Mapping housing discrimination: using HUD FHEO complaint data to find fair housing violations

Since 1968, federal law has prohibited discrimination in the sale, rental, and financing of housing. Since the late 1990s, the Department of Housing and Urban Development has published the complaint records from that enforcement regime — tens of thousands of cases spanning race, disability, national origin, familial status, sex, and religion. The database captures who was alleged to have discriminated, where, on what basis, and what happened to the case. Almost nobody analyzes it systematically.

The HUD Fair Housing and Equal Opportunity complaint dataset is one of the most underused civil rights data sources in the federal government. It covers every complaint filed through HUD's administrative process and its partner state agencies going back decades. The complaint-level records disclose the basis of discrimination, the issue type, the property type, the respondent category, and the disposition — including whether the complainant received monetary relief or a change in the respondent's practices. Cross-referenced with HMDA lending data, DOJ Civil Rights Division enforcement actions, and Census demographic data, it maps where housing discrimination concentrates and which protected classes drive complaint volume.

This article covers the Fair Housing Act's legal framework, the difference between the administrative and judicial filing paths, the data structure of the HUD complaint database, the key statistics about what 50,000+ complaints reveal, and three research approaches that produce defensible findings.

What the Fair Housing Act covers

The Fair Housing Act of 1968, codified at 42 U.S.C. §§ 3601–3619, was enacted eight days after the assassination of Martin Luther King Jr. as part of the Civil Rights Act of 1968. Congress's stated purpose was to replace the segregated housing patterns that had been reinforced by decades of racially restrictive covenants, redlining, urban renewal displacement, and government-sponsored suburban development that systematically excluded Black Americans.

The Act prohibits discrimination in the sale or rental of housing, in the terms and conditions of housing transactions, in advertising, in financing, and in brokerage services. The prohibited bases are race, color, national origin, religion, and sex — the original 1968 coverage. In 1988, Congress added two more: disability and familial status (households with children under 18, including pregnant women). Some states and localities add age, sexual orientation, source of income, and other categories, but the federal HUD database tracks only the seven federal bases plus a small number of additional protected characteristics recognized in specific HUD programs.

HUD's Office of Fair Housing and Equal Opportunity — FHEO — is the agency responsible for enforcing the Act. FHEO investigates complaints, conducts compliance reviews of HUD-funded entities, and refers cases for charge or conciliation. HUD also administers a network of Fair Housing Assistance Program (FHAP) partner agencies — state and local fair housing agencies that have laws “substantially equivalent” to the federal Act and that receive HUD funding to process complaints on HUD's behalf. Complaints processed by FHAP agencies are included in the HUD dataset.

Two filing paths: administrative complaint vs. federal court

A person who believes they have experienced housing discrimination has two independent legal routes. Understanding which route the FHEO database captures — and which it does not — is essential for interpreting what the data shows.

The first path is the HUD administrative complaint. The complainant files with HUD (or a FHAP agency), HUD investigates, and if HUD finds reasonable cause to believe a violation occurred, it issues a Determination of Reasonable Cause and a charge of discrimination. At that point, either party can elect to have the case heard in federal district court; if neither party elects, the case goes before a HUD Administrative Law Judge. This path is free, does not require an attorney, and has a 100-day investigation target under the Act (42 U.S.C. § 3610(a)(1)(B)). The FHEO database captures every complaint filed through this path.

The second path is a private right of action under 42 U.S.C. § 3613, which allows any aggrieved person to file suit in federal or state court without first going through HUD. This path requires an attorney, involves litigation costs, and runs on the standard two-year statute of limitations rather than HUD's administrative timeline. Private lawsuits brought directly under the Fair Housing Act are not captured in the HUD complaint database unless they were first filed as an administrative complaint. The DOJ Housing and Civil Enforcement Section also has independent authority to bring pattern-or-practice cases; those cases also do not appear in the HUD complaint count.

The implication: the HUD database is not a complete census of fair housing violations or even of fair housing legal activity. It is a census of the administrative complaint path specifically. It systematically undercounts the total volume of housing discrimination — most discrimination never gets reported at all — and it misses the private litigation track entirely. What it captures is the population of people who chose the administrative route, which tends to include lower-income complainants and those without attorney representation more than the private litigation track does.

