The Controversy Behind RealPage’s Rent Algorithm

AI-generated image of an apartment complex.

In August, . It’s been a long time coming. in May, identifying their software as a factor in increasing rent. I drew heavily on the original ProPublica piece on the company, originally published in 2022, and a look at their own site and marketing materials. I said it then, and I’ll say it now: When that it “outperforms the market by 2% — 5%,” that comes from renters.I’m thrilled to see the news about RealPage being sued by the DOJ.For the record, I am not a lawyer. I do not have the training or expertise to speak about the legal merits and flaws of the DOJ-filed suit. However, I am an IT professional with 20+ years of experience and have degrees in economics. I may or may not also have an educational 2×4 with a dictionary attached to it for those who misuse the term AI. So, despite my admitted bias, when I see that RealPage’s defense is that it’s just software and the real problem is housing affordability, I’m willing to give them the benefit of the doubt.

The Suit Itself

Again, IANAL, but the opening of the complaint makes it clear that the heart of the DOJ’s case isn’t the algorithm, but the actual business practices at RealPage. Even if they later feed that data into an algorithm, collecting and using non-public data on the rental market is where the collusion happens.When I touched on this in May, I discussed how in a market without RealPage, landlords would look at rental prices for other, similar properties to help gauge what range they should use to set their rental price. And if that’s what RealPage was doing, just on a larger scale, I would look askance at the DOJ’s case. Automating a standard business practice this isn’t. :

Landlords, who would otherwise be competing with each other, submit on a daily basis their competitively sensitive information to RealPage. This nonpublic, material, and granular rental data includes, among other information, a landlord’s rental prices from executed leases, lease terms, and future occupancy. RealPage collects a broad swath of such data from competing landlords, combines it, and feeds it to an algorithm.

That, combined with RealPage’s dominant market position — they control 80% of the market for rent-setting software, — is what makes this so dangerous. They really do have access to enough private data to fuel an algorithm to make recommendations. I don’t know if collecting that data and disseminating it reaches the level of collusion. Still, I do know as an economist that that level of market control puts a company in the driver’s seat.This becomes even more concerning when combined with RealPage’s push to automate their pricing recommendations, which the DOJ describes as:

AIRM and YieldStar provide daily price recommendations. RealPage has taken multiple steps to increase compliance with AIRM and YieldStar price recommendations. It designed AIRM and YieldStar to make it much easier to accept recommendations than to decline them. It built an auto-accept function and pushes clients to adopt it and increase its role. And its pricing advisors encourage landlords to follow AIRM and YieldStar pricing recommendations. Among their duties, pricing advisors review any request to override a price recommendation.

The business case for this is likely to help RealPage further tune its model, but the practice the DOJ describes is somewhat concerning. Rejecting the recommended increase requires a “strong sound business-minded approach”. That rejection and its reasoning are forwarded to a RealPage pricing advisor for the property. If the pricing advisor disagrees, they can escalate to a property manager’s regional manager. Not accepting the recommended price — which changes daily — can be escalated by the software vendor up a property’s management chain.

In my 20+ years in IT, I have been in-house, I have been external, I have been a consultant, and I have been a vendor. I have never, not once, in any of my roles, had the appropriate course of action been for a vendor to escalate an issue on how the software is used to a more senior manager. If that’s accurate, it’s mind-blowing.

Private Lease Data

If you’ve ever rented an apartment or a house, you know that a leasing agreement is a complex document. The leasing contract protects you, the tenant, as well as the landlord, by clearly defining what is expected of each party under the terms of the lease, from rent to fees, to repairs. It covers the security deposit, the late fee, and the date that rent is considered late, the pet fee tacked onto your rent each month for your four-legged companion. It also covers anything you may have negotiated for yourself, such as waiving those fees or guaranteeing that the carpet will be replaced in the bedroom, or the right to paint the walls. It also contains tenant demographics, income statements, and credit reports.Of all the information in private lease data that could influence a financial decision, which is more likely: the presence of an adorable puppy named Cookie or a 30% debt-to-income ratio?

No idea if their name is Cookie or not. It could be!
Photo by Harriet B. on

boils down to saying that this ‘private leasing data’ isn’t what drives their algorithm, but non-demographic, non-personal data focused on supply and demand. This combination of information means they don’t always recommend rent increases, and even when they do, the landlord can always turn them down.If this is the case, then the facts of the DOJ’s suit are wrong. I think this should be fairly easy to prove: show the courts the underlying models used by their software aren’t reliant on data gleaned from its dominant market position in revenue management software. Despite their efforts to obscure the issues, this is something they could do to rebut the DOJ’s arguments. Given their public stance is to negotiate to settle the suit, I don’t believe they can prove the suit is without merit.

