Do Land Use Plans Affirmatively Further Fair Housing?

Abstract

Problem, research strategy, and findings

The 1968 Fair Housing Act required local government recipients of federal money to take meaningful actions to affirmatively further fair housing (AFFH). Current fair housing analysis requirements are copious but do not request an assessment of how land use policies affect the potential for neighborhood integration. A recent California law requires local governments to include AFFH analysis in existing planning processes, and state guidelines encourage the measurement of the spatial distribution of planned sites for low-income housing with respect to opportunity. We propose and evaluate a fair housing land use score (FHLUS) that measures whether local governments’ land use policies promote inclusion across neighborhoods. We illustrate the FHLUS by examining zoning and housing plans for three municipalities in California that differ in terms of neighborhood variation in incomes. In all three cases, we found that municipal zoning and housing plans exacerbated patterns of segregation rather than reversed them. Our metric is more precise than existing approaches, but all measures of this phenomenon will be less useful in smaller, more homogenous jurisdictions. The analysis raises important questions about the geographic scale and outcome measures for AFFH analysis and expectations for municipalities of different sizes and levels of diversity.

Takeaway for practice

Our metric is a useful tool for advocates and planners at all levels of government. We recommend the federal government consider incorporating it into the AFFH toolkit and practicing planners employ the measure to analyze local zoning and investment decisions. The Technical Appendix is a step-by-step guide, including an Excel formula.

Keywords:

• affirmatively further fair housing
• California
• plan evaluation
• zoning reform

The Fair Housing Act (FHA) of 1968 not only prohibited racial discrimination in housing but also required all federal executive departments and agencies to administer programs related to housing and urban development to affirmatively further fair housing (AFFH). Yet the FHA did not specify how departments and agencies were to do this. It was not until 2015 that the U.S. Department of Housing and Urban Development (HUD) issued a substantial rule on AFFH, which empowered HUD to evaluate the actions of grantees—including local government grantees—across four broad goals: addressing overall disparities in housing and opportunity, reversing housing segregation, transforming racially and ethnically concentrated areas of poverty, and complying with civil rights and fair housing laws.

The 2015 rule did not provide grantee agencies with expectations about timeliness toward progress or preferred approaches to achieving these fair housing goals. Instead, the rule left these decisions up to grantees. The procedural logic was that each locality should first analyze its own history, context, and challenges and then develop appropriate programs to work toward fair housing based on this analysis.¹

When California incorporated an AFFH requirement into its existing statewide planning process in 2019, the state adopted the procedural logic of HUD’s 2015 rule, along with the data tools and expectations for analysis. The state also created a needed focus on land use planning in a way federal rules have not. California embedded the AFFH requirement into its existing planning mandate, which includes oversight of site-specific plans for housing growth. California’s Department of Housing and Community Development (HCD) now requires local governments to quantitatively assess the spatial distribution of land zoned to allow low-income housing² development across neighborhoods by various measures of opportunity. California law reflects the fact that high-density zoning is necessary, but not sufficient, to enable low-income housing development.

In this study, we asked how federal and state regulators can better assess local land use plans to further integration. We focused on land use plans because they continue to play a major role in perpetuating socioeconomic and racial segregation and because local government planners control this structural impediment to integration. The first-order challenge in achieving AFFH goals is forcing higher-income municipalities to allow multifamily and subsidized housing at all, and a regional perspective is therefore important. The distribution of multifamily zoning with respect to neighborhood opportunity within municipalities also matters, however, and is something higher levels of government could regulate.

HCD’s guidance on how jurisdictions should assess their land use plans is inadequate. We created a metric, the fair housing land use score (FHLUS), to address this problem and provide a potential approach for state and federal AFFH guidance. The FHLUS quantifies the distribution of land zoned for multifamily housing (including low-income housing) across neighborhoods ranked by an indicator of opportunity or segregation. The metric is flexible to definitions of its component variables. For this study, we used California’s definition of zoning for low-income housing and median incomes and racial/ethnic homogeneity as proxies for neighborhood opportunity. Analysts could use the FHLUS with other proxies, such as environmental conditions, access to jobs, or an opportunity index. The Technical Appendix includes a step-by-step guide to calculating the FHLUS.

Our analysis illuminates a distinction in AFFH goals between the location of subsidized housing production and land use planning more broadly. Higher land costs in a region’s high-opportunity neighborhoods might mean fewer units of low-income housing could be built with a given level of subsidy, compared with less affluent areas (Kelly & Ellen, 2022). Zoning reform, however, does not face this tradeoff. Regulators should therefore prioritize subsidy allocation and zoning reforms differently and rely on the FHLUS more to evaluate land use plans than to allocate housing subsidies.

To illustrate our metric, we used three California municipalities—Yorba Linda, Inglewood, and Santa Monica—as examples of places with different levels of neighborhood income diversity. We found that existing zoning and recently developed housing plans in these municipalities exacerbated patterns of economic segregation rather than reversed them. We also show that the FHLUS performs better than the approach recommended by HCD. We note the primary challenge to any measure of a spatial distribution is that it will be less useful in smaller, homogenous cities.

