Company that makes rent-setting software for landlords sued for collusion

RealPage worked with some of the nation’s largest landlords to raise rents, says lawsuit.

Read the whole story

 

[rosen380]

They just gotta amend the algo to not use nonpublic data, or only use nonpublic data from the client. It could still help clients decide rents to charge.

Seriously, let the "private information" thing go. Private information is not the problem. The problem is landlords colluding with each other to force prices up instead of competing with each other (which would bring prices down). 100% of the information could be public and they would still be operating as a cartel.

The problem is both the private data and the algorithmically determined rent increases across the board.

Without the private data, the algorithm wouldn't be able to raise rents nearly as aggressively, as there would be no way to know whether any given landlord was going to go along with it. Without the algorithm, the landlords wouldn't be nearly as efficient at setting rents at just-below-too-high-for-the-market-to-bear. Without either, the landlords would essentially be looking at FOR RENT ads and making their best guess - which is what should be happening in this (and every other) market.

If you are just going to use an algorithm to tell you to raise rents, why even bother basing it on data? Just keep raising it and do "normal" collusion.

 

[Adreaver]

They just gotta amend the algo to not use nonpublic data, or only use nonpublic data from the client. It could still help clients decide rents to charge.

Seriously, let the "private information" thing go. Private information is not the problem. The problem is landlords colluding with each other to force prices up instead of competing with each other (which would bring prices down). 100% of the information could be public and they would still be operating as a cartel.

The problem is both the private data and the algorithmically determined rent increases across the board.

Without the private data, the algorithm wouldn't be able to raise rents nearly as aggressively, as there would be no way to know whether any given landlord was going to go along with it. Without the algorithm, the landlords wouldn't be nearly as efficient at setting rents at just-below-too-high-for-the-market-to-bear. Without either, the landlords would essentially be looking at FOR RENT ads and making their best guess - which is what should be happening in this (and every other) market.

If you are just going to use an algorithm to tell you to raise rents, why even bother basing it on data? Just keep raising it and do "normal" collusion.

Because that only makes SOME of the money, not ALL of the money.

This is late stage capitalism at it's finest - if some scheme doesn't make ALL the money it possibly can, regardless of who is hurt in the process, it's "financially unviable" and something that hurts more people and makes more money replaces it.

 

[ardent]

They just gotta amend the algo to not use nonpublic data, or only use nonpublic data from the client. It could still help clients decide rents to charge.

Seriously, let the "private information" thing go. Private information is not the problem. The problem is landlords colluding with each other to force prices up instead of competing with each other (which would bring prices down). 100% of the information could be public and they would still be operating as a cartel.

The problem is both the private data and the algorithmically determined rent increases across the board.

Without the private data, the algorithm wouldn't be able to raise rents nearly as aggressively, as there would be no way to know whether any given landlord was going to go along with it. Without the algorithm, the landlords wouldn't be nearly as efficient at setting rents at just-below-too-high-for-the-market-to-bear. Without either, the landlords would essentially be looking at FOR RENT ads and making their best guess - which is what should be happening in this (and every other) market.

If you are just going to use an algorithm to tell you to raise rents, why even bother basing it on data? Just keep raising it and do "normal" collusion.

That's what landlords did before this option was made available, and it was markedly less efficient at extracting rents. In the Washington metro area there's been a clear spike in rents in Virginia now that they have far more clarity on the rents in Maryland and DC. Rent on a 1200 square foot condo jumped almost $600/mo during 2020-2022 timeframe, and coincidentally the landlords got access to RealPage in those timeframes.

 

[rosen380]

They just gotta amend the algo to not use nonpublic data, or only use nonpublic data from the client. It could still help clients decide rents to charge.

Seriously, let the "private information" thing go. Private information is not the problem. The problem is landlords colluding with each other to force prices up instead of competing with each other (which would bring prices down). 100% of the information could be public and they would still be operating as a cartel.

The problem is both the private data and the algorithmically determined rent increases across the board.

Without the private data, the algorithm wouldn't be able to raise rents nearly as aggressively, as there would be no way to know whether any given landlord was going to go along with it. Without the algorithm, the landlords wouldn't be nearly as efficient at setting rents at just-below-too-high-for-the-market-to-bear. Without either, the landlords would essentially be looking at FOR RENT ads and making their best guess - which is what should be happening in this (and every other) market.

If you are just going to use an algorithm to tell you to raise rents, why even bother basing it on data? Just keep raising it and do "normal" collusion.

