Dystopian Pricing Tactics: A Primer
Pricing and fairness in an age of AI.
This is Boondoggle, the newsletter about how corporations rip off our states, cities, and communities, and what we can do about it. If you’re not currently a subscriber, please click the green button below to sign up. Thanks!
The political buzzword of the moment is “affordability,” with politicians representing every spot along the political spectrum trying to work it into their remarks and legislating. Which makes sense! Prices are undeniably high, especially for groceries and utilities, and if they aren’t addressed soon they will be a political liability for the ruling party at the federal and state levels when elections roll around in November.
Part of the challenge legislators face when attempting to bring down prices is that new technologies are enabling novel pricing abuses by dominant corporations, as well as turbocharging and making opaque some old, already illegal practices. The continued creep of artificial intelligence and algorithmic price-setting into retail spaces, as well as the migration of so much commerce online, is enabling pricing shenanigans that simply aren’t possible in a brick-and-mortar setting (yet!) and that many laws on the books weren’t written to address.
Both last year and this, there was a flurry of legislation attempting to address these problems. There was and is also a lot of confusion — among elected officials, advocates, media, and the general public — about what, exactly, is happening out there in the economy when it comes to price setting and what solutions can and should be adopted in order to deal with it.
Many politicians, pollsters, and communications staffers also prefer terms such as “price gouging” or “predatory pricing” when describing all of these practices, even though those terms have particular legal definitions that don’t line up with what the politicians are actually trying to outlaw or rein in.
And as with so many things, there is no one-size-fits-all, easy-to-understand omnibus “make prices clearer and fairer bill” that can be passed in one quick vote.
Today, I’m going to break down five dystopian pricing tactics with which state legislators are attempting to grapple, explain where they overlap and are different, the effects they have, and what policies are being pursued in order to prevent them. I hope it’s helpful in both clearing up some of the confusion and in directing all the righteous energy high prices are creating onto something practical and useful.
Surveillance Pricing: Surveillance pricing is the use of personal data, collected via online surveillance or purchased from data brokers, to set individual prices for individual consumers. Last year, after asking several corporations about their data collection and pricing practices, Federal Trade Commission staff found “that consumer behaviors ranging from mouse movements on a webpage to the type of products that consumers leave unpurchased in an online shopping cart can be tracked and used by retailers to tailor consumer pricing,” and warned that, for example, a parent trying to buy a thermometer may be charged more than someone else because their data trail indicates they have a sick child at home. The exact inputs and algorithms used to determine those individual prices are, of course, a mystery, but there is certainly a large and growing business sector providing surveillance pricing infrastructure to retailers. Legislators in several states — including Colorado, Illinois, Virginia, Pennsylvania, Tennessee, and Washington — have bills to either ban surveillance pricing outright or eliminate certain types of data from being used to determine individual prices. The California attorney general also announced he is investigating the practice. New York last year adopted a law requiring that sellers disclose to consumers that their personal data has been used to set prices, but I don’t see much utility in that, as it puts the onus on the consumer to do … something … with that information, while allowing the tactic to continue unabated.
Dynamic Pricing: Dynamic pricing is the constant shifting of prices, supposedly based on some outside factor, via opaque inputs and algorithms. Though most closely associated with live events such as concerts or sporting events — remember the blow-up over dynamic pricing for World Cup tickets? — it happens on all sorts of online platforms, including Uber and Amazon. It’s distinct, but related to, “surge pricing” or price gouging, which is the exploitation of a particular event or market disruption to jack up prices in the short term, but dynamic pricing is more constant and even less transparent, as there’s often no obvious rhyme or reason dictating the price changes and no way to anticipate when they will increase or decrease. The result of dynamic pricing can clearly be different people being charged different prices for the same good, despite purchasing it at the same time or in relative proximity. Dynamic pricing can also very easily bleed into surveillance pricing, if the altered prices are offered to individual consumers based on a data point about the individual purchaser. Several state legislators are attempting to put new guardrails around dynamic pricing. The most ambitious is probably the Protection from Predatory Pricing Act in Maryland, which would require grocers to keep prices fixed for one business day.
AI-Powered Price Discrimination: The technologies enabling the above practices also enable price discrimination, such as that revealed by a recent investigation into the grocery delivery platform Instacart, which showed that shoppers purchasing the same goods from the same location at the same time were paying as much as a 23 percent difference, due to Instacart running constant, real-time tests to see which items it could increase the price of at any given moment, unconnected to any external factor. The underlying technology used by Instacart promises to drive higher margins via the use of algorithms that “figure out which categories of products our customers [are] more price sensitive on” and set prices “based on that information.” They were just constantly raising prices to see if they could. It’s unclear how these platforms choose which shoppers receive which prices, but this practice has the potential to blend the worst of dynamic pricing and surveillance pricing, creating a sea of constantly shifting prices based on unknowable inputs and unclear determinations about what factors go into each individual shoppers price. The New York attorney general announced an investigation into Instacart’s practices.