Relevant Literature: Implementing AFFH and Plan Mandates

Federal AFFH Rules and Analyses of Effectiveness

Decades of intense debate over AFFH stem in part from a disagreement about whether to encourage mobility into higher opportunity neighborhoods or focus on investment in disinvested neighborhoods (Bostic et al., 2021). Ultimately, HUD’s 2015 rule embraced both approaches in its four broad goals (stated above). AFFH’s first systematic rule, adopted in 1988, required grant recipients to conduct an analysis of impediments to fair housing choice and take action to address identified impediments. This rule had little direct impact (Steil et al., 2021) but created a procedural logic of grantee self-analysis and independent program development that continues to govern AFFH implementation.

The 2015 rule was more prescriptive, providing localities specific instructions for their self-analysis and data. HUD required grantees to a) engage their community using data; b) analyze segregation patterns, housing needs, and disparity in access to opportunity within jurisdictions; and c) generate an assessment of fair housing that identified key contributors to segregation and disparate outcomes (O’Regan & Zimmerman, 2019; Steil & Kelly, 2019). HUD’s data and mapping tool (U.S. Department of Housing and Urban Development, n.d.), however, lacked specific expectations about analysis of land use policy, which should be part of a strong planning mandate (Berke & Godschalk, 2009). Most important, HUD did not require grantees to analyze the role of zoning in segregation or require that programs addressing fair housing include zoning reform. The guidelines recommended considering land use policies as part of the problem, including them as possibilities among a crowded field: seven potential data areas and seven areas of planning. The guidelines also mentioned zoning in an evaluation checklist of 11 potential contributing factors³ to fair housing problems (HUD, 2015).

In addition, HUD’s expectations about progress toward established benchmarks have been limited. HUD emphasizes progress but only toward goals localities set for themselves. The 2015 AFFH rule thus depended on grant recipients making change through local processes and had weak enforcement mechanisms (Steil et al., 2021; Zasloff, 2020). An open-ended definition of fair housing outcomes creates an implementation challenge. As Kazis (2021) argued, an AFFH policy might do better to focus on eliminating specific practices that are well documented to contribute to unfair housing, such as large-lot zoning. This is an important potential complement to a required focus on land use plans and the use of outcome metrics.

California’s AFFH Rule and Statewide Planning Context

Local and state governments throughout the United States responded to the Trump administration’s expressed intent to dismantle the federal AFFH process (Abraham, 2021). California adopted Assembly Bill 686: the AFFH Law in 2018 (2018 California Legislative Serv. Ch. 958, AB 686, 2018), requiring state agencies and local governments to use federal, state, and local data to analyze housing inequality, segregation, and concentrated poverty and develop programs in response to this analysis (Affirmatively Furthering Fair Housing, 2018). California’s HCD enforces this new requirement as part of an existing plan mandate that requires cities and counties to update the housing element of their general plan every eight years (Affirmatively Furthering Fair Housing, 2018).

California’s housing planning system is complex and relies on multiple levels of government. The state, in cooperation with regional councils of government, periodically determines regional housing needs and allocates housing targets for different income levels to constituent local governments (Draft Allocation of Regional Housing Needs, 2020). After the local government receives its housing target, it must update the housing element of its general plan to accommodate the specified number of new units (Elmendorf et al., 2021). The housing element must either demonstrate that the jurisdiction has parcels with sufficient additional capacity in their zoning for new housing above these targets or commit to rezoning land to create the zoning space for growth (Monkkonen et al., 2023). A share of these housing targets is for low-income housing, and parcels used to meet low-income targets must satisfy criteria of size and allowable density.

Historically, this process has proved inadequate to spur necessary housing production (Baer, 2008; Lewis, 2005; Ramsey-Musolf, 2016) or dismantle exclusionary zoning in California (O’Neill et al., 2022). Several deficiencies have dulled its impact, but most relevant to AFFH has been the allocation of housing targets by regional governments to municipalities in a way that did not promote integration (Ramsey-Musolf, 2020).

California’s AFFH guidance (California Department of Housing and Community Development, 2021) requested that local governments analyze the distribution of parcels with potential for new housing with respect to segregation patterns. HCD suggested that the site inventory and accompanying analysis must identify and analyze selected sites, map the location of the sites, indicate the number of projected units for each site and represent the assumed affordability (i.e., lower, moderate and above moderate) for each site, and evaluate relative to socioeconomic patterns. (California Department of Housing and Community Development, 2021, p. 45)

HCD provided a suggested method for jurisdictions to do this analysis (California Department of Housing and Community Development, 2021). In brief, they asked jurisdictions to estimate the share of low-income units allotted to tracts that have an above-average share of low-income households as well as the share of units in tracts with more than twice the average share of low-income households.

