Sophisticated algorithms and access to online shopping data are increasingly enabling firms to tailor prices to individual consumers. This has provoked unease about the implications for privacy and fairness. In a new working paper, TSE’s Andrew Rhodes teamed up with Jidong Zhou (Yale) to investigate the impact of personalized pricing. Their research shows that market conditions – including costs, competition, and information imbalances – can determine whether firms and consumers stand to benefit.
How concerned should we be at the spread of personalized pricing?
Price discrimination has been documented in many industries, including retail, travel, and personal finance. According to a recent survey, around 40% of firms that have adopted AI for personalization use it to set real-time prices and promotions. In 2012, the Wall Street Journal revealed that retailers like Staples and Home Depot were personalizing prices on their websites, based on a consumer’s browsing history and distance from a competitor’s store. Firms’ ability to personalize prices surreptitiously, avoiding a potential consumer backlash, has likely grown significantly since then. Today, personalized prices are often concealed as personalized discounts sent by email or smartphone app.
Policymakers are often wary of personalized pricing, raising concerns that shareholders benefit at the expense of consumers. But some economists are supportive of the idea, arguing that it can improve both profits and fairness by expanding the market, allowing firms to sell to consumers who would otherwise be shut out by high uniform prices.
Personalized pricing is usually bad news for consumers in the absence of competition, as it allows firms with monopoly power to extract all the social surplus. But this impact can be completely reversed when a firm loses its monopoly position. Under duopoly, prior research shows that personalized pricing may lead to a reduction in the price paid by every consumer. The reason is that each firm tries to poach its rival’s consumers with low prices, forcing the rival to charge less. Moving to duopoly now harms firms but benefits consumers. This insight has been very influential for subsequent research on issues such as data privacy and data brokers.
How does your research help to inform the debate?
Very fine-tuned personalization is likely to become increasingly feasible. With this in mind, our paper focuses on perfect, or first-degree, price discrimination . We compare this with a uniform pricing regime, in which all consumers face the same price.
We first study the short-run case where the market structure is fixed. Contrary to existing research, we show that some personalized prices exceed the uniform price: although consumers who regard their two best products as close substitutes pay less under personalized pricing, consumers with a strong preference for one product end up paying more. But when the market is fully covered – that is, if all consumers buy a product – personalized pricing intensifies competition, harms firms, and benefits consumers.
However, for most markets it is more realistic to assume only partial market coverage. This can completely reverse the impact . In markets that are already competitive – for example, when production cost is low, or the number of competitors is large – personalized pricing intensifies competition and benefits consumers. But in markets where competition is relatively weak – for example, when product cost is high or the number of competitors is small – personalized pricing further reduces it, harming consumers.
We also study what happens in the long run when firms choose whether to pay a fixed cost to enter the market, and then engage in price competition. Personalized pricing allows the marginal firm to fully extract the increase in match efficiency caused by its entry. Consequently, personalized pricing leads to the socially optimal market structure, to the benefit of consumers.
What if only some firms have access to consumer data, using personalized prices to poach consumers from less-informed rivals? We show that when market coverage is relatively high, this mixed case can be the worst for consumers. Policies which prevent data-rich firms from price discriminating, or which force them to share their data, may therefore protect consumers.
What might explain why personalized pricing can benefit firms and harm consumers?
Personalized pricing can bring more consumers into the market but since they have low valuations for the product, these consumers may experience little benefit. At the same time, a relatively high number of consumers who would have bought under uniform pricing have a strong preference for one product, so personalized pricing can raise the average price they pay. This can make consumers worse off overall, even though it expands demand.
When marginal cost is higher, a firm may face a new “monopoly segment” of consumers who value only its product above marginal cost. This gives firms some monopoly power, which is an additional force for personalized pricing to harm consumers.
What are the incentives for consumers to allow their data to be used for personalized pricing?
Privacy policies like GDPR in the EU and CCPA in California give consumers some control over what data is harvested and how it is used. We study this issue in our companion paper on “Personalization and Privacy Choice” (2022). We find that when more consumers share their data, firms may increase their prices for consumers who choose to remain anonymous. In light of this novel externality, although privacy policies such as GDPR benefit consumers, we find that there are still too many consumers who choose to share their data. We also show that competition can harm consumers overall by inducing more of them to share their data.
The researchers’ working paper ‘Personalized Pricing and Competition’ is available to read at www.tse-fr.eu
Interview published in TSE Reflect, October 2022