Electronic Shelf Labels: One of the reasons dynamic and surveillance pricing are worrisome for many policymakers at the moment is the specter that they will migrate from online shopping to in-person shopping via the use of electronic shelf labels, which are what they sound like: Electronic price tags that can change in real time based (again!) on unclear inputs and algorithms, which in theory could enable unpredictable pricing jumps or even allow retailers to show no constant price at all, but instead mesh data on individual shoppers to craft individualized prices. There is one academic study showing that Walmart and Kroger have thus far not used electronic labels to engage in unpredictable price jumps, but the technology is obviously capable of enabling such action, so several state legislators have looked to either rein in the ability of those who use electronic labels to change prices in real time, or even to ban electronic shelf labels entirely (which seems like a make-work program for grocery store employees, to be honest).
Algorithmic Price Fixing: So far, I’ve only touched on technology that alters interactions between a seller and the consumer, but AI-powered algorithms can also enable sellers to collude with each other, as seen in rental housing, agriculture, hotels, or even frozen potatoes, resulting in higher prices for consumers. In this arrangement, centralized platforms act as price coordinators, collecting data and recommending prices to sellers in a particular market based on (you guessed it!) opaque inputs and algorithms, which allows those sellers to coordinate prices and ensure they don’t undercut each other. The result is good old fashioned price fixing, even if it’s dressed up in new technology. About a dozen municipalities, as well as the state of New York, have passed laws to rein in algorithmic collusion in rental housing, while California took a more comprehensive approach and enacted a new law to address algorithmic price fixing economy-wide.
I’m leaving out some practices — such as the use of junk fees or more traditional price discrimination between suppliers and sellers — that, while extremely problematic and the deserved source of much policymaking, aren’t being supercharged by AI and algorithms in the same way (at least that we know of).
All of this is important to understand, because these pricing tactics undermine the necessary feedback that keep a market-based economy functioning. If consumers can’t comparison shop, can’t escape personalized prices tailored exactly to their data profile, can’t avoid collusive seller behavior, and can’t predict some sort of reasonable price stability over time, then they will constantly face higher prices and be unable to facilitate the competition between sellers that’s necessary to lower them.
Corporate leaders always argue that all of this technology is only being deployed to offer people personal discounts and bargains, but that beggars belief: They’re not investing in all this fancy technology to lower prices for everyone and make lower profits. Just like they’ve warped loyalty and rewards programs to turn troves of data into traps and higher prices, so too with everything I’ve described above.
Distinct practices also require distinct policy solutions, so while it may be annoying, I’m going to keep trying to prevent folks from conflating all of the above or drafting Frankenstein’s monster bills that grab bits and pieces of different policy solutions and smash them together incoherently. Hopefully, then, some progress is made to prevent price dystopia from becoming our permanent reality.
SHAMELESS SELF-PROMOTION: I talked to The Register about the use of nondisclosure agreements in data center deals. Read the piece here.
SIMPLY STATED: Here are links to a few stories that caught my eye this week.
A Massachusetts court ruled that the online “prediction market” firm Kalshi needs to stop offering sports betting.
A trial of two utility executives at the center of the most significant public corruption scandal in Ohio history began this week.
“South Carolina has racked up $150 million in cost overruns as it works to fulfill pledges it made to lure electric vehicle maker Scout Motors to the state.”
Thanks for reading this edition of Boondoggle. If you liked it, please take a moment to click the little heart under the headline or below. And forward it to friends, family, or neighbors using the green buttons. Every click and share really helps.
If you don’t subscribe already and you’d like to sign up, just click below.
Thanks again!
— Pat Garofalo
This taxonomy is useful, and the instinct to keep the categories separate is right for legislation. But it's worth noting where they stack. Rental housing is probably the clearest example.
Algorithmic price fixing through RealPage sets the recommended rent. But the recommendation has structural weight because RealPage's clients collectively control roughly 70% of multifamily units in markets like Phoenix and Seattle. The algorithm coordinates pricing. The concentration ensures there's nowhere to escape to. And the platform explicitly recommends holding units vacant rather than lowering rent, which means the supply side is being managed by the same system setting the price.
That's #5 on your list doing the work of #2 and #4 simultaneously, in a market where consumers can't comparison shop their way out because the same data feed is pricing most of the available inventory.
Arizona's AG found 12-13% overcharges across tens of thousands of units. Stoller estimated up to a quarter of rental inflation between 2020-2024 may trace to this. And the DOJ case was dropped by Gail Slater earlier this year, so the enforcement window is closing even as the practice scales.
The legislative challenge is that a bill targeting algorithmic collusion alone doesn't address the concentration that makes the collusion effective. Housing may need both.
The solution to all these problems is a law mandating that the price for a product can only be changed once per day. I believe this is the law in NJ when it comes to gas presumably so that station owners don't have a white price and a Black price.