In the summer of 2012, the Wall Street Journal reported that the online travel agency Orbitz had sometimes shown Apple users hotel accommodations with a higher nightly rate than those it displayed to Windows users. The company discovered that Mac computer users spent up to 30 percent more per night on hotels. It was one of the first prominent instances in which the predictive capabilities of algorithms were demonstrated to have an effect on consumer-facing pricing.
Since then, the pool of data available to businesses about each of us (the information we’ve either volunteered or can be inferred from our web browsing and purchasing histories) has grown substantially, enabling businesses to create ever-more-precise customer profiles. Personalized pricing is now ubiquitous, despite the fact that many consumers are only now becoming aware of its existence. Users and fans have recently been astonished by other algorithm-driven pricing models, such as Uber’s surge and Ticketmaster’s dynamic pricing for concerts. In recent months, dynamic pricing, which is based on factors such as quantity, has increased the prices of some concert tickets, including those for Drake and Taylor Swift, even before they reached the secondary market. And while personalized pricing is slightly different, these examples of computer-driven pricing have generated headlines and social media posts that reflect a growing discontent with the role of data in determining prices.
The marketplace is said to be a realm of presumed equity, governed by the norms of competition, a neutral setting in which all consumers are treated equally. However, this concept is being undermined by the same opaque and perplexing programmatic data profiling that is steadily invading other aspects of our lives: algorithms. The government of Canada is presently contemplating new consumer protection regulations, including how to regulate algorithm-based pricing. Some consider stringent market regulation to be a political danger, but there may be an alternative solution—not at the point of sale, but at the point where our data is collected in the first place.
Theoretically, pricing algorithms are not inherently negative. A portion of the purchasing process could be made more efficient and personalized for each consumer if prices were more responsive to market forces and independent of human intervention. It may also result in pricing that is more transparent. And, technically speaking, by using data profiling to accurately assess a consumer’s willingness to pay (i.e., the maximum amount someone would spend on something), personalized pricing could meet people where they are financially, allowing some people to purchase items they otherwise could not afford.
Again, this is only a theory. It sometimes works differently in actuality. On the path to maximizing profits (the objective of pricing algorithms) strange things can occur. Researchers concluded in 2019 that “relatively simple pricing algorithms systematically learn to play collusive strategies” by continuously adapting to changes made by others, despite not being intended to do so or able to communicate with other pricing algorithms. Under this scenario, researchers noted in 2021 that “the largest gains accrue to a dominant firm with the most advanced technology and the largest market share.”
A similar conclusion was reached by a 2016 study that monitored for four months the top twenty Amazon sellers of over 1,600 products. In some instances, algorithms were altering the prices of goods “tens or even hundreds of times per day” (a frequency that would be difficult for a human to replicate), resulting in “a largely winner-take-all marketplace.” The study revealed that these sellers received more positive feedback, giving them an advantage in terms of Amazon page rankings. In other words, a website that appears to offer a vast selection of products may, when prompted by algorithms, offer only a handful of top-selling options. Algorithms may also result in an increase in the overall cost of goods, as lowering prices may only encourage competitors to undercut, reducing the incentive for anyone to reduce prices. If you feel like you’re constantly spending more money, it may be because you are.
Concerns also exist regarding the inherent biases of big data. In 2015, ProPublica revealed that prices for the Princeton Review’s online SAT tutoring packages varied by US ZIP code, resulting in a “unexpected effect . . . that Asians [were] almost twice as likely to be offered a higher price than non-Asians.”
All of this makes regulation both necessary and challenging, especially if the issue is addressed at the point of sale. A government could, for instance, institute price controls for consumer products, ensuring that we never pay more than a predetermined amount (an idea that has been discussed recently as a temporary solution to rising food prices). However, its implementation is not typical, and its history would make it politically unpopular. In actuality, any form of market regulation is likely to elicit a backlash from businesses that use techniques like personalized pricing to increase profits, as well as from opposition parties seeking ideological leverage. Given the recent history of the government’s update to the country’s broadcasting act, which also centered on the issue of the commercial value of data and sparked a frenetic discussion about censorship, a foray into personalized pricing regulation may be too risky to undertake.
Canada’s Competition Act was updated in the 2022 federal budget bill, but these amendments did not directly address algorithmic pricing. Innovation, Science, and Economic Development Canada noted in a discussion paper that the act needs improvement due to “the new challenges posed by how data-driven and digital markets operate.” The paper acknowledged that there are “valid reasons to limit grounds for intervention in private commerce,” but conceded that the question of how to do so is becoming increasingly complex. “TThe public interest is not well-served if competitive harm is identifiable but the [Competition] Bureau is not sufficiently empowered to intervene.”
Preventing Algorithmic Pricing Harms: Shifting Focus to Personal Data Access in Shopping
It may be possible to limit the negative effects of algorithmic pricing without directly intervening with a seller. To get to the core of the problem, policy could be targeted much earlier in the purchasing process and could concentrate on what personal data is available to sellers in the first place.
A more transparent declaration that prices are being customized by algorithms could be an initial step in addressing the issue. However, greater transparency may merely increase awareness without resolving anything. Moreover, individuals are already aware that computers are up to no good. Instead, Chapdelaine proposes legislative restrictions on the use of personal data that could be used to personalize pricing, as well as increased enforcement by the courts and privacy commissioner.
In light of the fact that many retailers and services are still recuperating from pandemic-related losses, it could be politically and practically risky for the government to restrict technology that businesses know can improve profit margins. Having been accused of losing contact with ordinary people, and specifically the bills they pay, the government may discover that the benefits of addressing algorithmic pricing and incorporating more fairness into our online marketplace outweigh the costs.
Source: The Walrus