I. Why Exploitative Self-Preferencing Requires More Attention
While self-preferencing behaviour in digital markets has typically been studied in light of exclusionary abuses of dominance, there is an increasingly relevant need to devote sufficient attention to exploitative self-preferencing behaviours as well. Digital platforms such as Amazon have amassed significant market share over the years-it is amply clear that, at the very least, the status quo will be maintained in the years to come. As mentioned before, these digital markets are characterised by strong network effects which amplify lock-in effects. With the vast scope for personalization made possible by the leveraging of cross-sectoral advantages, digital platforms are able to reinforce market power with the aid of algorithms and unprecedented collection of data.
The Amazon Buy Box case serves as a pertinent example in this regard. Amazon came under scrutiny for extending preferential treatment to marketplace sellers who use Amazon’s delivery and logistics services (Fulfilment by Amazon/ FBA). Even though sellers had the option of contracting with other logistics providers, they were at a disadvantage since Amazon had tied visibility and marketplace sales to the usage of FBA. Success in the Amazon marketplace was also dependent on other factors such as performance metrics; these metrics were not homogenous for FBA-affiliated sellers and those who were not using FBA. Particularly for companies with multiple lines of business, the ability to cross-leverage market power has allowed for exponential growth, both horizontal and vertical.
Due to informational asymmetry, complementors may not realise in a timely manner that any detrimental impact to online revenues is arising from self-preferencing strategies. But even if they do, it may not always be feasible for complementors to leave the marketplace for many reasons: for instance, they could face high switching costs due to lock-in effects, prohibitions on interoperability, contractual restrictions etc. As a result, barriers to exit are raised considerably.
For complementors who choose to stay on, imbalanced risk allocation between the
platform and the former could detract from the extent of independence that complementors should ideally exercise while making commercial decisions. This is exemplified by Google’s self-preferencing in the Google Search (“AdSense”) case. By imposing certain contractual obligations in its dealings with online publishers, Google ensured that prime advertising space was not available to its competitors based on an understanding of consumer choice architecture. Consumers are more likely to notice advertisement located at the top of any website, therefore Google sought to take advantage of this through premium placement conditions embedded in exclusivity clauses to boost profitability. The gold dust trailing out of the pockets of online platforms’ customers, thereby becomes all the more important in the age of algorithms and data wherein digital platforms (especially those who already wield significant market power) are willing to resort to anti-competitive means to ensure that they capture a greater share of attention markets- of which the advertising industry exists as
an integral subset.
II. Looking Forward
Countering self-preferencing with the imposition of an all-encompassing or per se ban
would prove to be destructive towards business ecosystems which have come to rely upon these methods as a means of monetization and revenue generation. Since online dual-role platforms like Amazon understand that there is inevitably a trade-off to be made between revenues generated from self-preferencing and revenues from platform traffic, consumers stand to benefit from self-preferencing insofar as the former realize that consumer value is dependent on this trade-off. For instance, economic analyses have shown that self-preferencing has competitive effects on both ends of the spectrum. Even if a particular method of self-preferencing has anticompetitive effects, they must be juxtaposed against any pro-competitive effects generated by this self-preferencing behaviour.
A case-by-case analysis is thus required in order to ensure that consumer welfare is not compromised in the pursuit of stringent regimes. It must also be noted that competitive markets upstream give rise to a greater probability of procompetitive effects. As a result of the aforementioned trade-off, digital platforms would be less incentivized to exploit their current consumer base if there was a possibility of future growth that seems more lucrative to pursue.
