I.3.1 Emphasis on Algorithmic Decisions or on the Price?

The literature on algorithmic price personalization, including the contributions to this volume, maps along an axis of whether the authors emphasize that an algorithm is taking decisions or that the decisions taken are about price. This axis is particularly conspicuous in discourse about the fairness and transparency of algorithmic price personalization.

In a first approximation, scholars focusing on algorithmic fairness are investigating price personalization in reference to an already pretty robust body of literature on the general fairness of algorithmic decisions. The price-focused literature in turn looks at algorithmic price personalization from the perspective of contracts and competition, and of market law more generally, to identify instances of fair as well as unfair algorithmic price personalization. Consequently, scholars emphasizing the transparency of algorithmic decision-making rely primarily on the vast literature on transparency and explainability (the algorithmic ‘black box’ problem) and on the ‘human in the loop’ idea of algorithmic bias and related problems.

The accounts tilting towards the price dimension refer more frequently to the economic parameters of personalized pricing, including its welfare and distributional outcomes as well as its effects on competition and antitrust law. In addition, the price-centred accounts often invoke transparency, which in this context refers not to algorithmic transparency as such, but to informing consumers about price and its basic parameters. Less commonly, this thread of literature also invokes the link between personalized pricing and classical price fairness doctrine as well as the general toolbox for reviewing price clauses (which includes such instruments as the EU Unfair Contract Terms Directive [UCTD]).

A deeply intertwined set of concerns, situated somewhere between price and fairness, is privacy.Footnote ²³ By its nature, algorithmic pricing builds on harvesting and processing personal data from the consumer market. Hence, algorithmic pricing clearly falls within the ambit of privacy regulations such as the General Data Protection Regulation (GDPR)Footnote ²⁴ and the California Consumer Privacy Act (CCPA).Footnote ²⁵ Consequently, algorithmic pricing is also incidentally subject to rules on automated profiling in the EU law, especially to Article 22 of the GDPRFootnote ²⁶ (notwithstanding that the real significance of this rule for price personalization is limited and cumbersome). From the algorithmic pricing perspective, privacy can be seen in a twofold way. First, if it is considered to constitute a stand-alone value in personalized pricing, privacy may be situated near the algorithmic-price-fairness end of the spectrum.Footnote ²⁷ Second, privacy can also be understood instrumentally, as a component of the transparency requirement. Seen through this prism, privacy is more of a procedural than a substantive standard, meant to protect other interests (e.g., not being exploited or discriminated) by prohibiting the use of consumer information in ways that will harm them.

In the background of these institutional and normative standpoints lies a broad cluster of empirical studies into consumer attitudes towards dynamic and personalized pricing schemes. All these studies point to a clear bottom line: consumers are generally strongly averse towards algorithmic pricingFootnote ²⁸ (and generally towards all instances of price differentiation in online and offline contextsFootnote ²⁹). Arguably, this aversion is rooted in consumers’ intuition that price differentiation carried out by a machine is substantially different from similar operations in the brick-and-mortar setting.Footnote ³⁰ In other words, popular perception holds algorithmic pricing to be ethically questionable, not only because it allocates different prices to different clientsFootnote ³¹ but also because the differentiation in question is based on personal details analysed by a de-humanized agent.

The relevance of the empirical studies in question is two-fold. First, the pedestrian view of price fairness can serve as a guideline in setting regulatory frameworks for personalized pricing. Although collective moral judgments are not definitive for the design of legal rules, they do however provide relevant signposts for understanding where the boundaries of price personalization should be placed.Footnote ³² Second, consumer backlash may dissuade firms from using algorithms at all (or at least, from using them opportunistically, as a way of rent-seeking). This pertains especially to what is known as ‘dual entitlement theory’, which provides an explanatory formula for social attitudes towards price changes identified through empirical studies. Dual entitlement theory holds that firms are generally averse to increasing prices solely due to fluctuations in demand or for certain other reasons that consumers generally do not deem acceptable. In turn, the dual entitlement view holds that firms are more inclined to increase prices to reflect increased costs of producing or obtaining a good (or at least, they are inclined to present price increases this way), as this is the only type of price increase that the majority of consumers tends to approve of. From a normative perspective, while dual entitlement theory taken by itself is merely one guide, its fitness with a broader fairness account needs to be stressed and is discussed in Section I.3.2.

