Rent-setting algorithms require targeted regulation, not new laws

The use of algorithms in housing rental markets to help assess and determine market pricing has come under increased scrutiny, with both federal regulators and state assemblies taking action. While many fear these services will inevitably lead to collusion that only raises rents, a closer look shows algorithmic pricing is not the boogeyman many have portrayed. Whether algorithms benefit or harm consumers depends on the data they use and the goals of the software.

A recent settlement between the Department of Justice (DOJ) and RealPage, the largest firm in the rent-setting algorithm market, shows that the federal regulatory system is working to preserve the positive features of these tools while disallowing the use of certain data that may facilitate price collusion between landlords.

Several states have also acted, either passing or considering legislation with goals similar to the antitrust settlement. However, many of these bills prove irrelevant at best and harmful at worst when considered alongside the federal settlement. A better approach would be to focus on enforcing existing federal regulations instead of layering new laws that add many costs for landlords and few benefits for renters.

Algorithmic rent-setting in practice

Algorithmic pricing uses computer software to recommend a price for a product based on user data, economic variables, and other information. While the rise of artificial intelligence (AI) has brought greater public attention to the practice, algorithmic pricing has been a part of many markets for decades, including airfare, ridesharing, and e-commerce.

In the rental market, companies like RealPage and Yardi use macroeconomic variables and property-level data to generate real-time, algorithmically recommended rents for each unit in a property. Landlords can choose to implement these prices or ignore them. Property occupancy, historical price data, and lease length are examples of data points used to understand each property’s internal supply-and-demand dynamics and ultimately inform final price recommendations. Current estimates find that 10.4% of rentals in the United States use algorithmic pricing, with users skewing towards larger multifamily buildings. An estimated 30% of buildings with 20 or more residential units use algorithmic pricing. In an estimated 90% of cases, landlords accept and implement the price recommendation generated by the algorithms.

Researchers and regulators separate the potential effects of rent-setting algorithms into two categories. First, algorithms make rents more responsive to changing market conditions. More efficient price-setting benefits the market overall. Like any other product or service, markets function best when landlords can quickly learn from and act on public information. For example, in 2008, when housing markets were in a nationwide downturn, properties that used algorithms had lower rents and higher occupancy rates. Algorithms quickly updated their recommendations based on new macroeconomic information, which resulted in lower prices. When the housing market recovered, the prices recommended by algorithms were higher than those charged in units that were not using them.

As this example shows, the proper use of algorithms is not biased toward either higher or lower rents; it simply helps landlords and renters find the right apartments and prices more quickly and efficiently. A 2026 study from the University of Pennsylvania and the National Bureau of Economic Research (NBER) confirmed that RealPage had this effect, making the discovery of market information and the setting of prices more efficient.

However, the second category of effects, algorithmic use of nonpublic information from individual landlords, has raised concerns about collusion. This data includes but is not limited to rents, leases, and occupancy rates for specific units. In effect, algorithm users could indirectly set rents informed by confidential information from their competitors.

The NBER/UPenn study found evidence that prices resulting from using an algorithm were consistent with “joint profit maximization,” or collusion among landlords, because of this shared nonpublic data. Across the 4.2 million rental units observed, this resulted in an estimated $25 increase in monthly rents. Concerns about this aspect of algorithmic pricing were the prime motivator of the 2024 DOJ antitrust suit against RealPage.

DOJ and RealPage reach a settlement

In August 2024, the DOJ, along with several states, sued RealPage, alleging the company’s software allowed price coordination between landlords. The suit focused on the use of nonpublic data to generate frequent price recommendations for participating landlords and algorithmic features that were designed to asymmetrically push prices up, even when they were not optimal. In November 2025, RealPage agreed to remove features that limit price decreases and to stop using current and granular competitive data to recommend rent prices.

The remedies agreed to by RealPage and the DOJ do not reject all algorithmic pricing as anticompetitive. RealPage continues to offer an algorithm that incorporates a landlord’s own property data, macroeconomic data, other public information, and some aged competitor data. The settlement seeks to preserve the efficiency-enhancing information properties of rent-setting algorithms while removing data and features that might aid collusion.

Reaching this settlement is evidence that the antitrust infrastructure already in place is working. While state-level challenges remain and are being resolved, the bottom line is that the DOJ and RealPage agreed these solutions are adequate to address antitrust concerns. A further question is whether additional legislation is needed.

State-level legislative efforts

To signal action on housing affordability, many states are now introducing their own legislation to restrict algorithmic rent-setting. Three states—New York, New Jersey, and California—have already enacted such laws, and states such as Maryland and Massachusetts are currently considering them.

