Designing competition policies for the age of AI
Source: By Avinash M. Tripathi: Mint
Recently, the Competition Commission of India (CCI) imposed a hefty penalty on Google for abusing its dominant position in the online search market. The company was accused of promoting its own verticals at the expense of its rivals. While the specifics of the case have received much attention, it has probably not been appreciated that this case is a kind of watershed moment for the competition policy in India. In future, competition policy will have to face the challenges of Artificial Intelligence (AI) and big data. Since Google is a leading AI company, this occasion may be used to think about these broader sectoral challenges.
Essentially, the regulatory challenges posed by AI fall in three broad categories: market foreclosure and related exclusionary practices; new forms of collusion; and new strategies to implement price discrimination. Additionally, AI will accentuate concerns about technological sovereignty and wealth inequality. While these concerns may seem underwhelming in comparison to the dystopian future shown in Hollywood movies, they are more realistic and have an immediate relevance for public policy.
In an influential article, investor and entrepreneur Marc Andreessen argued that software is eating the world. He elaborated: “More and more major businesses and industries are being run on software and delivered as online services—from movies to agriculture to national defence.” One corollary of Andreessen’s argument is that software and algorithmic solutions are complements to a whole array of goods and services. Complementary products refer to the goods and services which are more useful and valuable when consumed jointly; think of bread and butter and left-and-right shoes.
If a firm is dominant in one market (say search engine), it may try to extend its dominance in other complementary products (say email) by foreclosing (denying) fair access to its rival’s product (for example, by giving it low search ranking). Given their economy-wide cross-sectoral linkages, AI firms are particularly susceptible to these tactics, often called market foreclosure. Besides foreclosure, algorithmic markets and dynamic pricing bots are especially vulnerable to price-rigging and cartelization. Adam Smith had warned that when the businessmen of the same trade meet even for “merriment and diversion”, their conversations veer towards some “contrivance to raise prices”. These “contrivances”, more succinctly known as “collusion’, are a potential source of antitrust action. Smith would surely have known that dining, wining and other forms of human sociability are not required for collusion. In fact, with advances in technology, even human beings are unnecessary. In 2016, the UK competition watchdog, Competition and Markets Authority (CMA), fined online sellers over £160,000 for using their pricing software to implement an agreement not to undercut each other’s prices (among other things, they were selling Justin Bieber’s posters).
In a recent paper, legal scholars Ariel Ezrachi and Maurice Stucke analyse possible strategies to implement collusion in an algorithmic market such as Amazon. They find that such strategies differ in terms of the needed human intervention and legal deniability. In the simplest cases, like the one successfully prosecuted by the CMA, firms communicate their intention to behave collusively, leaving a clear-cut evidence trail that establishes their culpability. The trickiest case occurs where the collusion is “tacit”, that is, there is no agreement to fix the price, not even a communication. Pricing strategies are programmed to maximize profit over a long horizon and algorithms “learn” to behave collusively on their own. Since there is no human intervention and the objective being pursued is perfectly legitimate, two ingredients of the successful anti-trust prosecution—“intent” and “agreement”—are extremely hard to establish.
Finally, in the age of AI, price discrimination will most likely become the biggest bone of contention. Price discrimination is a practice where a firm, instead of charging a single, uniform price for a single good or service, charges different prices from different consumers, depending on their willingness to pay. Besides overcharging by creating artificial scarcity, price discrimination is another way a dominant firm can inflate its profit at the consumers’ expense. Price discrimination is known to have pretty strong distributional consequences.
Differential pricing is pervasive even in brick-and-mortar businesses, often in subtle and disguised forms. One of my friends was puzzled to learn that a hospital was charging differently for the same surgical procedure based on which ward the patient was occupying. But brick-and-mortar businesses and online outlets differ substantially in terms of their ability to price discriminate. Typically, an online business has both data mining tools and enormous informational database on the consumers. Consequently, it can segment its consumer base much more finely and predict consumer decisions with much more precision.
Economists and regulators have traditionally neglected price discrimination on the ground that it can only have distributional consequences which ultimately cancel out. A consumer’s loss is producer’s gain and vice-versa. This argument is however untenable in the case where the service-provider and consumers are not located in the same national jurisdiction. Should the antitrust authorities remain indifferent to the consumer’s loss provided it is compensated by the gain of the service provider located offshore? If the answer is no, then they should scrutinize tactics such as differential pricing much more closely.
Artificial Intelligence will probably become one of the biggest wealth-creating sectors in this century. Whether this wealth will raise general well-being or merely generate popular backlash is still an open question. Intelligent regulation and robust institutions will be the key to harness its potential. Otherwise, as economist Avinash Dixit once warned: “With good institutions, a good level of economic well-being can be sustained; without them, even great wealth can be fragile.”
[printfriendly]