I.3.2 Fairness

The bourgeoning use of personalized (that is, of individual or at least of granularized) prices, based on profiling of individual market preferences, triggers academic and policy discussion along several lines. One of the most foundational (and chronologically, the most seminalFootnote ³³) of them ponders the (un)fairness of price calculation schemes.Footnote ³⁴ The widespread use of various argumentative formulas, also in this regard, ultimately seems inconclusive, or at least not fully consistent in its ends.Footnote ³⁵ In many instances, fairness operates as the epitome of an unspecified set of values that embody the moral features of a ‘proper’ algorithmic price calculation. At the same time, fairness-based assertions occupy a particularly prominent position in the algorithmic pricing debate. The emergence and rapid proliferation of automated price calculation have brought the long-stagnant and ‘distinctly unfashionable’Footnote ³⁶ debate over price fairness to the fore again. It has revived foundational questions of the ethics of setting prices and is attempting to incorporate into them a partly altered domain of market phenomena and ethical judgements.

In the general debate on algorithms, the notion of fairness hence functions as something of a nebulous concept of certain values or moral principles that can serve as a yardstick for gauging whether the use of a particular algorithm is, broadly considered, ethical.Footnote ³⁷ In a nutshell, fairness-based arguments are underpinned by one of the two patterns mentioned earlier.Footnote ³⁸ They gravitate either towards the question of price per se or towards the fairness of the algorithmic procedure. Under the latter view, the concept of algorithmic fairness encompasses a diverse set of issues.Footnote ³⁹ Some of these issues involve imperfections in the algorithmic decision and the biases affecting it.Footnote ⁴⁰ Other issues may be more foundational, as they regard the right to be left alone, understood in this context as the wish to keep one’s ‘private’ life and persona separate from one’s ‘market’ persona.Footnote ⁴¹ Here, again, it is possible to operate along the lines of the fairness and transparency lenses. And in fact, it is perfectly imaginable that some portion of the population may have an intrinsic distaste for the merging of their private and market personas. Others, however, might be distinctly concerned that the merging of the two spheres will lower the quality of their market options. Apprehension on the part of some consumers suggests that the ethics of algorithmic pricing rest on a broader foundation than does the classic concept of a fair price. In this sense, moral concern about algorithmic pricing may far precede any contractual relationship between the parties. In this way, algorithmic calculation of prices extends the fairness argument into areas that normally fall outside the scope of axiological interest in contract law.

Accordingly, ‘fairness’ should be invoked with caution in the algorithmic context, as it often serves as an umbrella for various arguments related to the ethics of price differentiation introduced by an algorithm. It may be more fruitful to look at the fairness of price personalization from the perspective of contract law. Prices set using algorithms are part of the same axiological universe that emerged at the outset of modern contract law by developing the notion of laesio enormis, which puts a threshold on the substantive outcomes of the bargaining process. The current contract law debate has voiced concern over and against this idea, presenting it as too paternalistic and overly vague in its axiological fabric.Footnote ⁴²

Other, more optimistic views have pointed towards a large toolbox that could ensure the substantive fairness of personalized prices. This pertains especially to various devices in contract law, such as unconscionability, or to the judicial review of price terms, which a prevalent view would also hold is also applicable in the classical framework of contract law.Footnote ⁴³ Indeed, contributors to this Handbook have taken seriously the possibility of such substantive intervention. Chapters in all three parts of the Handbook have recognized that there is normative justification for the claim that personalized prices can be too high.

Thus, while the difficulty of identifying a benchmark might well justify limited legal review, the normative concern seems easy to articulate: exploitation. As Paul Krugman puts it, personalized prices are ‘undeniably unfair’ when ‘some people pay more just because of who they are’.Footnote ⁴⁴ This normative intuition finds deep roots in the idea of individual and, more precisely, of consumer sovereignty, as well as of its welfare equivalent: consumer welfare maximization. Indeed, in an account of market relations based on consumer sovereignty, ‘[a]s consumer the individual is sovereign; as producer he is subject’.Footnote ⁴⁵ This normative asymmetry actually relies on a strong notion of substantive equality: everyone as a consumer is sovereign and everyone as a trader is a subject. The emphasis is on traders endeavouring to offer consumers the best deals and then competing with one another on the merits to attract consumers rather than competing to indirectly maximize social or total welfare.