It would be wise, therefore, to adopt a more comprehensive approach to self-preferencing- perhaps one which establishes a distinction between the two types of self-preferencing. The first type is conceived of as those methods of self-preferencing which have been perpetuated as a tool of ‘legitimate self-promotion.’ On the other hand, the second type covers those forms of self-preferencing which give rise to exclusionary effects that can only be justified on anticompetitive grounds. The former has generally been adopted by brick-and-mortar stores without being per se considered anticompetitive. In fact, self-promotion may be useful in exposing consumers to a wider range of choices; on streaming services such as Netflix, where 80% of hours of content streamed are discovered through recommendation algorithms, self-preferencing cannot be considered per se anticompetitive. However, where platforms transcend self-promotion to actively exclude or demote competitors, this must be targeted and countered by regulatory intervention. This active exclusion can be undertaken through the following means: rescission of interoperability between products/services which were previously interoperable, or through a failure to provide a specific competitor with the same treatment as is provided to others.
More generally, it also needs to be acknowledged that entities are not obligated to engage in the sharing of competitive advantages. But while arguments along the lines of
consumer welfare can be made to substantiate support for self-preferencing, Lina Khan has succinctly captured why current metrics of consumer welfare are myopic and insular. Integration at various levels of the supply chain enables these platforms to function with a level of exactitude that is not fully reflected in competition laws across multiple jurisdictions. Adopting a simplistic understanding of consumer welfare is particularly counterproductive in the case of algorithmic abuse of dominance through self-preferencing as it could be argued that notwithstanding any exclusionary/exploitative effects, overall consumer utility increases as a result of
efficiency advantages.
Current definitions of consumer welfare must be sufficiently reflective of new theories of harm if consumer interests are to be truly protected; where procompetitive forms of self-preferencing arise, regulators must be able to distinguish these from anticompetitive self-preferencing on the basis of legal distinctions supported by economically sound understandings of consumer welfare. This distinction becomes particularly important in the cases of countries such as the EU, USA, and India where laws specifically encompassing self-preferencing are still in a state of flux.
From the standpoint of seeking to promote innovation, concerns have been expressed over the inadequacy of technical tools of inquiry such as vertical and horizontal integration analysis. With an increasing number of market behemoths such as Amazon, Google, Disney etc. foraying into different industries, methods of integration are no longer restricted to two dimensions. There is a stark possibility that the strengthening of market power will be undertaken through the manipulation of the “multiple nodes of dependency” established by these platforms.
In this light, self-preferencing strategies can be particularly detrimental to emerging markets with smaller market players who may, to differing extents, be dependent on any one or more lines of business of these platforms. Khan has also expressed the need for antitrust remedies to be reflective of the ways in which markets reward various business tactics: the speed at which remedies can effectively counteract self-preferencing behaviours is of vital importance since they can lead to exponential market visibility, growth, and retention of dominance at an unfair cost to smaller complementors. Where it can be shown that the procompetitive effects are overshadowed, regulators must- as mentioned before- be in a position to respond to self-preferencing speedily and effectively.
Conclusion
While the corpus of competition law as it stands today in various jurisdictions across the world is definitely not obsolete vis-à-vis digital markets, classical tools and methodologies may not be entirely adequate to challenge market positions of dominance established by powerful actors in digital markets. In this light, the significance of data, network effects, and cost savings are important factors which compel regulators to revamp competition laws to be reflective of the nuanced issues arising from the proliferation of algorithms in digital markets. Beginning with a brief exploration of how three major antitrust jurisdictions have so far attempted to regulate self-preferencing, this article has briefly explored the economic and legal implications of self-preferencing. What is common across these jurisdictions is the fact that there is not much scope to weigh the procompetitive effects of self-preferencing against its anticompetitive effects. Moreover, the exploitative effects of self-preferencing have not been explored adequately- at least in comparison to its exclusionary effects. Regulators across the world thus need to take these factors into consideration while establishing newer legislations which seek to tackle the problems posed by self-preferencing- in addition to updating current understandings of consumer welfare.
This article is authored by Mathanki Narayanan and Srinithi Shankari , 4th year and 5th year law students at OP Jindal Global Law School and Symbiosis Law School, Hyderabad respectively.
To view Part I of this article, Click here
