneous housing outcomes across different parts of the United States. How did city and regional economic and demographic growth patterns vary before and during the mortgage insurance program implemented through the Federal Housing Administration (FHA)? How might this have shaped loan guarantee patterns? How does preexisting racial housing discrimination relate to outcomes? Adopting an orientation that centers on Whiteness and the benefits of mortgage finance for certain groups and neighborhoods, this analysis uses a Bayesian hierarchical framework to investigate the degree of the FHA’s influence between 1940 and 1970, here proxied by the Home Owners’ Loan Corporation maps, on A or B (“AB”) graded neighborhoods versus C graded neighborhoods in different cities. This article studies how home values and homeownership change over time and whether there are regional variations in the influence of these grades. It also studies what longitudinal socioeconomic patterns might explain the persistence or decline of the AB effect over time. Findings show cities in the West Coast, Southwest, and northern central United States that saw the most housing construction also had the highest proportions of FHA loans to overall dwelling units. There is also a distinctive consistency and persistence of benefit on home value and homeownership to AB graded neighborhoods in these cities, possibly owing to regional shifts in the industrial landscape.

针对住房抵押贷款保险评级中的联邦标准化社区估价原则, 本文分析了其效果的区域差异。旨在强调导致美国不同区域的不同住房效果的歧视性和经济性附加条件。在联邦住房管理局实施抵押贷款保险计划之前和期间, 城市和区域的经济和人口增长模式有何不同？它如何塑造贷款担保模式？以前的种族化住房歧视与住房效果的关系如何？本文以白人为关注点, 以某些群体和社区的抵押贷款融资收益为核心, 通过房主贷款公司地图和贝叶斯层次框架, 研究了1940年至1970年联邦住房管理局对不同城市的A级或B级社区(AB)与C级社区的影响程度。研究了房屋价值和住房所有权随时间的变化, 以及社区等级影响的区域差异。探讨了哪些纵向社会经济模式能解释AB效应随时间的持续或降低。结果显示, 美国西海岸、西南部和中北部拥有住房最多的城市, 联邦住房管理局贷款的住房比例最高。可能是由于工业格局的区域变化, 这些城市AB等级社区的住房价值和所有权有显著的一致性和持续性。

Este artículo analiza la variación regional en los resultados de aplicar un principio federal de evaluación de vecindarios aparentemente estandarizado, que se usa en la calificación del seguro hipotecario de la vivienda. Lo que se busca es destacar las condiciones discriminatorias y económicas contingentes que mediaron en los resultados heterogéneos en materia de vivienda en diferentes partes de los Estados Unidos. ¿Cómo variaron los patrones de crecimiento económico y demográfico de ciudades y regiones antes y durante el programa del seguro hipotecario que se implementó a través de la Administración Federal de la Vivienda (FHA)? ¿Cómo podría esto haber configurado los patrones de garantía de los préstamos? ¿Cómo puede estar relacionada la discriminación racial preexistente en materia de vivienda con los resultados? Adoptando una orientación centrada en la blancura y los beneficios de la financiación hipotecaria para ciertos grupos y vecindarios, este análisis usa un marco jerárquico bayesiano para investigar el grado de influencia de la FHA, entre 1940 y 1970, aquí formulado en los mapas de la Home Owners’ Loan Corporation, en los vecindarios calificados como A o B (“AB”), versus los calificados como C en diferentes ciudades. Este artículo estudia el modo como cambian con el tiempo los valores y la propiedad de las viviendas, y si se presentan variaciones regionales en la influencia ejercida por estos grados. También estudia qué patrones socioeconómicos longitudinales podrían explicar la persistencia o la declinación del efecto AB a través del tiempo. Los hallazgos muestran ciudades en la Costa Oeste, el Sudoeste y en el centro norte de Estados Unidos donde se experimentó mayor construcción de viviendas y que tenían también las proporciones más altas de préstamos FHA respecto al total de las unidades de vivienda. Hay también consistencia y persistencia distintivas del beneficio sobre el valor de la vivienda y la propiedad de la vivienda, para los vecindarios con calificación AB en estas ciudades, posiblemente por los cambios regionales en el paisaje industrial.

Key Words:

• Bayesian methods
• historical redlining
• housing discrimination
• mortgage finance

关键词:

• 贝叶斯方法
• 历史红线
• 住房歧视
• 抵押贷款融资。

Palabras clave:

• demarcación excluyente histórica
• discriminación por vivienda
• financiación hipotecaria
• métodos bayesianos

Notes

1 The exact geographic coverage of the Sunbelt has been variously debated (Abbott 1988; Bernard and Rice 2014), but is generally understood as those states (or portions of states, as in the case of California) that fall southward of the 37th parallel and are characterized by faster economic and population growth due to defense spending, other federal outlays such as FHA insurance and federal road infrastructure financing, a favorable business climate, and an appealing quality of life (Bernard and Rice 2014).

2 It should also be noted that patterns are not expected to align entirely along regional patterns, as local influences, including the importance of particular cities in World War II–related manufacturing, the variation of local implementation practices and compliance (U.S. Commission on Civil Rights 1959; Biles 2011), and the role and influence of the local real estate industry on governmental mortgage insurance at all scales of jurisdictions (Weiss 2002; Winling and Michney 2021), all could create more idiosyncratic effects. Describing the attempt by the National Association of Real Estate Brokers to implement standardized national practices, Glotzer (2020) described realtors as making “daily, often inconsistent decisions about exclusions based on local conditions” (3).

3 This analysis uses the default No-U-Turn Sampler (NUTS; Hoffman and Gelman 2011) extension of the Hamiltonian Monte Carlo algorithm with an adaptive diagonal mass matrix based on the variance of the tuning samples as an initialization method. This is the default sampling method in the PyMC library. The 95 percent credible intervals are used in evaluating parameters in the posterior distribution. Credible intervals here represent a 95 percent probability that the true parameter falls within these bounds. As shown in Figures A.1 and A.2, almost all of these credit intervals are all positive and do not include zero, the prior.

4 All trace plots for estimated parameters converged and are not shown in this article.

5 The exceptions are New York City; Toledo, Ohio; and Richmond, Virginia, by 1970.

Additional information

Notes on contributors

Wenfei Xu

WENFEI XU is an Assistant Professor in the Department of City and Regional Planning, Cornell University, Ithaca, NY 14850. E-mail: [email protected]. Her research questions how housing policies, practices, institutions, and technologies have shaped urban inequality, with an orientation toward methods in urban analytics.