It is apparent that we are far from an ideal of consumer sovereignty if traders are using consumer data to extract value from them. Moreover, this normative concern has significant empirical support. As the dual entitlement theoryFootnote ⁴⁶ emphatically illustrates, price variations are subject to widely shared social beliefs (that are mostly at odds with using price personalization to charge consumers higher prices). Second, at least in the EU, similar considerations also motivate vast branches of market regulation.Footnote ⁴⁷ The development of EU energy law provides direct examples of this. For instance, the Court of Justice of the European Union (CJEU) has interpreted a vague reference to a right to reasonable prices in Directive 2003/55/EC as a proxy for the right to a competitive price.Footnote ⁴⁸ Subsequent legislative reforms in the energy sector have made this individual right to benefit from competition even more explicit.Footnote ⁴⁹

I.3.3 Transparency

While prices might be difficult to regulate directly, transparency-based interventions might go a long way by empowering consumers to look after themselves. But the appeal of transparency-based tools also derives from individual autonomy; in particular, transparency appeals to individual preferences and decisional processes while shielding consumers from third-party assessment.Footnote ⁵⁰

The issue of transparency in algorithmic pricing sits at the juncture of two more generic questions: one, of transparency in consumer transactions; and two, of the overall cognizability of algorithmic decision-making (often referred to in scholarship as ‘algorithmic fairness’).Footnote ⁵¹ Understood in this way, transparency of algorithms operates not merely as a procedural requirement but as an indirect guarantee that the process of price calculation will be understandable for the individuals subject to price personalization and will allow them to scrutinize its premisesFootnote ⁵² and to make a decision about opting out of the personalization scheme. However, the exact content of the transparency requirement is itself not very transparent. Depending on one’s vantage point, the prescription may encompass a ‘thinner’ or ‘thicker’ version of transparency. Thinner transparency relates to the transparency of price personalization as such, mandating clarification that the particular price is differentiated for various consumers through algorithmic decision-making. Thicker transparency refers to the setting in which not only the fact of personalization but also the criteria used to calculate the price are revealed to the consumer. This thicker transparency may rest either on criteria set ex ante (indicating the premises to be taken into account by an algorithm) or on ex post elucidation of why the particular price was calculated in a certain way (and hence tending towards ‘algorithmic explainability’).Footnote ⁵³

In the existing EU legal framework, requiring transparency is the most basic approach to price regulation, including personalized pricing. First, the UCTDFootnote ⁵⁴ eschews price control and leaves it to the parties’ autonomous choice. Article 4(2) UCTD excludes from the unfairness assessment terms that determine ‘the main subject matter of the contract’ as well as ‘the adequacy of the price and remuneration … as against the services or goods supplies in exchange’. The application of this exclusion is conducive to transparency of the price term (i.e., when it may be assumed that a consumer was able to understand the price and make a meaningful market choice). Hence, the UCTD approach to price rests on a strong assertion that fairness-based interventions into price are only legitimate as long as the price as such is not transparentFootnote ⁵⁵; the procedural (formal) premise supersedes the substantive one. However, this is a weak policy choice, since the directive provides only a minimum of harmonization, and some member states have opted for allowing substantive review even if the price was not found opaque.

The detailed considerations that underpin the application of the UCTD in the context of personalized pricing are discussed in Chapter 7. Building on a thorough analysis of the economic premises of algorithmic price control, they conclude that the existing instruments are of limited applicability in this regard. Most importantly, they do not see how it is feasible to construct convincing benchmarks (such as an average market price) that could allow the conventional contract law toolbox to apply. Instead, they advocate seeking outside the contract law domain, and indeed outside the perspective of individual consumer-to-business relations.

