Algorithmic Anti-Discounting in F.T.C. v. Amazon: Neo-Brandeisian Lessons for Competition Law

[Chytanya S Agarwal is a 3rd-Year B.A., LL.B. (Hons.) student at the National Law School of India University, Bangalore]

Recently, the American Federal Trade Commission (‘FTC’) sued Amazon alleging that its practices unlawfully maintained its monopoly. The complaint was based on two main grounds – first, the algorithmic processes used by Amazon to curate its search results penalise the retailers who offer lower prices in other platforms by hiding their products in search results; and second, by coercing retailers into availing Amazon’s expensive fulfilment services for obtaining ‘Prime’ eligibility.

In this post, I limit my analysis to the first ground called ‘anti-discounting’ by the FTC. Using the background of FTC’s action against Amazon in the United States (‘US’), I argue that the Indian competition law does not adequately safeguard against Amazon’s anti-discounting measures. This gap, I conclude, can be remedied by incorporating the Neo-Brandeisian undercurrents of the FTC complaint. To make this argument, firstly, I briefly break down the reasoning in the FTC’s complaint. Secondly, I contextualise anti-discounting in India by analogising it with ‘price parity clauses’.  Lastly, through a perusal of case law, I argue that the Indian Competition Act (‘the Act’) and its jurisprudence inadequately account for such anti-competitive behaviour. Such legal gaps, I submit, should be filled by incorporating the Neo-Brandeisian undercurrents of the FTC complaint. Put simply, the Neo-Brandeisians argue that competition law and policy must widen its focus (from prices, efficiency, and consumer welfare) to become more structure-oriented to prevent any form of monopolistic market concentration.

The Main Findings of the FTC

Anti-discounting, according to the FTC complaint, refers to the strategy of penalising retailers who charge lower prices in other online marketplace platforms. The FTC notes that despite similar practices (such as price parity clauses) being discarded, Amazon effectively undertakes anti-discounting using algorithms. Per the FTC, Amazon’s Project Nessie algorithm ensures that a seller’s product is pushed towards the end of the search results if the algorithm detects lower pricing by such seller elsewhere. This makes such retailer’s products virtually hidden for majority of consumers who do not go beyond the first few search pages. By disincentivising retailers from charging lower prices elsewhere, the algorithm maintains the perception that Amazon offers lower prices than its e-commerce competitors. Consequently, the rivals of Amazon are unable to undercut the prices charged by it and the prices are not reflective of the market clearing price in competitive markets (see paragraphs 326-332 of the complaint).

The second malpractice alleged is that Amazon has mandated the subscription of Amazon’s Fulfillment (or transportation) services (‘Fulfilled By Amazon’ or ‘FBA’) for any seller in order to be ‘Prime’ eligible. The FTC stresses that, for a retailer, ‘Prime’ eligibility is necessary to access Amazon’s large customer base of ‘Prime’ members. In this regard, the data adduced by the FTC shows that large numbers of retailers who had forgone FBA ended up being wiped out from the storefront (see paragraphs 358-360 of the complaint). This practice had the anti-competitive effect of restricting market access for those retailers who did not opt for the costly services of FBA. Additionally, it raised the operating costs of the retailers who had no option but to opt in for FBA (see paragraphs 364-367), and harmed other competitors in the market for online retail fulfilment services (see paragraphs 375, 387-394). Cumulatively, anti-discounting and mandatory requirement of FBA for ‘Prime’ eligibility have worked to entrench Amazon’s monopoly power in two relevant markets – the market for online superstores and online marketplace services (see paragraphs 410-415).

Contextualising FTC’s Findings in the Light of Competition Act, 2002 in India

Similar to anti-discounting, online platforms earlier used ‘price parity clauses’ in their contracts with every retailer. ‘Price parity clauses’ contractually restrict the retailers from offering lower prices in competing platforms. Such clauses are shown to have anti-competitive effects (see here and here) because they act as entry barriers for other low-cost entrants in the e-commerce market. They also distort the prices across platforms, adversely impacting competition and reifying the dominant position of the certain marketplace entities. Such clauses have been banned in several jurisdictions. Illustratively, Article 5.3 of the EU’s Digital Markets Act prohibits a ‘gatekeeper platform’ from preventing its ‘business users’ from charging lower prices at other platforms.

In the Indian context, price parity clauses were flagged for the first time by the 2020 Market Study on E-Commerce (see paragraph 93) by the Competition Commission of India (‘CCI’). Later, in Federation of Hotel & Restaurant Associations of India v. MakeMyTrip India Pvt. Ltd. (MMT case’), the CCI penalised OYO and MMT-Go for using anti-competitive price parity clauses in their business agreements with hotel partners. However, it exempted a class of parity clauses – called narrow price parity clauses. Such narrow parity clauses restrict lower pricing only on the retailer’s own website and not in other online marketplace platforms. Under the test laid down in the MMT case, one must undertake a comparative assessment of the pro-competitive and anti-competitive effects of a strategy that ensures price parity. On such analysis, if the pro-competitive effects of narrow parity clauses exceed its anti-competitive effects, then they are deemed justifiable on grounds of market efficiency (MMT case, paragraph 265). However, this reasoning ignores that the anti-competitive effects of price parity strategies are mostly producer-centric whereas its pro-competitive effects are predominantly consumer-centric.