HCD’s proposal for measuring the distribution of parcels across neighborhoods is a laudable rough assessment but imprecise. Their measure does not reflect the full distribution of opportunity in a city, focusing instead on tracts that are below (or above) average in terms of incomes, racial composition, disability status, or other characteristics. The disadvantage is the HCD approach treats tracts as identical no matter their size and treats all below-average income tracts the same. As we show in our analysis, the HCD approach can give a false impression that a housing plan distributes sites relatively evenly across a city’s neighborhoods when it in fact does not. The measure does not differentiate among tracts that are below average income or how many units are planned for each one. The HCD measure is slightly easier to calculate, but the FHLUS offers a more precise measure using the same data.

Measurement Challenges in Regulating Land Use to Reduce Segregation

Segregation occurs both within and across municipalities in a metropolitan area, and households are segregated by socioeconomic status as well as race/ethnicity. Exclusionary zoning creates inaccessible and segregated neighborhoods within diverse and homogenous municipalities. To measure progress toward removing barriers to segregation, planners have to consider the geographic scale of integration and the measure of neighborhood conditions. Most land use regulation is local, but segregation patterns and housing markets span entire metropolitan areas.

If an urban area had only one municipal government, the municipality could relatively easily examine whether its land use policies entrenched or dismantled patterns of segregation and design reforms accordingly. But for an urban area with many municipalities that are diverse both in size and in relative homogeneity, the task is more complex. Moreover, state and federal regulators are limited in their ability to treat municipalities differently, although they can prioritize enforcement of AFFH requirements.

How Should Regulators Prioritize Reforms for Small, Homogenous Municipalities?

A fundamental question facing policymakers and regulators in enforcing an AFFH mandate is whether it is sufficient to zone for multifamily housing anywhere in municipalities that are relatively homogenous and wealthy. Or should the goals of AFFH require that all municipalities allow multifamily housing in all types of neighborhoods, especially their most affluent and exclusionary?

On the one hand, many exclusionary suburbs are small and homogeneous (Marantz & Lewis, 2022), so allowing multifamily housing anywhere in these places would promote integration at a regional level. The extensive literature on the impact of neighborhood conditions on life outcomes also focuses on the role of neighborhood differences within regions (for a recent summary, see Chyn & Katz, 2021). It is unclear, therefore, whether relative neighborhood incomes within a high-income municipality affect individual life outcomes. Moreover, if AFFH implementation prioritizes subsidies in the most affluent neighborhoods, fewer low-income housing units may result because of higher land costs (Kelly & Ellen, 2022).

On the other hand, neighborhoods differ even in relatively homogenous municipalities. The goal of integrated and balanced neighborhoods suggests all cities should share responsibility for AFFH, not just those that are larger and more diverse. Moreover, AFFH strategies to reform land use can be distinct from those for allocating subsidies. The administrative costs for rezoning are low no matter the land value. Rezoning in higher-income neighborhoods will also yield more multifamily development overall because of demand, which can alleviate regional housing unaffordability (Phillips et al., 2021). In California, forcing cities to identify sites for low-income housing in high-income neighborhoods could lead to more rezoning because the existing zoning tends to be low density.

Thus, an important nuance is that the approach to allocating housing subsidies does not need to be the same as reforming land use more broadly. AFFH seeks to improve residential outcomes for households in subsidized housing but also to make housing more widely available at various income levels. Regulators should use separate measures to decide how to maximize the impact of housing subsidies compared with pushing for land use reform.

Measuring Opportunity

A measure of neighborhood conditions or opportunity is essential to evaluate local action on AFFH. Many opportunity indexes exist, though, as Zheng et al. (2021) have demonstrated, simpler approaches may be preferable. A spatial metric could employ an index to rank neighborhoods across several variables, but we prefer a simple proxy: median household income.

California’s local governments rely on regional opportunity maps to assess progress toward fair housing goals. These maps rank urban census tracts within regions using an index of 11 indicators, ranging from poverty rates and home values to job accessibility and local environmental quality (for details, see California Fair Housing Task Force, 2021). Regional analysis is important, but relying exclusively on regional-level rankings masks considerable variation within municipalities.⁴

As our case cities illustrate, opportunity maps may support a positive assessment of the distribution of multifamily housing in exclusionary places, even as these places continue to concentrate their sites in a handful of their lowest-income neighborhoods. California’s opportunity maps classify roughly half of the state’s urban municipalities as either entirely low resource or entirely high or highest resource. For these places, the maps cannot usefully guide internal analysis of housing plans even though within these cities neighborhood income and demographic characteristics do vary.

Our measure also builds upon existing efforts to identify whether lower-rent multifamily housing is present in opportunity neighborhoods. In research evaluating the success of the Housing Choice Voucher program at facilitating access to higher opportunity neighborhoods, for example, McClure (2010) assessed the supply of such neighborhoods and the presence of housing units below the fair market rent. He found that 52% of block groups had poverty rates less than 10% but only 28% of rental units below fair market rent were in those neighborhoods. Further, using more expansive measures of neighborhood opportunity (e.g., race, education, and employment data), McClure estimated only 1 million rental units would be available and only 13% of block groups would meet those criteria.

