💰 quantitative finance

Measuring Racial Disparities in Rent Growth Under Algorithmic Landlord Concentration in U.S. Metros

This study provides the first tract-level evidence that corporate landlord concentration is associated with disproportionately higher rent growth in majority-minority neighborhoods compared to white neighborhoods across ten U.S. metropolitan areas.

Original authors: Advay Ranade

Published 2026-06-29

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Advay Ranade

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: The "Price-Fixing" Game

Imagine you are trying to buy a pair of sneakers. If five different stores sell the exact same shoe, you expect them to compete with each other to get your money. But what if those five store owners all use the same secret computer program that tells them, "Don’t lower your price. Keep it high. We are all in this together."

That is essentially what the U.S. Department of Justice accused five massive apartment companies (called REITs) of doing. They used a software called RealPage to coordinate their rent prices instead of competing. The government says this kept rents artificially high.

This paper asks a specific, deeper question: Does this "price-fixing" hurt everyone equally, or does it hurt certain neighborhoods more than others?

The Detective Work: How They Measured It

To answer this, the researcher, Advay Ranade, had to play detective. He couldn’t just ask the companies what they were doing, so he used public records.

Finding the "Big Landlords": He looked at official government filings (SEC 10-Ks) from the five accused companies. These filings list every single apartment building they own. He mapped these buildings to specific neighborhoods (census tracts) to see where these "corporate landlords" were most concentrated. He calls this measure CLC (Corporate Landlord Concentration).
Measuring Rent Growth: Instead of using slow government surveys that average data over five years (which blurs the details), he used Zillow’s rent data. This is like using a live stock ticker instead of a yearly report—it shows exactly what rents were doing in real-time between 2019 and 2023.
Controlling for "Pre-Existing Stress": He knew that big landlords might just pick neighborhoods that were already expensive or crowded. To make sure he wasn’t just measuring that, he created a new score called the AHBI (Algorithmic Housing Burden Index). Think of this as a "stress test" for a neighborhood before the algorithmic pricing started. It measures how tight the housing market was and how much rent burden people already had.

The Findings: What Happened?

1. The General Effect (H1)
The study found that in neighborhoods where these corporate landlords were heavily concentrated, rents grew faster. Specifically, if you doubled the presence of these corporate landlords in a neighborhood, rent growth was about 2.8 percentage points higher than in neighborhoods with fewer corporate landlords.

2. The Racial Disparity (H2) – The Core Finding
This is the most important part of the paper. The researcher looked at whether this rent hike affected White neighborhoods and Minority neighborhoods (defined as neighborhoods where Black and Hispanic residents make up more than 50% of the population) differently.

In White Neighborhoods: Surprisingly, having more corporate landlords was actually associated with slightly lower rent growth within the same city. The paper suggests these are often newer, upscale suburban complexes where the market dynamics are different.
In Minority Neighborhoods: The trend flipped. In these neighborhoods, higher corporate landlord concentration was linked to significantly higher rent growth.

The Analogy: Imagine two runners in a race. One runner (White neighborhoods) is on a smooth, paved track. The other runner (Minority neighborhoods) is on a track with hidden hurdles. When the "algorithmic pricing" whistle blows, the runner on the smooth track doesn’t slow down much. But the runner on the hurdle-filled track gets tripped up harder, causing their "rent burden" to spike much higher.

The data showed that within the same city, minority neighborhoods with high corporate landlord presence saw rent growth that was 5.9 percentage points higher than comparable White neighborhoods.

3. The AI Check (H3)
To make sure this wasn’t a fluke or a mistake in the math, the researcher used an Artificial Intelligence model (XGBoost). He fed the AI all the neighborhood data from 2019 and asked it to predict rent growth for 2023.

The AI didn’t know about the "racial disparity" hypothesis; it just looked for patterns. The AI independently confirmed the finding: it predicted that corporate landlord concentration would push rents up in minority neighborhoods and down (or have less effect) in White neighborhoods. This gives the researchers high confidence that the pattern is real and not just a statistical accident.

Why Does This Happen?

The paper suggests that minority neighborhoods are more vulnerable to this algorithmic pricing for a few reasons:

Fewer Alternatives: Renters in these areas may have fewer affordable places to move to if their rent goes up.
Existing Stress: These neighborhoods often already had higher "housing stress" (the AHBI score), meaning residents had less financial cushion to absorb price hikes.
Less Competition: There may be fewer small, independent landlords to compete against the big corporate algorithms.

The Bottom Line

This paper provides the first neighborhood-level evidence that the use of algorithmic rent pricing by large corporate landlords is not a neutral event. It appears to disproportionately drive up rents in communities of color, potentially widening the racial wealth gap.

In short: The "secret computer program" used by big landlords doesn’t just raise rents for everyone; it raises them more for minority neighborhoods, likely because those neighborhoods are already more vulnerable and have fewer options to escape the price hikes.

Technical Summary: Measuring Racial Disparities in Rent Growth Under Algorithmic Landlord Concentration in U.S. Metros

Problem Statement
The paper addresses the gap in existing literature regarding the distributional effects of algorithmic rent pricing. While prior research (e.g., Calder-Wang and Kim, 2024; White House Council of Economic Advisers, 2024) has established that algorithmic pricing software (such as RealPage’s YieldStar) raises rents above competitive levels, it remains unknown whether this burden falls disproportionately on racial minorities. Specifically, the study investigates whether census-tract-level corporate landlord concentration (CLC) is associated with rent growth between 2019 and 2023, and whether this association is significantly stronger in majority-minority neighborhoods compared to majority-white neighborhoods within the same metropolitan areas.