In the Indian context, price parity clauses would be governed by section 3 of the Act. For instance, the MMT case’s reasoning hinged on the presence of an agreement for price parity (in paragraphs 235, 291). Interestingly, as alleged by the complaint in FTC v. Amazon, anti-discounting price parity is not contractual but algorithmic, excluding the application of section 3. Even if we assume that there existed an agreement/arrangement/understanding between Amazon and the sellers regarding price parity, the causation and motive for anti-discounting would be extremely difficult to explain if Project Nessie employs inductive and deep learning AI systems.

If a section 3 enquiry is precluded by the absence of any agreement, the only safeguard against anti-discounting is section 4(2) of the Act. First of all, section 4 requires Amazon to be in a “dominant position” – a requirement that is negated by the existing precedents. In Lifestyle Equities v. Amazon Seller Services Pvt. Ltd. (paragraph 23), Amazon was held to be not in a dominant position. Nonetheless, Lifestyle Equities dealt with a more restricted relevant market, namely, the “market for services provided by online platforms for selling fashion merchandise in India.” In contrast, in the FTC complaint, the relevant markets was more widely defined as “markets for online superstores and online marketplace services” (see paragraph 24 of the FTC complaint). However, even with respect to the e-commerce market, in All India Online Vendors Association v. Flipkart, the contention alleging Flipkart’s dominant position was rejected, despite its market share in the e-commerce industry being around 50%. As observed in Delhi Vyapar Mahasangh v. Flipkart and Amazon (in paragraphs 10,15), Amazon owns a much smaller market share compared to Flipkart, and there exists no joint dominance by both. A conjoint reading of All India Online Vendors Association and Delhi Vyapar Mahasangh eliminates the likelihood that the status of “dominant position” would be accorded to Amazon.

In the Indian context, going by the standard applied in the MMT case, a comparative assessment of the anti-competitive and the pro-competitive effects of such strategy would have to be undertaken. Even in Matrimony,com v. Google LLC (at paragraphs 437-440), where search bias was penalised by the CCI, the analysis rested on a trade-off between pro-competitive and anti-competitive effects of Google’s actions. This standard of search bias in can be applied to Amazon since anti-discounting also perpetuates a form of search bias. However, this still leaves ample scope for condoning anti-discounting behaviour if its pro-competitive effects, which are mostly consumer-centric, exceed the anti-competitive effects on sellers and other online e-commerce platforms. Moreover, according to Sanwar Mal Agarwal v. Punjab National Bank, the standard for proving abuse of dominant position is a high and onerous burden. As noted earlier, the algorithmic nature of Amazon’s anti-discounting is more difficult to prove due to the ‘black box’ problem, or the inherent difficulty in explaining the internal functioning and motives of AI systems. This further weakens the potential for a section 4 remedy to anti-discounting.

Some Concluding Neo-Brandeisian Lessons

As mentioned earlier, the Neo-Brandeisian economists view monopoly and market concentration as detrimental in the long run. The Neo-Brandeisians focus on the structures and processes of market competition and not its outcomes (such as efficiency, pricing, and consumer welfare). Consequently, they are also averse to a consumer welfare view of competition law, since several monopolistic structures (such as predatory pricing) may benefit consumers greatly yet have an adverse impact on sellers. Thus, a Neo-Brandeisian view of antitrust law would view arrangements (such as anti-discounting) that seek to eliminate price competition as unlawful.

According to Lina Khan, the current FTC chairperson, antitrust regulation of the e-commerce industry is better suited to a Neo-Brandeisian outlook due to two reasons: first, platforms such as Amazon prioritise growth over profits; and second, such online marketplaces function as critical market intermediaries. Due to these reasons, the current view of competition laws, exemplified by the insufficiency of the Indian Competition Act in accounting for algorithmic anti-discounting, risks downplaying the long-term risks of market concentration. As a course correction, the Act and its jurisprudence must, apart from comparing the pro-competitive and anti-competitive effects of anti-discounting and price parity arrangements, outlaw such market practices and structures for being per se frustrative of market competition and competitive pricing. This Neo-Brandeisian view also aligns better with the studies that show the dangers of allowing even narrow price parity, despite the fact that their pro-competitive effects may seem to be greater than the anti-competitive effects.

Chytanya S Agarwal

About the author

Add comment

Top Posts & Pages


Recent Comments


web analytics

Social Media