Data structure: what the FHEO complaint database contains

The FHEO complaint database as published by HUD contains complaint-level records with the following fields:

Complaint identification and timing

complaint_id     # HUD-assigned case number (unique per complaint)
filing_date      # Date complaint was filed with HUD or FHAP agency
closure_date     # Date the case was closed
days_open        # Days from filing to closure (derived)
agency           # HUD (direct) or the FHAP agency abbreviation

Basis of discrimination

# Seven federal bases, plus additional categories in some records
basis_race             # Boolean: race or color alleged
basis_national_origin  # Boolean: national origin alleged
basis_religion         # Boolean: religion alleged
basis_sex              # Boolean: sex alleged
basis_disability       # Boolean: disability alleged
basis_familial_status  # Boolean: familial status (children) alleged
basis_age              # Boolean: age alleged (state FHAP cases only)

# A complaint can allege multiple bases simultaneously.
# When disability is alleged, the specific theory often appears in
# issue_type (see below): reasonable accommodation, reasonable
# modification, or accessibility (design-and-construction).

Issue type

issue_type
  # Categories include:
  # Terms/conditions/privileges of rental or sale
  # Refusal to sell
  # Refusal to rent
  # Discrimination in financing
  # Discriminatory advertising
  # Harassment/hostile environment
  # Reasonable accommodation (disability)
  # Reasonable modification (disability, physical changes to unit)
  # Accessibility (design-and-construction violations, new multifamily)
  # Failure to make accessible (disability)
  # Coercion/intimidation/interference

Property type

property_type
  # Single-family home
  # Multi-family residential (2+ units)
  # Condominium / cooperative
  # Mobile home / manufactured housing
  # Vacant land
  # Commercial / mixed use (rare; FHA coverage limited)

Respondent type

respondent_type
  # Individual landlord / property owner
  # Corporate property management company
  # Real estate broker / agent
  # Mortgage lender or servicer
  # Homeowners association (HOA)
  # Housing authority / public housing
  # Condominium association
  # Property developer (design-and-construction cases)

Disposition and relief

closure_reason
  # No reasonable cause (NRC): HUD investigation found insufficient evidence
  # Conciliation/settlement: case resolved with agreement before charge
  # Administrative closure: complainant withdrew, failed to cooperate, or
  #   elected to pursue private litigation after reasonable cause finding
  # Charge issued: HUD found reasonable cause; elected to federal court
  # Successfully conciliated: agreement included relief for complainant
  # Referred to DOJ: pattern-or-practice element referred to Civil Rights Div.

monetary_relief    # Dollar amount of monetary relief, if any (may be $0
                   # for non-monetary conciliation agreements)
relief_obtained    # Boolean: whether any relief (monetary or equitable)
                   #   was obtained for the complainant

Key statistics: what 50,000+ complaints reveal

HUD receives approximately 7,000 to 9,000 fair housing complaints per year through its direct intake and FHAP partner agencies combined. The volume has been relatively stable over the past decade, with year-to-year variation driven partly by economic conditions (rental market tightness tends to increase landlord-tenant complaints) and partly by changes in HUD outreach and FHAP agency capacity.

The most striking fact in the aggregate data is the dominance of disability as a complaint basis. In recent years, disability has accounted for approximately 55 percent of all fair housing complaints filed — more than all other bases combined. Race, historically the Act's primary concern and the basis for which it was enacted, accounts for roughly 20 percent of annual complaint volume. National origin is third at approximately 10 percent, followed by sex (8 percent), familial status (7 percent), and religion (2 percent). These shares have been relatively stable since 2010, though disability's dominance is a post-2008 phenomenon discussed in the next section.

HUD's closure statistics reveal a system where most complaints do not result in a finding against the respondent. Roughly 60 to 65 percent of complaints are closed with a no-reasonable-cause determination — meaning HUD's investigation found insufficient evidence to support the allegation. Approximately 20 to 25 percent are resolved through conciliation or settlement, with the remainder split between administrative closures (withdrawal, failure to cooperate), charges issued, and referrals to DOJ. Of closed complaints, only about 15 percent result in any relief — monetary or equitable — for the complainant.

The conciliation success rate — the share of cases where HUD achieves a negotiated resolution — varies significantly by complaint basis. Disability cases involving reasonable accommodation (ESA requests, parking, grab bars) have among the higher conciliation rates because the legal standard is relatively clear and respondents often prefer to modify their policies rather than litigate. Race cases, where the evidentiary standard for discriminatory intent is higher, have lower conciliation rates and higher no-reasonable-cause closure rates.

The disability surge: reasonable accommodation and the ADAAA effect

Disability was added to the Fair Housing Act in 1988 by the Fair Housing Amendments Act. From 1989 through approximately 2007, disability complaints grew steadily but did not dominate the complaint docket. After 2008, disability began its ascent to more than half of all complaints — a shift with two distinct explanations.