Owens (2019) also measured housing segregation at various geographic scales and found that the segregation of housing types is strongly tied to the segregation of people. Owens also demonstrated the importance of measuring housing segregation at multiple geographic scales by finding that housing segregation between jurisdictions (places) was responsible for only 40% of renter-to-owner segregation between neighborhoods. Thus, nearly 60% of housing segregation operated within city boundaries. This suggests housing types contribute substantially to segregation by income, and land use plans determine where different types of housing can exist.

Caveats and Considerations in Land Use Reforms to AFFH

Multifamily housing is not automatically affordable to people with lower incomes, but it is considerably more affordable than single-family housing. People with higher incomes and White people are much more likely to live in single-family housing. In 2019, 64% of housing units in the United States were detached single-family homes and were home to 67% of Whites. In contrast, only 48% of Blacks and 49% of households with incomes less than $30,000 lived in single-family homes (U.S. Census Bureau, 2019).

Racial desegregation motivated the 1968 FHA and recent AFFH rules. Racial and economic segregation are not synonymous, even if they overlap, and the past and present forces that shape racial segregation are not simply differences in income and wealth, even if wealth and income gaps contribute to racial segregation (Kucheva & Sander, 2018; Ong et al., 2016). California has worked to further fair housing in land use plans by correlating housing density with affordability potential and thus economic integration, in part because this is the best existing policy option. State law requires a default standard of 30 dwelling units per acre for urban jurisdictions for low-income sites, as well as a size between 0.5 and 10 acres (Inventory of land suitable for residential development used to identify sites, 2022) in recognition of this fact. We also relied on this density threshold in our categorization of land use plans.

Jurisprudence prohibits racially explicit land use and housing practices, meaning governments in the United States cannot mandate racial/ethnic integration. Governments can, however, make racial integration possible by rezoning land in all neighborhoods to allow for the production of housing that is affordable to people with lower incomes. This land use reform would also facilitate a more dispersed construction of subsidized housing, which in the United States is almost entirely multifamily.

Methods: Quantifying AFFH in Land Use Plans

Land use planning can best promote integrated living patterns by allowing housing for all income levels in all neighborhoods. For plans to reverse the legacy of segregation, they must disproportionately change land use rules in higher opportunity neighborhoods. The FHLUS quantifies how the local zoning for multifamily or subsidized housing is distributed across neighborhoods ranked by a measure of opportunity. Measuring the spatial distance of different income groups poses a challenge, which is why a traditional measure of spatial evenness like the dissimilarity index does not work (for a discussion of ordinal segregation measures, see Reardon & Firebaugh, 2002).

The FHLUS is flexible in three ways. It can evaluate different planning actions (e.g., existing zoning or plans for new low-income housing), it can measure opportunity using different indicators (e.g., income or racial–ethnic homogeneity), and it can function at different geographic scales (e.g., block groups or tracts within cities, or municipalities within regions). Because the FHLUS ranks neighborhoods using a variable like income, it can be applied to jurisdictions with diverse geometric shapes, including fragmented unincorporated county areas.

California’s housing elements provide a useful way to illustrate the FHLUS because they must identify sites available for low-income housing. Figure 1 presents two hypothetical plans for low-income housing. Plan A disproportionately sites future low-income housing in the municipality’s lower-income neighborhoods, which does not advance fair housing goals. Plan B advances fair housing goals because it disproportionately sites low-income housing in higher-income neighborhoods.

Figure 1. Two hypothetical plans for low-income housing development overlaid on neighborhood median household incomes.

Source: Authors.

The difference between the two plans is clear in Figure 1 because it is dramatic. Yet quantifying the distributions will allow regulators to compare less obviously different plans and to track changes within the same municipality over time.

The FHLUS is based on the Gini coefficient, a standard measure of income inequality and the basis for one index of segregation (Rothwell & Massey, 2010). To generate the FHLUS, we first measured the share of the municipality’s low-income sites and the share of municipal land in each neighborhood. Just as the Gini coefficient measures income inequality by comparing the cumulative distribution of households by income with a perfectly equal distribution of incomes, we measured the distributional inequality of sites by neighborhood income. Because municipalities differ in the distribution of land area by neighborhood income, we had to standardize the distribution of sites by the distribution of land in each neighborhood. We then ranked neighborhoods by income from lowest to highest to compare the cumulative share of sites with the cumulative share of land.

For a detailed, step-by-step guide to calculating the FHLUS, see Technical Appendix A and the example spreadsheet with the formula in Technical Appendix B.

Figure 2 shows the hypothetical distribution of sites across neighborhoods. They correspond to Plan B in Figure 1 in which most sites are in lower-income neighborhoods. Panel A shows the share of housing units available in each neighborhood and the share of the municipality’s land each neighborhood represents. Panel B shows the cumulative share of units and land in each neighborhood ranked by income. When the red line (online version), which represents the share of sites for low-income housing, is higher than the blue line (online version), which represents land area, the municipality’s sites are disproportionately located in lower-income neighborhoods.