Methodology
The study employs a multi-method approach combining econometric regression and machine learning, utilizing novel data construction techniques:

Data Construction:
- Corporate Landlord Concentration (CLC): The primary independent variable is constructed from SEC EDGAR 10-K Schedule III filings for five major residential REITs named in the 2024 DOJ antitrust complaint (Avalon Bay, Equity Residential, Essex Property Trust, Mid-America, and UDR). Property addresses and unit counts from fiscal years 2016–2022 were geocoded to 2020 census tracts. CLC is defined as the share of renter-occupied units owned by these REITs in a given tract.
- Algorithmic Housing Burden Index (AHBI): A novel composite index created to control for pre-existing market vulnerability. It combines two American Community Survey (ACS) measures: Pre-existing Rent Burden (share of renters paying ≥30% of income on rent) and Housing Market Tightness (complement of vacancy rate). Both components are standardized as z-scores and averaged.
- Rent Growth Outcome: To avoid the smoothing effects of ACS 5-year estimates, the study uses the Zillow Observed Rent Index (ZORI), a transaction-based monthly rent dataset. ZIP-code-level ZORI data (2019–2023) was crosswalked to census tracts using HUD USPS allocation factors.
Empirical Strategy:
- H1 (Average Effect): An OLS regression estimates the relationship between 2019 CLC and 2019–2023 rent growth, controlling for AHBI, demographic composition, and market characteristics. This specification excludes metro fixed effects to capture between-metro variation.
- H2 (Racial Heterogeneity): An OLS regression with metro fixed effects tests the interaction between CLC and a majority-minority indicator (combined Black and Hispanic share >50%). This isolates the within-metro effect to determine if CLC impacts rent growth differently based on racial composition.
- H3 (Machine Learning Validation): An XGBoost model is trained on 2019 tract-level features to predict out-of-sample rent growth. SHAP (SHapley Additive exPlanations) analysis is used to decom feature contributions, providing a non-parametric check on the racial disparity findings.

Key Contributions
The paper claims four primary contributions:

Data Pipeline: Introduction of a publicly replicable method for measuring algorithmic landlord concentration at the census tract level using SEC EDGAR filings, a data source previously unused for neighborhood-level housing analysis.
Index Construction: Creation of the AHBI, a generalizable composite measure of housing vulnerability derived from ACS data.
Empirical Evidence: Provision of the first tract-level estimates suggesting that corporate landlord concentration is associated with disproportionately higher rent growth in communities of color.
Methodological Insight: Demonstration that ACS 5-year estimates are insufficient for detecting acute rent growth due to multi-year smoothing, necessitating transaction-based data like ZORI.

Results

H1 (Overall Effect): Doubling REIT concentration in a census tract is associated with approximately 2.8 percentage points higher rent growth (2019–2023). This result is statistically significant under HC1 robust standard errors ( $p = 0.030$ ) and marginally significant under metro-clustered standard errors ( $p = 0.086$ ). The inclusion of AHBI as a control is critical, as pre-existing housing stress is positively associated with rent growth.
H2 (Racial Disparity): The effect of CLC on rent growth is significantly larger in majority-minority tracts. Within the same metropolitan area, high-CLC majority-minority tracts exhibit 5.9 percentage points higher rent growth than comparable majority-white tracts ( $p = 0.039$ ). The interaction term remains significant when using a continuous minority share variable ( $p = 0.002$ ). Notably, the within-metro effect of CLC is negative in majority-white tracts ( $\beta_1 = -0.034$ ) and positive in majority-minority tracts ( $\beta_1 + \beta_3 = +0.024$ ).
H3 (ML Validation): The XGBoost model predicts 44% of the out-of-sample variance in rent growth ( $R^2 = 0.439$ ). SHAP analysis confirms that CLC ranks fifth among eight features in importance. Crucially, CLC’s predicted contribution to rent growth is positive in majority-minority tracts (mean SHAP value = +0.0057) and negative in majority-white tracts (mean SHAP value = -0.0011), independently corroborating the OLS interaction results without imposing functional form assumptions.

Significance and Implications
The paper argues that these findings provide the first tract-level evidence consistent with corporate landlord concentration driving disproportionately higher rent growth in communities of color. The authors suggest that algorithmic pricing imposes coordinated price pressure that minority renters are less able to resist due to lower tenant mobility, higher pre-existing rent burden, and fewer affordable alternatives.

The study highlights implications for:

Antitrust Enforcement: Suggesting that competition policy should consider distributional impacts and racial equity, not just aggregate consumer welfare.
Fair Housing Law: Noting that the documented disparity is consistent with a disparate impact theory under the Fair Housing Act.
Regulatory Oversight: Advocating for greater transparency, such as requiring REITs to disclose revenue management software usage in 10-K filings, to enable real-time monitoring of algorithmic pricing effects.

The authors maintain a modest tone regarding causality, noting limitations such as the cross-sectional design, the small number of metropolitan clusters, and the potential for unobserved confounders. They recommend future work using instrumental variables and expanding the geographic scope.

The Big Picture: The "Price-Fixing" Game

The Detective Work: How They Measured It

The Findings: What Happened?

Why Does This Happen?

The Bottom Line

More like this