The first explanation is the ADA Amendments Act of 2008 (ADAAA), which significantly broadened the definition of disability under the ADA. Although the ADAAA applies directly to the ADA (employment and public accommodations) rather than to the Fair Housing Act, which has its own definition of disability at 42 U.S.C. § 3602(h), the ADAAA's broader cultural and legal framing shifted how tenants, advocates, and courts understood disability accommodation obligations. HUD itself has interpreted its housing disability definition in ways consistent with the post-ADAAA broadening.

The second and more operationally significant explanation is the explosion of reasonable accommodation requests related to emotional support animals (ESAs). Landlords with no-pets policies began receiving requests — often accompanied by letters from online “ESA certification” services — to waive their pet policies for tenants claiming disability-related need for an animal. When landlords denied these requests, tenants filed FHEO complaints. By the mid-2010s, ESA-related complaints had become a substantial fraction of all disability complaints filed with HUD. HUD issued guidance in 2020 and again in 2024 attempting to clarify the documentation standards for ESA requests, but complaint volume has not declined substantially.

Other high-volume disability complaint categories include accessible parking (landlords denying reserved accessible spaces to tenants with mobility impairments), physical modifications to units (landlords refusing to permit grab bars, ramps, or threshold modifications), and design-and-construction cases (multifamily buildings built after 1991 that fail to meet the Act's accessibility requirements for new construction). The design-and-construction cases are structurally different from the others: they typically involve DOJ pattern-or-practice referrals rather than individual complainants, and the respondent is a developer rather than a landlord.

How to access the data

The HUD FHEO complaint data is publicly available through two channels. The primary channel is the HUD Fair Housing complaint data page, accessible at:

https://www.hud.gov/program_offices/fair_housing_equal_opp/fair_housing_data

HUD publishes summary data in Excel format, broken down by complaint basis, issue type, closure reason, and geography (state). The annual summary tables are the easiest entry point for aggregate analysis. Complaint-level data with individual case records is available through FHEO annual reports, which HUD submits to Congress and posts on the HUD website.

The FHEO annual reports are the most analytically useful published documents. Each report includes tables showing complaint volume by basis, by issue type, by respondent category, by state, and by closure reason — broken down by fiscal year. Reports from fiscal year 2001 through the present are available on the HUD website, providing a two-decade longitudinal dataset. The tables are published in PDF with underlying data extractable via standard PDF-to-table tools.

The most detailed public data comes through FOIA. Researchers and journalists have obtained complaint-level records — one row per complaint, with individual case identifiers, filing dates, closure dates, basis, issue, and relief information — through FOIA requests to FHEO. The complaint-level file is not routinely published as a bulk download, unlike HMDA, but it has been released in response to specific FOIA requests and is available through sources including the National Fair Housing Alliance, academic datasets, and journalism investigations.

Three research use cases

Geographic concentration of race-based complaints

State-level complaint counts, available in HUD's published tables, can be normalized to population to produce per-capita race-based complaint rates by state. This analysis consistently reveals geographic concentration: a small number of metropolitan areas — Chicago, Detroit, Memphis, Baltimore, Philadelphia — account for a disproportionate share of race-based housing complaints relative to their population. These are metropolitan areas with documented histories of racial residential segregation, large Black urban populations adjacent to white suburban areas, and active nonprofit fair housing testing organizations that generate complaints.

The per-capita normalization matters because larger states naturally generate more complaints in absolute terms. When complaint rates are computed per 100,000 Black or minority residents rather than per total population, the geographic pattern sharpens: Midwestern cities with high levels of residential segregation (dissimilarity index above 0.70 in Census data) cluster at the top of the per-capita complaint rate distribution for race-based cases. This correlation between segregation and complaint rate is not proof that more discrimination occurs in those markets; it may reflect differences in complainant awareness, FHAP agency capacity, or legal aid availability. But the geographic signal is strong enough to direct investigative attention.

Lender vs. landlord breakdown by complaint basis

The respondent type field enables a breakdown of complaint volume by the category of housing actor alleged to have discriminated. For race-based complaints, individual landlords and corporate property management companies account for the majority of respondents, with mortgage lenders and HOAs each accounting for smaller shares. For disability-based complaints, the pattern differs: corporate property management companies dominate (ESA and accessibility policy denials affect managed rental properties disproportionately), with HOAs — which set community-wide accessibility and pet policies — accounting for a rising share of complaints.

Lender complaints are the most interesting for cross-reference purposes. A fair housing complaint against a mortgage lender for race or national origin discrimination in financing — refusal to lend, discriminatory loan terms, or predatory targeting — maps to the same institution visible in HMDA data. The lender's LEI or FDIC certificate number in the HMDA dataset can be matched to the complaint record by institution name and date range. This cross-reference identifies lenders who simultaneously show elevated racial denial rate ratios in HMDA data and face fair housing complaints alleging discriminatory financing.