Figure 2. A hypothetical distribution of units across neighborhoods: Percentage per neighborhood (A) and cumulative percentage (B). These represent the map shown in Figure 1, Plan A. Source: Authors.

Like the Gini coefficient, the FHLUS is the area of difference between the two lines. It ranges from −1 to 1. A score of −1 indicates all sites are in the lowest-income neighborhood, a 0 represents an equal distribution of sites across neighborhoods, and a 1 indicates all sites are in the highest-income neighborhood. Negative scores mean sites are disproportionately located in the lower-income neighborhoods of the municipality and positive scores mean sites are in the higher-income neighborhoods. The FHLUS for the hypothetical distribution in Figure 2 is −0.6.

Limitations of the FHLUS

Creating a useful measure of relative spatial distributions is challenging. The advantage of a measure like that proposed by HCD is interpretability. The HCD measure translates to “What percentage of a plan’s low-income housing is sited in the lower-income half of the city?” However, the simplicity can be misleading. Using the HCD measure, two very different distributions could have the same score. A city with all of its multifamily zoning in the lowest-income neighborhood would have the same score as a city with all its multifamily zoning in a neighborhood just below the average. The FHLUS measure for the first city would be much lower than for the second city. Yet, the interpretation of the FHLUS is a slight limitation. A score of −0.6 is clearly worse than −0.2, but it does not translate into words as easily.

The other limitation of the FHLUS is one that applies to any measure of a spatial distribution, including the HCD approach. Spatial measures of this nature are less meaningful in places with few neighborhoods, places that are homogenous, and places with a very small presence of the phenomenon being measured (i.e., very little multifamily zoning). Some municipalities only span a few census tracts, which makes any spatial measure less useful than in a place with dozens of tracts. Similarly, the impact of zoning’s spatial distribution matters more in more heterogeneous places because neighborhoods are more distinct from one another. Regulators should be aware of this issue and might consider a different measure for municipalities with fewer than four census tracts. Municipalities that cover four or more census tracts should have a positive FHLUS as the expectation for the AFFH requirement.

Three Cases Illustrate the FHLUS

To illustrate the FHLUS, we selected three cases, municipalities⁵ with low (Yorba Linda), medium (Inglewood), and high (Santa Monica) neighborhood income diversity. Yorba Linda was in the bottom 10% of California cities in terms of neighborhood income diversity, Inglewood was near the median on this measure, and Santa Monica was in the top 10%. To give some context, tract-level median household incomes in Yorba Linda ranged from $76,000 to $171,000, in Inglewood from $32,000 to $99,000, and in Santa Monica from $67,000 to $234,000. The variation in ranges across cities is notable, yet even in the relatively homogenous municipality of Yorba Linda, the variation between neighborhoods is nontrivial.

We also chose cities that highlight the varied interaction between socioeconomic status, race/ethnicity, and housing. In the United States, many communities of color are not low income (Patillo, 2005; Wen et al., 2009), and exclusionary suburbs are not always predominantly White (Orfield & Luce, 2013). Yorba Linda was a majority White municipality with a substantial and growing high-income Asian population and had very little multifamily housing (less than 1% of its total land area). Inglewood was almost entirely Black and Latino, had relatively lower incomes than the state’s median municipality, and had a slight majority of housing in single-family neighborhoods. Santa Monica was a majority White, high-income city, but a slight majority of its housing stock was multifamily. Inglewood had been a predominantly Black community but, like much of Los Angeles, lost a significant share of its Black population.

The cities also varied in the amount of multifamily zoning they contained. In Santa Monica, 22% of the residentially zoned land allowed multifamily housing, which was more than in Yorba Linda (where only 8% of residential land allowed multifamily) but less than in Inglewood, where 38% of residential land allowed multifamily housing (O’Neill et al., 2022).

Table 1 presents descriptive statistics for the three cities and the median California municipality, as well as correlations across neighborhoods between these social (income and race/ethnicity) and housing (share multifamily and zoning) attributes. We measured diversity using an entropy score (Reardon & Firebaugh, 2002) based on the share of four different racial/ethnic groups (Black, non-Hispanic White, Asian, Hispanic) in each tract. Entropy scores were higher in tracts that were more heterogeneous, approaching 0 when there was an equal proportion of each group and larger than 1 when only one group was present.

Table 1. Descriptive statistics: Three cases and median California municipality.

Variables and correlations	Yorba Linda	Inglewood	Santa Monica	Median California municipality
Population (’000s)	74	114	92	32
No. tracts	19	29	19	20
Median income (USD ’000s)	132	54	103	83
Income variation across neighborhoods	0.29	0.39	0.50	0.36
% Black	1	38	4	2
% Hispanic	17	52	15	32
% Asian	21	2	10	7
% Non-Hispanic White	57	4	65	44
% Housing in multifamily buildings	7	38	56	15
% Land zoned multifamily (30 dwelling units per acre)	4	27	16	NA
Correlation between neighborhood racial/ethnic diversity and multifamily housing as a share of stock	0.15	0.50	0.20	0.20
Correlation between neighborhood household income and multifamily housing as a share of stock	−0.59	−0.42	−0.58	−0.48
Correlation between share of neighborhood population that is Black and household income	−0.16	0.33	−0.42	−0.12
Correlation between share of neighborhood population that is Asian and household income	0.38	−0.09	0.11	0.01
Correlation between share of neighborhood population that is Latina/o and household income	−0.29	−0.22	−0.30	−0.33
Correlation between neighborhood racial/ethnic diversity and household income	0.08	−0.17	−0.22	−0.01

Source: Authors (5-Year ACS 2015–2019 and local zoning maps from the three cities).