Table 1: Enacted legislation as of October 2026

State	Law & year	Summary
New York	S. 7882 (2025)	Adds a new “agreement to not compete” provision covering residential rentals. It defines a “coordinating function” as any software or algorithmic device that (1) collects historical or current rent, supply, or lease-date data from two or more unaffiliated landlords, (2) processes that pooled data computationally (including training an algorithm), and (3) recommends rental prices, lease terms, occupancy levels, or renewal terms back to a landlord.
California	AB 325 (2025)	Makes it unlawful to use or distribute a “common pricing algorithm,” a tool used by two or more parties, relying on competitor data, to recommend or set prices or commercial terms, as part of a conspiracy to restrain trade. Also makes it unlawful to use or distribute such an algorithm if the operator coerces another business into adopting its recommended price. Unlike New York’s law, AB 325 isn’t rent-specific; it applies to pricing algorithms across any industry, with rental housing as one example legislators cited.
New Jersey	A3497 (2026)	Makes it unlawful for a landlord to pay for or use a “coordinator,” algorithmic revenue-management software or a device performing a “coordinating function,” where that function is defined to include collecting competitively sensitive data (prices, occupancy, lease terms) from multiple landlords to train a pricing algorithm, or setting/recommending the same or substantially similar prices to two or more landlords via the same underlying algorithm even without shared data, if it facilitates “parallel pricing coordination.”

Table 2: Examples of state-level legislation currently being considered   

State	Law & year	Summary
Maryland	HB 434 (2026)	Would bar a landlord from using an “algorithmic device” that uses, incorporates, or was trained on “nonpublic competitor data,” which is defined as rent prices, occupancy, or lease dates from another landlord in the same or a related market, to set rent, renewal terms, occupancy levels, or other lease conditions.
Massachusetts	H. 5222 (2026)	Would forbid any landlord from contracting with any outside company that collects rent and lease data from multiple outside properties, runs it through software, and produces a price recommendation, regardless of whether this data is public or private.

All the state bills overlap substantially with the terms of the antitrust settlement between RealPage and the DOJ. Nevertheless, they raise concerns. First, with federal regulators already on the case, even small variations between bills risk creating a regulatory patchwork across states that would raise compliance costs for landlords and algorithm providers without providing any additional benefit to consumers. For example, states differ on whether using nonpublic data, or any competitor data at all, constitutes coordination.

Second, by using overbroad terms and definitions, state bills risk banning algorithmic rent-setting that facilitates the efficient flow of market information but does not assist in collusion. A primary example is the use of older competitor data. During its settlement, the DOJ determined that algorithms may still use nonpublic data if it is at least 12 months old and not part of an active lease. However, all five state bills in the tables below provide no similar carve-out and ban the use of all such data. These provisions will add uncertainty and complexity to rental markets.

Each bill contains different provisions and defines terms differently. In Massachusetts, for example, “coordinate” is defined as collecting data on historic or current rent prices, supply levels, occupancy rates, or lease dates; analyzing this data through software; and recommending a rental price. This definition makes no distinction between public and nonpublic data, capturing virtually all versions of algorithmic pricing software, regardless of any connection to price collusion. This definition also deviates from most other states, which specify the non-public nature of the data they target.

The differences between laws don’t end with definitions. They also vary substantially in which activities they target. Some states, like New Jersey, directly address “parallel price coordination,” while Maryland, for example, targets the training of models on nonpublic data. These differing provisions across state bills range from redundant to overly broad.

Fragmented, vague legislation creates unnecessary risk and uncertainty about how platforms and users can operate without triggering legal penalties. To complicate things further, several city governments, like Bellingham, Wash., and Montgomery County, Md., are proposing to layer their own laws over state and federal laws. Even if these local rules are well-targeted, they risk creating an even more fragmented regulatory landscape.

Instead of adding additional laws and regulations on an increasingly granular scale, focusing on the use of nonpublic data while relying on the existing legal code can achieve the same goal of market fairness directly. As the RealPage lawsuit demonstrated, a robust, nationwide legal infrastructure already exists to address antitrust concerns. Price collusion is already illegal, and the system already addresses instances when it occurs. Relying on these existing, functioning legal frameworks is a clearer path to enabling both innovation and accountability, without creating a patchwork of state-specific laws that tech companies and landlords must now navigate.

The post Rent-setting algorithms require targeted regulation, not new laws appeared first on Reason Foundation.

Source: https://reason.org/commentary/rent-setting-algorithms-require-targeted-regulation-not-new-laws/