Second, EU law reinforces the disclosure-based paradigm even further by indicating, in Article 6(1)(d) of the UCPD,Footnote ⁵⁶ that what constitutes a misleading commercial practice is not informing, or at least deceptively communicating to, a consumer about ‘the price or the manner in which the price is calculated, or the existence of a specific price advantage’. The lack of proper disclosure amounts to an unfair practice only as long as it may substantially distort the client’s market behaviour (i.e., when it ‘causes or is likely to cause him to take a transactional decision that he would not have taken otherwise’). This provision, paired with the UCTD, quite conspicuously indicates that EU contract law concerns itself mostly with the question of the procedural fairness of algorithmic pricing, rather than with its substantive outcome. In general, under EU law, price personalization as generally practiced is acceptable as long as the fact of personalization has been properly disclosed to a consumer. Finally, Article 4(4)(a)(ii) of the ‘Omnibus’ DirectiveFootnote ⁵⁷ takes the most straightforward step towards algorithmic pricing transparency. It mandates that all traders who apply such pricing techniques must inform consumers ‘that the price was personalised on the basis of automated decision-making’.

This provision from the Omnibus Directive does not specify what it understands personalized pricing to be, which is quite perplexing, given the multitude of meanings discussed earlier.Footnote ⁵⁸ The essential question of transparency in algorithmic pricing, however, lies elsewhere. As Jabłonowska, Lagioia, and Sartor explain in their chapter, none of the three provisions discussed in the foregoing paragraphs indicates what particular issues should be taken into account in assessing the procedural fairness of algorithmic pricing. The only clear and universal obligation they impose is that providers must disclose the fact that the price was personalized by an algorithm. The exact parameters of price calculation, the internal architecture of the algorithm, and the pool of data used for algorithmic pricing need not be revealed to any extent. Even the UCPD’s provision, which could possibly mandate that business should reveal more detail about the premises of price calculation, is of negligible practical importance. It applies only to the extent that the lack of proper disclosure may negatively impact consumer market behaviour, which admittedly will not be the case for every situation in which consumers do not receive detailed information about the algorithmic pricing scheme. Quite the opposite: in many instances, awareness of the details may be much less important to consumers than knowing that the price they are paying is not excessive compared to the market average.

Consequently, as Bagchi’s contribution to this volume elucidates, the optimal disclosure model should go further than merely informing consumers that the price was calculated in a personalized way. It also should provide consumers with a certain degree of clarity as to the extent to which their individual features have contributed to the final price. Hence, as Bagchi explains, a well-conceived duty of disclosure should ‘satisfactorily alert consumers as to the personal characteristics reflected in their prices, each one of which plausibly informs buyer’s decision to buy’.Footnote ⁵⁹ Her analysis is motivated by a concern for making algorithms transparent to consumers, which is one of the key challenges posed by personalized pricing.

However (and as Bagchi, as well as Jabłonowska, Lagioia, and Sartor rightfully point out), one must be mindful that providing complex information to consumers has obvious limits. Even if consumers could act upon it in theory, one can never be sure what actually drives their behaviour.

Moving away from the concern for price fairness leads to emphasizing mostly (if not exclusively) the right to know the impersonal price. According to this view, instead of knowing the parameters used to calculate the personalized price, it is more attractive to make sure consumers know the impersonal price along with the personalized price. After all, the material information is the price to be paid, not how it was calculated. As we explain later, one way, and it is not without challenges, to fulfil this blueprint may be to disclose an impersonal price to consumers.Footnote ⁶⁰ This information seems simpler both to provide and to take into account, and it seems more informative than the main parameters leading to a personalized price.

To illustrate, suppose a website offers identical tablets to each editor of this volume, one for 549.99 EUR (Mateusz) and the other for 489.99 EUR (Fabrizio). Each is told that the price was calculated taking the following parameters into consideration: age, profession, country of residence, and familial status. It is apparent that this information is not very useful for either of us. Instead, suppose we are both told that the impersonal price is 489.99 EUR. Fabrizio is reassured; Mateusz becomes alarmed that he is being asked to pay more than the impersonal price, and at least he can exit the website and return, this time denying access to his personal data and thus being offered the impersonal price of 489.99 EUR.Footnote ⁶¹ For this reason, the impersonal price appears more important from a consumer protection perspective than the personalization parameters.

