Investigation of Alleged Rental Price-Fixing via “Algorithmic Collusion” on RealPage and Other Revenue Management Software

Problem & Motivation

In October 2022, ProPublica reported allegations that RealPage’s Revenue Management software contributes to rising rental prices by facilitating algorithmic collusion among landlords. This alleged practice could effectively replace explicit price coordination with implicit, algorithm-driven price increases. Such price hikes significantly impact renters’ affordability, reduce real incomes, increase housing instability, and disproportionately affect landlords not using similar technologies. Given the stakes for renters and landlords alike, nonpartisan research is critical for informing ongoing antitrust litigation and regulatory scrutiny.

Data Source & Data Science Approach

The investigation employed publicly accessible data from Zillow’s rent indices, the U.S. Census Bureau, RealPage's client disclosures, and web-scraped data from property management websites listed in lawsuits. The team used:

• Web scraping techniques, including BeautifulSoup and Selenium, to compile a robust dataset of rental properties.
• Fuzzy matching algorithms to accurately pair properties across various datasets.
• A Propensity Score Matching (PSM) model to compare similar properties using and not using RealPage software.
• Predictive modeling methods, including Random Forest, Gradient Boosting, and Neural Networks, to further assess rental price differences.
• Difference-in-Difference (DiD) and Synthetic Control methods to evaluate the impact of RealPage's 2017 merger with Lease Rent Options (LRO).

Evaluation

The PSM model demonstrated a statistically significant association between RealPage usage and higher rents, estimating an average increase of $0.23 per square foot monthly. Predictive modeling via a Neural Network confirmed this price differential, though Mann-Whitney U tests revealed non-significant distribution differences.

However, comprehensive analyses using DiD and Synthetic Control methods found no consistent, statistically significant evidence of rent increases specifically due to RealPage’s acquisition of LRO. Although isolated cases, such as Phoenix, indicated possible increases, these findings were inconsistent across regions and methodologies.

Key Learnings & Impact

This research confirmed a meaningful correlation between RealPage usage and higher rents, providing evidence of increased rental costs associated with algorithm-driven pricing. Despite these findings, conclusive evidence of deliberate price coordination due to RealPage’s market consolidation via the LRO merger was lacking.

The project's significant contributions include:

• Providing robust, publicly accessible data for legal and regulatory purposes.
• Delivering insights via an interactive dashboard, allowing policymakers and the general public to explore RealPage's influence independently.
• Highlighting the complexities surrounding algorithmic pricing and its regulatory implications, reinforcing the need for further transparency from RealPage and further investigative research.

This work aims to aid policymakers, legal professionals, housing activists, and researchers by enhancing transparency, supporting informed decision-making, and strengthening antitrust evaluations.

Acknowledgements

We acknowledge Timothy Majidzadeh for his outstanding leadership, management, and expertise during this project, Chelle for her visionary website architecture and commitment to a quality final product. We acknowledge Ahmad Allaou and Patrick Yim for their tireless pursuit of great data, great models,  and great analysis. Finally, we acknowledge Peter Benzoni for his flexibility and ingenuity in filling the gaps throughout the project and crafting compelling narratives. Special thanks to Econ Consulting and academic contacts for providing crucial feedback. We likewise acknowledge the Washington Post, ProPublica, and Zillow for providing meaningful data and background research