Automating Standard Business Practices

I said this in my initial review of RealPage’s business practices and I’ll say it now, in bold: a landlord’s increased profits come from renters. Unless the rental property just became a small country and began using its own currency, they are bound by the same calculus that constrains every other business. If a business wants to increase how much it makes in profit, it must do some combination of reducing costs and increasing prices.

For a rental property, those opportunities to reduce costs come with their own downsides. Rental properties have a high fixed cost and not a lot of variable costs, Those variable costs are focused on management and maintenance. Skimping on maintenance or management can come with legal repercussions. Even skimping on things like landscaping can contribute to making the rental property less attractive to tenants.For a rental property to outperform competitors year-over-year, those austerity measures, if any, would need to be aided with increases in revenue — an increase in rent.Whereas the normal forces of a competitive market constrain rent increases, RealPage’s algorithm essentially gives the landlord an information advantage that they can use to increase rents more than would be expected in a competitive market.Want an example? Here’s one that holds to the rules RealPage says they work by:ABC Apartments is across the street from the XYZ Complex. Both are about the same age, are well-maintained, offer similar amenities, have similar apartments for rent, and are located right by each other. Given their similarities, ABC and XYZ probably charge approximately the same amount of rent. When ABC is reviewing how much it should charge for its 2-bedroom 2-bath units, the presence of a competitor like XYZ will keep the rent increase, if any, reasonable for the market. Maybe inflation and rising costs necessitate increasing the rent from $1000 a month to $1025, a 2.5% increase. That makes sense and it’s right in line with the $1015 a month XYZ charges for the same unit.Enter RealPage. ABC is using RealPage, as is XYZ. RealPage knows what ABC doesn’t — what it can’t know: all of XYZ’s 2-bedroom 2-bath units are occupied and under lease for at least the next nine months. So, RealPage recommends increasing the rent on 2-bedroom 2-bath units to $1100 a month. Maybe ABC hesitates because that’s way more than they’ve ever done before and it seems steep for the market. Maybe they only increase it to $1075 instead. On a 12-month lease, that’s an extra $600 per unit. The recommended rent is an extra $900 a year per unit.None of XYZ’s data was exposed by RealPage when making this recommendation. Nor was any of ABC’s data. Yet, it still used RealPage’s access to confidential data to make recommendations. , you don’t need a PhD in Economics to understand how this can be to the landlord’s benefit. Nor do you need one to see that that level of information access could be doing real economic harm to tenants.

Technology Isn’t On Trial

Very little of the DOJ’s 115-page filing is focused on RealPage’s use of machine learning itself. Despite alarmists’ cries to the contrary, this suit is not about pushing data science and AI out of the marketplace. Frankly, I think you could remove every single mention of generating recommendations more advanced than manually inputting data into an Excel spreadsheet and the case against RealPage doesn’t change. The DOJ isn’t suing because RealPage did something novel with technology. RealPage’s work to collect confidential information from its customers and use that to generate pricing recommendations, which it then pushes its customers to adopt automatically is the problem. The algorithm is just a tool that serves that purpose. The DOJ is suing because RealPage’s business practices are anti-competitive and harmful to renters.I’m looking forward to what happens next.

References

Apartment Revenue Management Software | RealPage. . Accessed 1 Sept. 2024.

Board, The Editorial. “Opinion | The New Housing Inflation Scapegoat.” WSJ, 4 Sept. 2024, .

“DOJ Files Antitrust Lawsuit against RealPage.” RealEstateNews.Com, 23 Aug. 2024, .

Gregg, Aaron, and Eva Dou. “Why Is Rent so High? The Justice Department Blames a Tech Firm’s Algorithm.” Washington Post, 23 Aug. 2024, .

How a Secret Rent Algorithm Pushes Rents Higher — ProPublica. . Accessed 1 Sept. 2024.

Peck, Emily. “How RealPage Transformed the Apartment Rental Market.” Axios, 29 Aug. 2024, .

“RealPagePublicPolicy.” Your Site Title, . Accessed 10 Sept. 2024.

“Statement from RealPage: Addressing Recent Allegations.” RealPage Blog, . Accessed 1 Sept. 2024.

Originally published at on October 21, 2024.