Be that as it may, disclosure duties (even if more sophisticated than under the current EU model) should only be part of a much more complex regulatory system. Consumer protection based merely on information mandates can in many instances prove illusory if not coupled with meaningful market choice for the consumer. As explained earlier,Footnote ⁶² even when properly informed about the basic parameters of the price calculation process, consumers may not be able to find a substantially different option. For these reasons – as several contributors to this volume point out – supplying consumers with information must be coupled with other regulatory instruments, in particular with regulating competition and data flows on the market. From this vantage point, price personalization comes with major caveats against the clustering of legal analysis into conceptual silos and strongly incentivizes systemic thinking.

I.3.4 Efficiency and Distribution

Applying the traditional economic categories to price personalization poses another clear issue. The traditional approach to price assessment harbours a tension between the efficiency of algorithmic price personalization and its distributive effects. Under this view, algorithmic price personalization can increase market access and therefore total economic welfare, which is normally taken for a synonym of increasing the efficiency of market allocations. However, while some consumers may benefit from algorithmic pricing, others will lose – and the aggregate effect is therefore hard to measure. The inability to go beyond this finding hampers some scholarly analyses of personalized pricing and seems to impede the search for concrete and actionable conclusions.

The specific challenges of studying algorithmic prices can be eased by following two complementary routes. The first hinges upon the notion of institutional necessity. In other words, it asks to what extent it is necessary to harm some consumers to increase market access for those consumers who are offered a lower price. As pointed out by Brinca, Costa, and Martinez,Footnote ⁶³ nothing in the mechanics of algorithmic price personalization makes harming some consumers necessary to ensure that other consumers benefit from algorithmic price personalization in the form of lower prices. Indeed, as several chapters in this Handbook point out (especially Chapters 6 and 9–13), legal systems are willing to balance a practice’s positive against its negative effects if the negatives are found necessary to obtain the positives; otherwise, there is no need to balance them.

This point might be especially hard to grasp in the traditional economic analysis of law because – as recently pointed out by Bayern – the field’s methodology is overshadowed by an ‘analytical failure’ in failing to consider more nuanced institutional arrangements.Footnote ⁶⁴ In the context of algorithmic price personalization, this analytical failure manifests in not taking into account institutional arrangements in which price personalization for some consumers is allowed only if it is beneficial to them, regardless of the impact of the practice on other consumers. In other words, trade-offs between consumer groups are unnecessary.

Davola, Esposito, and Grochowski offer some insights in this regard. They expose two crucial limitations to the idea that price personalization should be considered necessary every time traders invest in the personalization of other contractual terms. The first is that the market these traders are operating in is perfectly competitive, because when it is not, traders need not engage in price personalization in order for term personalization to be feasible. Second, the market under consideration is one in which the contract has no relevant effect on third parties. To illustrate, the authors point out that risk distribution in insurance markets may justify limiting price personalization in light of socially desirable distributive effects. A case in point is the law forcing insurance companies to ignore gender as a variable in calculating premiums.

Nevertheless, traditional economic analysis can be invoked to insist that consumer harm is irrelevant or less relevant than pure efficiency considerations (in the total welfare maximization sense). The strong version of this approach struggles to acknowledge the primacy of consumer interest. Notably, this strong version is challenged in its very conceptual foundations by the economic view that consumer welfare is the only relevant consideration for judging the effects of the relative efficiency of market outcomes. According to this rediscovered view, the distributive concern is part of the efficiency concern, not complementary to it.Footnote ⁶⁵ From this perspective, it becomes apparent that there is no such thing as a merely distributive concern about the impact of algorithmic prices in the relationship between consumers and traders: every cent, penny, real, rupee, and yuan traders make by harming some consumers is a cause of inefficiency, and therefore an alternative market arrangement without such harm is more desirable in pure efficiency terms.

The weaker version of this total-welfare approach consists of an institutional shell gameFootnote ⁶⁶ in which, when analysing legal route A, it is determined that the issues are best left to legal route B to address. Then, the analysis of route B leaves it up to route C, and so on, up to an unavoidable outcome either of going a route where there is no political will to address the issueFootnote ⁶⁷ or of closing the circle by once more indicating route A as the relevant one.

I.4.1 Framework

One of the key factors motivating the contributions to this volume is mapping the existing approaches to price personalization from analysis of actual legal systems. EU law plays a prominent role, but some of the most insightful findings come from other jurisdictions. They allow us not only to understand various regulatory mindsets in response to algorithmic pricing but also to look critically at the view that the EU is the global champion in regulating digital commerce. Personalized pricing makes a vivid case for a more contextualized approach towards comparative studies in the domain of online contracting, digital platforms, and consumer privacy protection.