That's what landlords did before this option was made available, and it was markedly less efficient at extracting rents. In the Washington metro area there's been a clear spike in rents in Virginia now that they have far more clarity on the rents in Maryland and DC. Rent on a 1200 square foot condo jumped almost $600/mo during 2020-2022 timeframe, and coincidentally the landlords got access to RealPage in those timeframes.

What I was suggesting was that if an algorithm is telling landlords to up their rent to $1200/mo, why not just have everyone up it to $1500? Or $2000?

Or is the algorithm essentially including estimates of rates in which tenants will fail to be able to pay and have to get evicted (which has costs to the landlords)?

IE
@$1200 expect 1% eviction rates
@$1250 expect 3% eviction rates
@$1300 expect 7% eviction rates
@$1350 expect 15% eviction rates
@$1400 expect 30% eviction rates
@$1450 expect 50% eviction rates
@$1500 expect 90% eviction rates

And then does some math to figure out what the highest rate is, where once you include the eviction costs, the total is the highest...?

 

[marsilies]

What I was suggesting was that if an algorithm is telling landlords to up their rent to $1200/mo, why not just have everyone up it to $1500? Or $2000?

Or is the algorithm essentially including estimates of rates in which tenants will fail to be able to pay and have to get evicted (which has costs to the landlords)?

You're making it too complicated. The maximum revenue a landlord can extract from their properties is average rent * total occupied residences. The algorithm figures the optimal revenue based on these. So while it will allow for additional vacancies if the overall revenue goes up due to the rent increase, it's not going to raise the rent to the point where the landlord's overall revenue goes down. So there's limits on how high and how fast it's going to raise rent in an area.

 

[rosen380]

What I was suggesting was that if an algorithm is telling landlords to up their rent to $1200/mo, why not just have everyone up it to $1500? Or $2000?

Or is the algorithm essentially including estimates of rates in which tenants will fail to be able to pay and have to get evicted (which has costs to the landlords)?

You're making it too complicated. The maximum revenue a landlord can extract from their properties is average rent * total occupied residences. The algorithm figures the optimal revenue based on these. So while it will allow for additional vacancies if the overall revenue goes up due to the rent increase, it's not going to raise the rent to the point where the landlord's overall revenue goes down. So there's limits on how high and how fast it's going to raise rent in an area.

To me that feels like somethign that would be high value to landlords with small numbers of units though. Once you are one of the big dog with hundreds or thousands of units in the area (which my impression is largely the companies using the data), it feels like your own data likely gets you 90-95% of the way there anyways.

So does it suddenly become OK if RealPage pivots to only using a particular landlord's data for that landlord? Then it is largely the same rent-setting without it looking like collusion.

 

[marsilies]

To me that feels like somethign that would be high value to landlords with small numbers of units though. Once you are one of the big dog with hundreds or thousands of units in the area (which my impression is largely the companies using the data), it feels like your own data likely gets you 90-95% of the way there anyways.

So does it suddenly become OK if RealPage pivots to only using a particular landlord's data for that landlord? Then it is largely the same rent-setting without it looking like collusion.

There's always a benefit to collusion, otherwise companies wouldn't keep trying it and/or looking for loopholes like letting an algorithm do the collusion.

But yes, on a legal level, if the Realpage algorithm was looking at solely one particular landlord's private data (along with public data for all the other rental properties) to form recommendations for that landlord, then it wouldn't be collusion. If one landlord is so large as to be able to drive the market on their own though, then there's other monopoly/anti-trust laws that should come into play.

The idea of competition is that if there's meaningful competition, then any one company can't unreasonably raise prices, as people will switch to using the competitor. If competition becomes basically non-existent, either through collusion or just one company basically becoming a monopoly, then something is wrong and needs to be fixed.

 

[JohnDeL]

What I was suggesting was that if an algorithm is telling landlords to up their rent to $1200/mo, why not just have everyone up it to $1500? Or $2000?

Or is the algorithm essentially including estimates of rates in which tenants will fail to be able to pay and have to get evicted (which has costs to the landlords)?

IE
@$1200 expect 1% eviction rates
@$1250 expect 3% eviction rates
@$1300 expect 7% eviction rates
@$1350 expect 15% eviction rates
@$1400 expect 30% eviction rates
@$1450 expect 50% eviction rates
@$1500 expect 90% eviction rates

And then does some math to figure out what the highest rate is, where once you include the eviction costs, the total is the highest...?

