Digital Threats to Competition: Shaping Tools and Regulations for a Fairer Future

Third year law student at National Law Institute University, Bhopal

I. Introduction

Any kind of collusive practice in the market is prohibited by the Competition Act . The word “any” is wide enough to include any kind of activity which allows the competitors with reasonable degree to ascertain the pricing policy of other competitors. Even mere exchange of non-public competitive sensitive information (“NPCSI”) is enough to attract liability under Section 3(1) and 3(3) irrespective of the implementation of agreement based on such information as it takes away the independent decision-making ability of the participants as held in the Paper Cartel case.

With the advent of pricing algorithms (“PA”), collusion in the market has taken on a modern form raising serious issues about competitiveness. The impact of these algorithms on market dynamics is becoming a major concern throughout the world, as their influence on pricing tactics and customer choice has the potential to undermine fair competition in the digital marketplace. Further, they can mirror collusion like behaviour and facilitate transmission of NPCSI The recent case of RealPage highlights such concerns.

II. Realpage and the Algorithmic Web: A New Face of Market Manipulation

US department of Justice (DOJ) has recently launched investigation against RealPage citing competition concerns arising from the use of pricing algorithm which may lead to collusion-like effect in the market. Real Page is a company which sells revenue management software to landlords. The company amasses NPCSI from the landlords through its software and this combined data is then used by the algorithm to make pricing recommendations to the landlords. Further, investigations reveal that prices are not mere recommendations as RealPage also ensures compliance with these recommendations through automated setups on its platform.

The DOJ’s main concern is the collusion-like situation which occurs when landlords who use RealPage's software exchange NPCSI and set rents using the same algorithm. Instead of relying on independent judgment or market forces, landlord’s pricing is shaped by the software’s data, resulting in similar pricing strategies across the market. Even though landlords do not communicate directly with one another, the software's recommendations cause prices to rise in a coordinated manner which resembles collusion this becomes more concerning as the investigations reveal that landlords understand that sharing their data on the platform will provide them access to their competitor’s data and the same data will be used to influence pricing decisions for both themselves and their rivals.

III. From Human Intent to Machine Independence: Advanced Detection Systems and Regulatory Reforms

Various legal scholars and professionals are of the view that pricing algorithm can probably lead to collusion in the market and thereby distorting the market forces. There are studies indicating that PA might hinder competition by discouraging firms from undercutting prices, as competitors would immediately retaliate with similar reductions, maintaining prices above competitive benchmarks without direct collusion. In this regard it is important to note that algorithm pricing is mainly divided into two categories, human- relatable conduct and purely automated conduct. Cases of human related conduct can range from cases where the algorithm is itself designed in a way to promote collusive practices or where different competitors rely on third party for such algorithm which help them communicate NPCSI and streamline the collusive practices. This category falls closer to Hub and Spoke antitrust conspiracies. Hub and spoke conspiracy is where the competitors (spokes) are in no direct communication amongst themselves rather a third party (hub) acts as a rim to communicate amongst the spokes. In modern world the same can be facilitated through digital platforms or algorithms as in the present case of RealPage.

Such cases are covered under Section 2 and Section 3(3) which presumes such direct or indirect arrangements to be prohibited unless rebutted. The challenge lies with second category, involving purely automated conduct, since it relies entirely on algorithms to determine pricing strategies. These sophisticated algorithms can independently devise ways to align prices across competitors, potentially leading to tacit collusion without any human intent or direct communication. Thus, here the issue of determining the liability becomes crucial.

The concern deepens as companies utilizing PA often operate in a discreet manner. Detecting algorithmic pricing is, therefore, a vital first step in uncovering collusion or cartel activity. However, traditional detection methods may fall short due to the vast amount of data generated by these algorithms. Further, the enormous amount of price fluctuations and pricing data created by algorithmic pricing software makes it unfeasible to conduct audits without using automated systems. For example, recently Amazon was found to have implemented 2.5 million price changes per day. Currently CCI majorly uses market information, complaints, market studies and economic tools to detect and investigate anti-competitive behaviour. But given the unique challenges which algorithmic pricing driven market poses it is high time for India to have a technological tool which advance and specialise in detecting algorithmic collusions.

Several nations have already made steps to address these difficulties, the forerunner being UK's CMA which established the Data, Technology, and Analytics (Data) section in February 2019. The Data Unit's notable accomplishment is the creation of an internal program meant to detect retail price maintenance (RPM) schemes by finding anomalous patterns in scraped pricing data. The CMA intends to employ this technique to examine price activity in other industries for potential abnormalities. Other initiatives involve Columbia’s “Sabueso” project which uses bots to collect data and monitor pricing of products by different online retailers.

Given the challenges which algorithms poses in the markets, India must also come up with a legislation specifically focusing the risks associated with algorithm pricing. US has taken significant steps in this regard by proposing Preventing Algorithmic Collusion Act of 2024 which is still under consideration for its final approval. The abovementioned act regulates the use of PAs in the market and promotes transparency by mandating detailed disclosures from persons or platforms using PAs The act requires submitting detailed report to the commission regarding the information as to who is responsible for development of pricing algorithm along with role of third party if any. The act also requires clarification on whether the algorithm is programmed to recommend prices based on certain rules or pricing is done automatically through algorithm through deep learning which involves no human intervention. This is significant to determine liability in case of autonomous algorithm as discussed above. Further, the act also aims to regulate the design of the algorithm and requires information relating to the data sets used to train algorithm. Authorities may also pursue civil action if an algorithm is trained using non-public, competitively sensitive data. Additionally, they can initiate investigations if pricing practices resemble collusion.

Such steps are crucial to flag anti-competitive practices even if they are done under the garb of technological advancements. With the increasing use of pricing algorithm by companies like Amazon, Walmart etc. in Indian Market the need for such regulations is crucial.

IV. Digital Oversight in India :A New Approach to Unresolved Challenges

India is considering to bring Digital Competition bill which marks that India is aware of the unique challenges to competition in the digital world. The bill aims to prevent abuse of dominance by the companies on the basis of significant financial strength test and significant presence test and requires certain disclosures and compliance for the same. However, the bill does not specifically focus on the problems associated with algorithm pricing as discussed above. Therefore, the author believes the bill should also involve discussions related to algorithmic pricing taking inspiration from other countries while catering to the unique challenges faced by Indian market.

Further, it is crucial to focus on training the automated tools with good quality data from relevant market to effectively detect any collusion like activities facilitated through algorithm pricing. Also, rather than relying on the empirical methods to check algorithmic pricing tools which can only detect unusual patterns based on inputs and outputs, adoption of technical tools must be considered. This will be beneficial particularly to address the issue of ascertaining liability in cases of automated pricing as these tools are capable of reading the mind of such algorithms and detect the underlying code itself which is the reason of such pattern. Lastly, testing pricing algorithms prior to its rollout with appropriate market data, must be discussed consistent with ex-ante approach. However, such requirements should focus on undertakings with large user bases or companies with considerable market power to prevent extensive testing ensuring balance between regulation and innovation.

V. Conclusion

To address the issue of algorithmic pricing in India, a proactive approach to policy and research is required. India must pull up its socks and should develop a regulatory framework that effectively addresses these emerging challenges, ensuring that anti-competitive practices cloacked as technological advancements do not go unchecked. Further, technological challenges must be addressed through technology itself, using advanced tools to detect and prevent algorithm-driven market manipulation while safeguarding fair competition.

Note: This article has been reviewed by Mr. Anshuman Sakle (Partner (Competition/Antitrust Law) at Khaitan & Co.) at the Tier II Stage.