Two correlations are consistent across the three cities: neighborhoods with more multifamily housing stock had higher racial/ethnic diversity and lower median household incomes. Because single-family homes are less affordable, neighborhoods where they predominate have higher incomes. A more diverse set of households can live in multifamily housing because of its affordability (Owens, 2019). Other correlations differed from city to city. In Inglewood, neighborhoods with more Black households were higher income and those with more Asian households were lower income, patterns that contrasted with most municipalities in the state. In Yorba Linda, neighborhoods with more Asian households were higher income, also different from most municipalities. In the median California municipality, racial/ethnic diversity did not correlate with household income. This correlation was negative in Inglewood and Santa Monica and positive, although small, in Yorba Linda.

Two Applications of the Metric: Existing Zoning and Sites for Low-Income Housing

We used the FHLUS to assess whether local land use policies—existing zoning and municipalities’ 2013 and 2021 plans for low-income housing development—were working against or exacerbating neighborhood socioeconomic segregation. To illustrate FHLUS’s flexibility, we calculated scores for the three municipalities using neighborhood income and racial/ethnic composition. Before presenting the full results, we describe the calculations for Santa Monica in detail to illustrate the operation of the FHLUS.

Where Is Existing Multifamily Zoning?

Figure 3 shows an overlay of Santa Monica’s zoning (density and use controls) on the median household incomes of census tracts. The map suggests that there is limited overlap between the multifamily zoning that allows at least 30 dwelling units per acre and the higher-income neighborhoods of the municipality.

Figure 3. Santa Monica multifamily zoning and tract median household incomes. Sources: Authors; City of Santa Monica, n.d.; U.S. Census Bureau, 2019.

Figure 4 is a graphic representation of the cumulative share of land area and multifamily zoning across neighborhoods, corresponding to the map in Figure 3. The FHLUS is −0.43. Recall that the metric ranges from −1 to 1, with scores below 0 indicating a distribution that is disproportionately in the lower-income neighborhoods of the municipality.

Figure 4. Cumulative distribution of land zoned for at least 30 dwelling units per acre and all land of neighborhoods ranked by income. Sources: Authors; City of Santa Monica, n.d.; U.S. Census Bureau, 2019.

Planning for New Low-Income Housing in Lower-Income Neighborhoods

In the inventory of sites that local governments in California submit to the state in their housing element, jurisdictions listed parcels zoned to allow low-income housing development. These sites must meet a minimum threshold of allowable density (30 dwelling units per acre in urban areas) and size (more than 0.5 acre). We geocoded sites from the three municipalities in 2013 and 2021. To illustrate, Figures 5 and 6 overlay Santa Monica’s 2013 and 2021 housing element sites on choropleth maps indicating median household incomes across census tracts. The maps are paired with graphs of the cumulative distribution of low-income housing units across neighborhoods ranked by median household incomes (City of Santa Monica 2013, 2021).

Figure 5. Santa Monica’s 2013 low-income housing sites and neighborhood incomes and cumulative distribution of land and low-income housing sites by neighborhood incomes in 2013. Sources: Authors; City of Santa Monica, 2013; U.S. Census Bureau, 2013.

Figure 6. Santa Monica’s 2021 affordable housing sites by neighborhood incomes and cumulative distribution of land and affordable housing sites by neighborhood income in 2021. Sources: Authors; City of Santa Monica, 2021; U.S. Census Bureau, 2019.

As with the city’s zoning, a visual inspection of sites selected by Santa Monica for its housing element in 2013 suggests sites were disproportionately located in lower-income neighborhoods. The map of sites in 2021 looks less concentrated, but we cannot be certain without quantifying the relationship.

The FHLUS for the 2013 sites is −0.64 and for 2021 it is −0.55. The 2021 sites are an improvement over 2013, yet both plans for low-income housing development worked against the goal of integration. Moreover, the 2021 plan’s score was lower than the FHLUS score for its existing zoning, which means the municipality was identifying sites in its lower-income neighborhoods that have multifamily zoning for future low-income housing development.

Assessing plans using a metric rather than a map to compare and track progress is more rigorous.⁶ Neither federal (HUD, 2015) nor state (California Department of Housing and Community Development, 2021) AFFH data analysis tools require such quantification. Instead, those tools rely on maps and, in the case of HCD, an exercise that does not produce precise or comparable results.