Conciliation rate analysis: which bases receive monetary relief most often

Not all complaint bases are equally likely to result in relief. Computing the conciliation rate — the share of closed complaints resulting in any form of relief — by basis reveals a consistent pattern. Disability cases involving reasonable accommodation failures have among the higher relief rates because the legal standard is clear and the remedy is often operational (modify the policy, allow the accommodation) rather than requiring a finding of discriminatory intent. Race cases, which generally require evidence of intentional discrimination or a disparate impact showing, have lower relief rates. National origin cases, which often involve language barriers in landlord-tenant communications and are easier to document, fall between race and disability in the conciliation rate distribution.

Monetary relief amounts, where disclosed in the FHEO annual reports, also vary substantially by basis and respondent type. Individual landlords in conciliation settlements tend to pay smaller amounts than corporate property management companies or lenders. The largest monetary conciliation agreements in the residential rental space involve large multifamily operators facing pattern complaints about accessibility failures in their building portfolios — cases where the relief includes both backpay to individual complainants and structural retrofits to multiple properties.

Limitations

The FHEO complaint database has four limitations that bound what conclusions can be drawn from it:

First, the administrative complaint record is a severe undercount of actual housing discrimination. National fair housing testing studies — in which paired testers (one minority, one white) inquire about the same available unit — have consistently found discrimination rates far higher than complaint rates would suggest. The audit studies find discriminatory treatment in 20 to 25 percent of rental inquiries in some markets; the complaint rate implies discrimination at a tiny fraction of that frequency. The gap reflects the difficulty of recognizing discrimination when it occurs, the cost of filing a complaint, and the widespread practice of subtle differential treatment that is difficult to document.

Second, a complaint is not a violation. Only approximately 15 percent of HUD complaints result in any relief for the complainant. HUD closes more than 60 percent of complaints with a no-reasonable-cause determination. This does not mean those complainants were not discriminated against; it means HUD's investigation, constrained by the evidentiary standards for administrative proceedings and the resources available to FHEO investigators, did not produce sufficient evidence to support a charge. The no-reasonable-cause closure rate is partly a reflection of investigative capacity, not just the underlying merit distribution of complaints.

Third, the database has respondent-selection bias. A complainant can only file a complaint against a respondent they can identify — a specific landlord, a specific lender, a specific HOA. Discrimination by actors who conceal their identity, who operate through intermediaries, or who implement discriminatory policies through facially neutral screening criteria (criminal background checks, income-to-rent ratios, credit score minimums) that have disparate impact on protected classes appears in the complaint database only if a complainant connects the outcome to discriminatory motivation. Structural discrimination is systematically undercounted relative to interpersonal discrimination.

Fourth, the geographic resolution of the publicly published data is at the state level, not the metropolitan area or census tract level. Complaint-level data with finer geographic resolution requires FOIA requests. The state-level aggregation masks the extreme metropolitan concentration of complaint volume described in the geographic analysis use case above.

Cross-references: HMDA, DOJ, and CFPB

The FHEO complaint database reaches its analytical potential when read against three adjacent federal datasets.

The HMDA loan-level dataset from the CFPB provides mortgage application records for every lender above the reporting threshold, including action_taken(originated, denied, withdrawn), derived_race, census_tract, and the institution's Legal Entity Identifier. Combining census-tract-level HMDA denial rate disparities by race with the geographic distribution of HUD race-based lending complaints identifies markets and institutions where both the statistical signal and the complaint volume are elevated simultaneously. The two signals are methodologically independent — HMDA denial rates do not depend on complainant behavior — making their convergence more meaningful than either in isolation.

The DOJ Housing and Civil Enforcement Section maintains a public database of settlements and consent orders from pattern-or-practice fair housing cases at justice.gov/crt/housing-and-civil-enforcement-section-cases. These cases represent the cases that HUD referred to DOJ or that DOJ initiated independently — the cases with the strongest evidence and the largest affected populations. Matching DOJ settlement geographies against FHEO complaint concentration and HMDA denial rate disparities identifies which markets have attracted the full stack of enforcement attention. Markets with all three signals — high complaint rate, elevated HMDA denial rate ratio, and DOJ enforcement history — are the places where housing discrimination has been most systematically documented.

The CFPB's fair lending enforcement actions under the Equal Credit Opportunity Act (15 U.S.C. § 1691 et seq.) cover discrimination in credit, including mortgage lending. CFPB consent orders — publicly available at consumerfinance.gov/enforcement/actions, filterable by statute — name specific lenders and sometimes specific geographic markets. These actions frequently cite HMDA data in the statement of facts, and the named lenders appear in both HMDA records and, for any lender also facing a HUD complaint, in the FHEO complaint database.