In exploring the regulatory options for personalized pricing, the Handbook focuses on individual consumer transactions and competition issues. In so doing, it spotlights making personalized prices per se fair (which so far has also been the main gird of the discussion on algorithmic price calculation globally).Footnote ⁶⁸ The analysis does not delve into instruments that may mitigate the negative consequences of price personalization after a contract is formed. One of these instruments is taxation, which in the case of personalized pricing may follow one of two models. In particular, a tax is conceivable that redistributes directly to recover the ‘consumer surplus’Footnote ⁶⁹ from the firms applying personalized pricing and redistributes it back among consumers. In this way, every firm would be entitled to a certain profit margin below the total profit that could be accrued through personalized pricing, while the rest of the revenue is clawed back as a ‘consumer surplus tax’.

The jurisdictions covered in this volume have been chosen along two lines. First, our aim was to investigate the legal systems that have explicitly engaged with algorithmic price personalization (the EU, China, and Brazil). Second, as a counterpoint to the latter, the volume also discusses jurisdictions in which, despite the lack of direct regulatory measures for algorithmic pricing, the issue has entailed several difficulties and challenges, coupled with scholarly and policy output.

Contributors to this volume have been asked to take as a benchmark a framework of three questions, which can now be used as tertium comparationisFootnote ⁷⁰: (1) does a right to an impersonal price exist in the considered legal framework? And do consumers know it? (2) Do consumers have a right to opt out of the personalized price in favour of the impersonal one? If so, how easy is it to opt out? (3) Is there a way to avoid manipulation of the impersonal price?

I.4.2 Focus: Impersonal Price

The expression ‘impersonal price’, to the best of our knowledge, was introduced into this debate by Esposito.Footnote ⁷¹ Nevertheless, Carlton used it in the 1989 edition of the Handbook of Industrial Organizations, in which he wrote of ‘impersonal markets’ and ‘impersonal prices’ to refer to situations in which traders rely only on consumer information derived from ‘an impersonal (auction) market that clears by price alone’.Footnote ⁷² Indeed, this is the intuition behind the idea of the impersonal price as the opposite of the personalized price; or, more precisely, it is ‘a price that was not set on the basis of the processing of personal data (including automated decision-making)’.Footnote ⁷³ Legal systems traditionally recur to the idea of a market price or value, often to fill contract gaps. For example, the Vienna Convention on the Sale of Goods establishes that if the parties have not determined the price, then it is assumed they have ‘made reference to the price generally charged at the time of the conclusion of the contract for such goods sold under comparable circumstances in the trade concerned’ (Article 55). Similarly, under the Draft Common Frame of Reference (DCFR),Footnote ⁷⁴ the price can be determined by looking at ‘the price normally charged in comparable circumstances at the time of the conclusion of the contract or, if no such price is available, a reasonable price’ (II. – 9:104 DCFR).

The main argument raised against this benchmark is its alleged impracticality. A general market valuation of a good or service cannot be established, goes the critique, with a sufficient degree of certainty.Footnote ⁷⁵ Hence, it would be unreasonable to rely on impersonal prices as a rubric for market regulation or for the review of prices in particular agreements. But the following observation could be made in counterargument: suppose that an online shopper refuses to consent to the processing of their personal data for the purposes of personalizing the price. The price offered to said person is impersonal; personalized prices higher than said price are personalized surcharges, and those lower are personalized discounts. In reference to the traditional legal solutions mentioned earlier, issues of time and place – which would relate to dynamic pricingFootnote ⁷⁶ – could be accommodated easily into the determination of the benchmark. Or, at the very least, the determination of this benchmark does not seem to encounter challenges much different from those customarily faced in an unfair-price case under competition lawFootnote ⁷⁷ or from the challenges of determining the usury interest rate.Footnote ⁷⁸

The concept of impersonal price can be a promising avenue for regulating algorithmic pricing practices. This triggers the key question of how the disclosure duties for impersonal consumer prices should be designed. The first intuition suggests that disclosing the impersonal price together with the personalized one is a simple and effective way to limit the diffusion of personalized surcharges.Footnote ⁷⁹ In other words, as is often the case, this book reflects on the best possible transparency-based regulatory intervention before venturing into substantive controls.