Summary of the Three Cases

Table 2 presents the FHLUS for the three municipalities using tract data on median household incomes and the entropy measure of racial/ethnic diversity. We also calculated the assessment approach proposed by California HCD.

Table 2. FHLUS and HCD scores for multifamily zoning and housing plans in 2013 and 2021.

	Yorba Linda		Inglewood		Santa Monica
Variable	FHLUS	HCD (%)	FHLUS	HCD (%)	FHLUS	HCD (%)
Median household income
Zoning (2019)	−0.14	100	−0.23	50	−0.43	79
Fifth-cycle sites (2013)	−0.42	52	−0.23	48	−0.64	84
Sixth-cycle sites (2021)	−0.52	60	−0.49	70	−0.55	78
Racial/ethnic homogeneity
Zoning (2019)	−0.10	100	−0.27	94	−0.30	71
Fifth-cycle sites (2013)	0.30	69	−0.56	96	−0.42	79
Sixth-cycle sites (2021)	0.43	38	−0.36	84	−0.46	77

The FHLUS shows that all three municipalities had more land zoned for multifamily housing in their lower-income neighborhoods. Yorba Linda had the lowest segregation of multifamily housing, with a score of −0.14, and Santa Monica had the highest at −0.43. Recall the score ranges from −1 to 1, and if it is negative, then multifamily zoning is disproportionately in lower-income tracts. A score of −0.43 means a city would have to rezone many of the parcels in above-average income tracts to yield a score above 0.

The FHLUS for the three municipalities’ housing element sites indicate that they were disproportionately located in the cities’ lower-income neighborhoods. Inglewood’s score was the same for zoning and its 2013 housing plan (−0.23), meaning they selected sites for low-income housing evenly across neighborhoods zoned for multifamily housing. In Yorba Linda and Santa Monica, however, scores for low-income housing plans were lower than for their existing multifamily zoning. Among the areas that allowed multifamily housing, these cities identified sites in the relatively lower-income neighborhoods for low-income housing production.

Comparing the distribution of sites from 2013 to 2021 plans with that of 2021 to 2029 plans shows that Yorba Linda and Inglewood have done worse at siting low-income housing in higher-income neighborhoods in 2021 compared with 2013. Santa Monica has done better, as indicated by its score going from −0.64 to −0.55.

In theory, the HCD measure should be negatively correlated with the FHLUS. If most sites are in higher-income neighborhoods, the HCD score should be lower than 50% and the AFFH score should be positive. In some cases, the two measures match. Yet Inglewood’s zoning evaluation revealed the importance of a more precise measure. Its HCD score of 50% means half of its multifamily zoning was in tracts with below-average incomes. The FHLUS of −0.23, on the other hand, revealed that among those lower-income tracts, the zoning was disproportionately located in the lowest-income tracts.

Yorba Linda’s scores were somewhat challenging because only two tracts had multifamily zoning and its housing plans proposed low-income sites in very few tracts (three or four). In 2013, one of the few tracts with many sites was barely above the average income, meaning that Yorba Linda did well on the HCD score (52%), whereas the more precise FHLUS (−0.42) indicated that low-income housing sites were disproportionately located in lower-income neighborhoods.

Turning to racial/ethnic diversity, the FHLUS measures whether zoning and sites are disproportionately located in the already diverse neighborhoods of a city. The HCD measure is the share of a city’s zoning or sites in tracts with above average levels of diversity. Yorba Linda’s limited zoning for multifamily housing and low-income housing plans were in relatively homogenous neighborhoods, whereas Inglewood and Santa Monica had disproportionately more existing multifamily zoning and plans for low-income housing in their more racially and ethnically diverse neighborhoods.

Diversity was different in these three places. In Inglewood, for example, two of the four tracts with the most multifamily zoning had an above-average share of Black residents, and the other two tracts were majority Hispanic. There was no consistent pattern in Santa Monica. Most tracts with multifamily housing had a lower share of White residents than average but differed in which racial/ethnic groups were more well represented than in the rest of the city.

For context, recall that only 6% of Yorba Linda’s households were classified as poor, whereas in Santa Monica this share was 10%, and in Inglewood it was 17%. In Yorba Linda, the two tracts with all of the city’s multifamily zoning were among those with the highest poverty rate in the city, although that was only 7%. The situation was similar in the other cities, where a handful of tracts had most of the cities’ multifamily zoning and the highest poverty rates. In Santa Monica and Inglewood, however, these poverty rates were between 15% to 17% and 17% to 25%.

In their 2021–2029 housing element, Santa Monica plans to accommodate 53% of its low-income housing units in only three tracts, which are the most diverse and lowest income in the city, ranging between $65,000 and $75,000 for the median household. Yorba Linda plans for 52% of its low-income housing units in only two tracts: One is the second most diverse tract in the city and has an average median income for the city ($130,000), and the other is majority White and has a median household income of $85,000. These are not low-income tracts, but they are lower income than most of the city.

