From an economic standpoint, personalised pricing is not a novel (theoretical) concept. However, this practice has become topical thanks to digital technological developments that make it actually feasible, even if there is very little evidence that the feasibility of personalised pricing has led to its widespread implementation so far.
The current debate explores the circumstances in which intervention under not only competition law, but also consumer law and data protection law, would be warranted. The focus is primarily on exploitative outcomes under imperfect competition, whereby firms with substantial market power charge consumers high prices that could be deemed excessive and/or unfair. There is a consensus that enforcement against such practices would be challenging. For example, it is not straightforward to establish under a consumer welfare standard that consumers are in aggregate worse-off under personalised pricing. This is because personalised pricing can entail lower prices for consumers who would otherwise not buy the product in question, thus leading to a welfare enhancing output expansion.
Whilst this ambivalence in terms of welfare implications would make enforcement contentious against a dominant company, the case for intervention ought to be more straightforward under joint monopolization. Finding that this exploitative outcome is underpinned by a collusive agreement would likely override any efficiency enhancing defence.
This article, available here. provides a comprehensive analysis of when the parallel adoption of personalised pricing (or lack thereof) across rival firms could be indicative of the existence of collusion and, thus, within the scope of competition law. In short, the author finds that, in a competitive market, personalised pricing will likely be the result of collusion when it relates to various firms being able to identify customers’ willingness to pay. Collusion is also likely when firms adopt uniform pricing despite horizontal product differentiation.
Section II looks at how to identify collusion under personalised pricing.
The economic analysis of price discrimination under non-coordinated imperfect competition (i.e. oligopolistic competition) helps one to identify those conditions where an observed market outcome is incompatible with ‘normal competition’. In particular, they assist one in distinguishing between coordinated and non-coordinated collusion – including as regards personalised pricing.
From the demand side, personalised pricing typically requires the firms to be able to distinguish between customers’ willingness to pay. This willingness to pay can be influenced by horizontal product differentiation – whereby consumers may not only exhibit a general preference for variety, but also differ as to which variety they like the most (at the same level of quality and price) – and the demand-side frictions that can emerge from high search and switching costs – since firms are aware that they can charge consumers with high search and/or switching costs comparatively higher prices, whilst still retaining them.
From a supply side perspective, the ability to take advantage (i.e. in a non-coordinated way) of observed differences in customer willingness to pay depends on the type of strategic interaction that prevails among rival firms – in particular whether rival firms agree about which consumers can be charged a higher price. The same applies to their perception of demand-side frictions that can allow them to charge higher prices. In contrast, preference for the product’s of one firm, or for specific product, entails a specular permanent disutility when buying from a less preferred firm. Rival firms would therefore have to try to defeat this preference by means of offering lower prices, which leads to a lower propensity to collude on personalised pricing.
Under normal market conditions, rival firms should not be able to take advantage of different willingness to pay due to the threat of price undercutting between them (i.e. as under Bertrand competition). Therefore, rival firms’ ability to identify heterogeneous levels of consumers’ willingness to pay ought to be inconsequential under competition – and the existence of price discrimination is indicative of collusion. This conclusion can be qualified in a number of circumstances, such as the presence of demand-side frictions related to search costs (e.g. the client is time poor and relies on branding).
However, the situation may be different when there are high levels of product differentiation. Firms are collectively better-off under uniform pricing, but each firms has a unilateral incentive to target rivals’ customers with poaching offers tailored to the corresponding degree of brand preference. As a corollary, the fact that rival firms stick to uniform pricing, when it is feasible and convenient to adopt personalised pricing (or even less granular forms of price discrimination), may be indicative of the presence of a collusive agreement.
Section II also devotes particular attention to how personalised pricing will typically not be collusive when customers have different search costs.
It is well established that consumer search costs lead to prices being set above competitive levels. This exploitative effect is most pronounced when consumers have low levels of preference for variety (i.e. the product is homogeneous) and relatively high search costs. An example of such a situation is provided by diamonds, hence this effect is sometimes called the ‘Diamond paradox’. When buyers face search costs, they rationally anticipate that all firms will take advantage of this situation by setting high prices; as a result, consumers do not shop around. At the same time, firms lack the incentives to cut prices, since they anticipate that disillusioned consumers would not be on the lookout for better bargains. The result is that prices are set at a homogenous supra-competitive level.
The economic literature has also posited that such markets may have two different types of customers: costly searchers who are uninformed (also labelled ‘tourists’), and costless ‘shoppers’ who have a full grasp of market prices (also labelled ‘locals’). Assuming that firms are able to identify each type of customer, price dispersion emerges as the result of trade-offs between the opposing incentives of attracting ‘locals’ by undercutting rivals, and ripping off ‘tourists’. In any case, under both configurations the companies’ (differentiated) pricing behaviour may seem to be collusive, but does not result from firms coordinating their pricing conduct.
Section III looks at the implications of personalised pricing being used as a means to increase search costs.
In the discussion above, the presence of heterogeneous search costs is considered a given (i.e. exogenous) feature of the market. However, the adoption of personalised pricing could be aimed at artificially (i.e. endogenously) raising search costs. This could lead to the Diamond paradox resurfacing, whereby where firms are able to ratchet up prices up to monopolistic levels without inducing consumers to shop around.
This line of argument is reminiscent of the peculiar market configuration labelled ‘confusopoly’, whereby rival firms engage in price obfuscation in order to thwart (otherwise intense) pricing rivalry (as occurs in utility services such as retail energy, communications and finance). To rectify this anticompetitive outcome, various proposals have been considered – such as duties to place customers on the cheaper available tariff, or mechanisms to improve comparability. In the digital sphere, the adoption of personalised pricing could amount to a means of counteracting the ubiquity of third-party comparison services (e.g. companies could provide such services with the listing price, while in reality pricing at a personalised level via discounts). Such a pricing strategy would be equivalent to a conduct aimed at boycotting group bargaining/purchasing facilitated by digital agents. However, the sustainability of such a strategy is questionable, since there are strong unilateral incentives for firms to deviate from the common obfuscating course of conduct and undercut rival firms through third-party platforms, thus reinstituting intense pricing rivalry under uniform pricing (i.e. Bertrand competition). In any event, the common conduct of refusing to quote prices, whether personalised or uniform, to a third-party price aggregator would be contrary to ‘normal competition’, and thus indicative of collusion.
I have to admit to not having previously thought of personalised pricing outside the realm of abusive (exploitative) practices. This paper provides a very interesting overview of how to identify instances of collusion based on whether personalised pricing is to be expected or not, and outlines a number of theories of harm to boot.
Personally, I would have liked a more detailed discussion of how the market characteristics that the author identifies could be used as screens. Furthermore, the author does not articulate whether the theories of harm he identifies lead to collusive arrangements that would be prohibited per se or by object, even though he does not address whether there may be countervailing efficiencies at play. I think that engagement with these questions would have made for a more complete – and useful – paper in the end.