Economist: Are American rents rigged by algorithms? That is what Department of Justice prosecutors allege

Imagine that you are about to enter a room with a group of nine other people. You will display a number—any between, say, 2,500 and 3,000. Once the group enters the room other players will start to come in. Each will choose one of your group, picking the lowest number. You do not know how quickly or slowly the other players will trickle in to pick from the group. What is the highest number you can display while still getting picked quickly?

Now imagine that before entering the room eight of you submit your proposed numbers into a computer. It thinks for a moment, then tells you a number to pick: 2,850. You can choose a different one if you really want, but you will need to enter a reason for your decision in a text box.

In which scenario might the average number displayed by the group be higher? If you think it would be in the second then you are joined by the Department of Justice, which on August 23rd filed an antitrust complaint against RealPage, a property-software platform that landlords use to help them set rents for flats. According to the doj’s lawsuit, RealPage enables landlords to collude, pushing up rents on properties across America. RealPage says that its software is “built to be legally compliant” and notes it has worked with the doj to ensure this is the case.

Now is a fruitful time to be attacking firms for raising prices or keeping them high. Both Kamala Harris and Donald Trump, America’s presidential candidates, have alighted on the rising cost of living—of which rents and housing are perhaps the most important slice—as a crucial issue in the forthcoming election. Ms Harris wants to take aim at companies for price gouging. The doj has filed briefs and motions in a handful of class-action lawsuits brought by homeowners against estate agents (or realtors, as they are known in America) alleging anti-competitive behaviour.

The RealPage case is interesting because it targets a pricing algorithm, rather than landlords gathering in a smoke-filled room. RealPage’s commercial-revenue-management software has an 80% market share for multifamily housing rentals, which include apartment blocks and condo buildings. Landlords submit private information about their properties and the rents they command. The firm’s software then makes suggestions for rents.

In the gamified version, it seems like competitive forces might still be able to work (surely the incentive is to undercut the suggestion made by the computer, even by a little?). But prosecutors at the doj argue that the company makes it cumbersome to do this. Although landlords are not bound to do as the software says, the system makes it much easier to accept suggested rents than reject them. If a property manager is dealing with several properties she can accept all suggestions in bulk, but rejecting or overriding must be done one by one. If she chooses to override a suggested rent, the software generates a text box, asking for an explanation. That explanation is sent to a rep at RealPage. If he deems it insufficient, it can be escalated to the property manager’s supervisor.

This design, which the doj says gives RealPage the power to influence rents for most of the market, could mean that landlords are in effect banding together and keeping prices high, rather than competing. Perhaps the most compelling evidence of this is how the company and its users describe what is going on. RealPage has described itself as a tool that “helps curb [landlords’] instincts to respond to down-market conditions by either dramatically lowering price or by holding price when they are losing velocity and/or occupancy”. According to the doj, a RealPage executive said that if enough landlords used the firm’s software, they would “likely move in unison versus against each other”. A landlord even described one of RealPage’s products as “classic price fixing”.

The firm’s lawyer has said that the doj is cherry-picking quotes. It does not take a sophisticated algorithm to work out whether such comments will be good or bad for the firm’s case.