It bases the eviction rates on the number of evictions at that price point for that demographic in that area, which is something it can do thanks to the fact that all of those landlords are sharing all of that data with RealPage. Thanks to regression to the mean, while the data from one leasing company might be skewed when all of them are put into the same pot you get a much truer picture of what is happening.

To see this in action, flip a coin four times and count the number of times it comes up heads. You should get an about equal number of heads and tails but can easily get four heads or four tails because the sample is so small. Now flip twenty coins four times each and count the total number of heads you get. Odds are, this time you got a number that is very close to 40 heads (somewhere between 32 and 48) and if you get 80 heads you know that the coins are biased.

Now imagine doing that with rental prices. That's the power of this tool.

The other thing that is getting hidden is that this tool can easily be used to discriminate against certain groups. "The algorithm says that in order to rent to you, we need to have a higher deposit because you are more likely to default" is more defensible in court than "we require blacks to give us an extra month's deposit".

To me that feels like somethign that would be high value to landlords with small numbers of units though. Once you are one of the big dog with hundreds or thousands of units in the area (which my impression is largely the companies using the data), it feels like your own data likely gets you 90-95% of the way there anyways.

Except that the large landlord is likely to have their properties spread across a variety of locales, each of which will have its own optimization function based on demographics, location, etc. So you might get 80% of the way there but not 95%. And when you are talking about millions per month in rent income, that adds up fast!

What this company does is let you coordinate with every other property owner in a given area to maximize revenue for all of you. That such coordination is collusion is just a technicality (right, sure).

So does it suddenly become OK if RealPage pivots to only using a particular landlord's data for that landlord? Then it is largely the same rent-setting without it looking like collusion.

If they do that, then they lose the "killer app" part of the app.

 

[Bondles_9]

So does it suddenly become OK if RealPage pivots to only using a particular landlord's data for that landlord? Then it is largely the same rent-setting without it looking like collusion.

No, because it's still collusion if RealPage is giving "advice" to all of the landlords with the objective of driving up all rents instead of getting the best outcome for the one landlord. If one landlord was using RealPage, and another was using RealStage, and another was using RealPhage, then that would be fine.

The question is around the motivation for one landlord to undercut the others to attract more tenants. The whole point of RealPage is to remove that incentive (whether it's using private data or not), which removes competition from the market, benefitting all of the landlords at the cost of all of the tenants.

 

[Statistical]

They just gotta amend the algo to not use nonpublic data, or only use nonpublic data from the client. It could still help clients decide rents to charge.

Seriously, let the "private information" thing go. Private information is not the problem. The problem is landlords colluding with each other to force prices up instead of competing with each other (which would bring prices down). 100% of the information could be public and they would still be operating as a cartel.

The problem is both the private data and the algorithmically determined rent increases across the board.

Without the private data, the algorithm wouldn't be able to raise rents nearly as aggressively, as there would be no way to know whether any given landlord was going to go along with it. Without the algorithm, the landlords wouldn't be nearly as efficient at setting rents at just-below-too-high-for-the-market-to-bear. Without either, the landlords would essentially be looking at FOR RENT ads and making their best guess - which is what should be happening in this (and every other) market.

If you are just going to use an algorithm to tell you to raise rents, why even bother basing it on data? Just keep raising it and do "normal" collusion.

That's what landlords did before this option was made available, and it was markedly less efficient at extracting rents. In the Washington metro area there's been a clear spike in rents in Virginia now that they have far more clarity on the rents in Maryland and DC. Rent on a 1200 square foot condo jumped almost $600/mo during 2020-2022 timeframe, and coincidentally the landlords got access to RealPage in those timeframes.

What I was suggesting was that if an algorithm is telling landlords to up their rent to $1200/mo, why not just have everyone up it to $1500? Or $2000?

because it may then go unrented. Even a few months of vacancies can bring down your yearly income average.

 

[Defenestrator]

Strange how the free market always steers towards turning people into serfs. Wonder why that is.

Because actually competing in a free market is brutally difficult. Unless there are actively-enforced up-to-date rules to prevent it, it's often much easier to take advantage of or create market distortions and increase your profit margins.

 

[MyersVandalay]

Strange how the free market always steers towards turning people into serfs. Wonder why that is.

Yeah, it's kind of basic capitalism 101. (And feudalism, and most the systems from history). Those with more, are in better positions to get more. It's a god damn circle. You have more, you make more, you use what you make to have even more, which makes even more, until limited resources force you into competition with others, in which case, the one who already has more has a strong advantage, and unless massively outplayed will take from the one with less.