These three cases illustrate the precision of the FHLUS and provide context for what a negative FHLUS score means in a relatively homogenous, higher-income municipality. Applying the goal of balanced and integrated cities as per the AFFH mandate in a consistent manner means that relative neighborhood opportunity should matter for land use plans, even in places that are more homogeneously higher opportunity. The case of Yorba Linda also suggests that two additional measures—a municipality’s share of multifamily zoning and its concentration—would be useful additions to measures of spatial distribution. Regulators should establish minimum expectations about how much multifamily zoning municipalities should have and how dispersed it should be as an important step toward achieving balanced and integrated living patterns.

Discussion and Conclusion

Here we present the FHLUS, a new measure of whether municipal land use policies allow for integration across neighborhoods. The FHLUS will assist state and federal agencies in tracking progress toward the AFFH goal of replacing segregated living patterns with truly integrated and balanced living patterns. We used three California municipalities to illustrate the advantages of our approach, measuring whether their zoning for multifamily housing and plans for low-income housing were distributed across neighborhoods in a way that would alter the segregated status quo. We found that even in recently created and state-certified housing plans with extensive AFFH analysis, land use policies exacerbated segregation rather than worked against it.

Our analysis suggests that regulators can and should measure whether land use policies promote integration. Although any spatial measure of land use policy is slightly less important in smaller or more homogenous cities, we contend that state and federal regulators should add the FHLUS to existing requirements in fair housing analysis along with expectations about what constitutes an acceptable score. Implementing agencies can use a variety of strategies to incorporate the FHLUS while recognizing the diversity of contexts. Our recommendation is that HUD and California’s HCD set a positive score as a minimum expectation for local land use plans of municipalities that span four or more census tracts.

If jurisdictions believe that reforming land use plans so that they obtain a positive score is not possible, the burden of proof should fall on them to justify the reason to regulators. Just as the National Environmental Policy Act encourages compliance with the threat of a lengthy and costly full review (Kazis, 2021), we suggest that municipalities would be more likely to meet quantitative benchmarks rather than undergo a costly review process.

HUD and HCD could employ a similar metric to assess progress on transforming racially and ethnically concentrated areas of poverty and the implied place-based goals like the distribution of parks and other public amenities. By measuring the spatial distribution of non-housing investments across neighborhoods, state and federal agencies can track progress toward a more equal local environmental quality as a complement to the measure of land use policies’ role in segregation.

Land use reform is only one aspect of one goal of the broad AFFH mandate; on its own it will not integrate urban areas by race, ethnicity, and socioeconomic status. Land use policies can, however, make it impossible for integration to occur, and reforms can enable integration that is prohibited today. Informational barriers and the housing search process are important drivers of residential segregation (Bergman et al. 2020; Sander et al., 2018), which need to be addressed separately. But all the information in the world cannot integrate neighborhoods that do not have housing affordable to all income groups. In addition to debating how to measure integration in land use policies, government agencies need progress metrics for other fair housing goals.

Research Support

This project was funded by a grant from the Robert Wood Johnson Foundation.

Acknowledgments

The authors thank Michael Manville, Noah Kazis, Aaron Barrall, anonymous reviewers and the editor of the Journal of the American Planning Association, and participants at the 2021 Association of Collegiate Schools of Planning Conference, the 49th Annual Pepperdine Law Review Symposium in March 2022, and the Second Annual Housing Forum at the University of Oregon in June 2022 for their insightful comments and suggestions.

Supplemental Material

Supplemental data for this article can be can be accessed online at https://doi.org/10.1080/01944363.2023.2213214.

Additional information

Notes on contributors

Paavo Monkkonen

PAAVO MONKKONEN ([email protected]) is a professor at the University of California, Los Angeles (UCLA).

Michael Lens

MICHAEL LENS ([email protected]) is an associate professor at UCLA.

Moira O’Neill

MOIRA O’NEILL ([email protected]) is an associate research scientist at the University of Virginia.

Christopher Elmendorf

CHRISTOPHER ELMENDORF ([email protected]) is a professor at UC Davis.

Gregory Preston

GREGORY PRESTON ([email protected]) is a PhD student at UCLA.

Raine Robichaud

RAINE ROBICHAUD ([email protected]) is a researcher in Berkeley, California.

Notes

1 The Biden Administration issued a rulemaking notice on measurable improvements advancing equity and providing access to opportunity for underserved populations in the June 2022 Unified Agenda of Regulatory and Deregulatory Actions. (Affirmatively Furthering Fair Housing, 2023).

2 We use the term low-income housing to refer to deed-restricted housing built with subsidies for people with low incomes, recognizing that there are varying definitions.

3 Examples of the other factors are community opposition, private discrimination, and lack of investment.

4 California has a regional fair share housing planning system. Currently, federal AFFH implementation does not have an avenue to create regional fair share systems, however, even if the between-municipality redistribution of low-income housing is more important than its within-municipality distribution.

5 We chose from municipalities in Southern California because their housing element update timeline was the same and before other regions in the state.

6 Changes in FHLUS scores could reflect changes in the underlying composition of neighborhoods as well as land use changes.