The literature has been paying primary attention to comparisons between the particular personalized price offered to a person and the personalized prices paid by others.Footnote ⁸⁰ Upon reflection, it becomes apparent that the impersonal price is the price being paid by a specific other: the perfect stranger. Instead of having to discuss the various complexities (statistical, design, cognitive) or the usefulness of disclosing the price paid by everyone else or by an average consumer, the simplicity of the impersonal price seems worthy of consideration.

Moreover, and somewhat contradictorily, the policy analysis considers the possibility of providing an opt-out from personalization,Footnote ⁸¹ which indeed logically presupposes the existence of a price to fall back on without personalization, which it seems will have to be the impersonal price. Moreover, once the right to opt-out is considered, it is apparent that for the consumer to make an informed decision about opting out, the alternative should be known.Footnote ⁸²

The last component of the analytical framework aims to address the strategic response of traders. As Jabłonowska, Lagioia, and Sartor put it in this volume, ‘The impersonal price may be misleading when set to an artificially high level, just in order to convey the idea that the proposed personalised prices are highly convenient’.Footnote ⁸³ The goal is to find a price that is somewhat meaningful and that can be used as reference point for calculating the level of discounts and surcharges. Something similar is already provided by EU law for discounts: a discount must be calculated using as a benchmark a ‘reference price’, which consists of the lowest price applied by the trader in the previous thirty days (Article 6a(2) Directive 98/6/EC). Even more specifically, Portuguese law introduces a price cap for dynamic pricing in platform-based local transportation (Uber, Lyft, Bolt, etc.), where the price increase is limited to 100 per cent of ‘the average price charged for the services provided in the immediately preceding 72 hours’ on that platform (Article 15(5) Law 45/2018 of 10 August). A claim along similar lines has also been advanced by marketing scholars,Footnote ⁸⁴ who have lamented the diffusion of ‘fictitious prices’ and advocated for the disclosure of the ‘true normal price’, namely the most commonly applied price in the relevant business period.

I.4.3 Findings

Based on Chapters 9–13, the surveyed legal systems can be divided according to the replies to the three questions posed earlier: an explicit affirmative answer (EA); an affirmative answer via a controversial interpretive and integrative legal analysis (IA); an explicit negative answer (EN); a negative answer via a controversial interpretive and integrative legal analysis (IN); and no answer in light of the proposed analysis (NA).

The following two tables are designed to reflect these categories. Table I.1 looks at the EU as a whole and compares it with the other jurisdictions surveyed in Part III of the Handbook. Table I.2 zooms in on the different branches of EU law as surveyed by the different chapters. Jurisdictions are presented in rows, while the three columns refer to the right to the impersonal price; the right to opt out; the right to the integer impersonal price.

The findings on EU law have been aggregated and included in Table I.1 in the following way: the conclusion that a right exists in one branch of EU law is sufficient to find that said right exists. This same approach also informs, albeit implicitly, the analysis of the other jurisdictions included in Table I.1.

Table I.1 shows that the system that most clearly endorses the regulatory framework described in Section I.4.2 is the Chinese one. China recognizes both a right to the impersonal price and a right to opt out. The recognition of these rights derives from a series of ad-hoc provisions introduced through different legal instruments, which, as Ge points out in his chapter, raises some coordination problems.Footnote ⁸⁵ EU, Canadian, and US law appear to be similarly situated in terms of rights and their origins and limits. In fact, like in the Chinese legal system, only the right to the impersonal price and the right to opt out have been identified. But the list of similarities continues, in that the origin of the rights in each case is a careful analysis of the privacy law framework. The right to an impersonal price can be inferred from an analysis of privacy law for the Indian legal system as well. Regarding US law, the finding is built on the analysis of California’s privacy law and is therefore limited in its geographical ambit. Moreover, in each case, the right to opt out is narrow in scope, in that it only allows consumers to limit access to their data after it has been used for personalization. The outcome is that consumers can access the service again (or stay within a contractual relation) without giving access to personal data. This is admittedly a cumbersome procedure that also seems to be quite lacking in terms of effectiveness of the right.