To obviate the need for regulation , India must map use cases for different types of datasets, before creating new rules for the road
By Megha Patnaik & Vivan Sharan
The proposed $5.7 billion Facebook investment in Jio has reignited questions on the digital economy and the centrality of data. Data is often called the “new oil” because digital economy businesses, like social media platforms, use it as fuel to generate value. More access to data is generally better for such a business. For instance, larger datasets lead to better predictions of user behaviour. Accurate platforms attract more users, thereby generating larger datasets. This virtuous cycle at a business level concentrates market power, and is a reason why India might mandate sharing of privately collected data. The expectation is that this will diffuse the dominance of large platforms.
The first mention of regulatory mandates for data sharing appeared in the draft e-commerce policy of 2019. The draft compared datasets to public resources, like a spectrum. While the draft was scrapped, the issue continues to resurface in data governance debates. The ministry of electronics and information technology recently established a committee to look into the governance of data.
This committee, headed by Kris Gopalakrishnan, co-founder of Infosys, is looking at the “economic dimension of data” and the “platformisation of the digital economy”. It is expected to submit its findings to the government this year.
At first blush, data-sharing regulation may seem like the correct approach to reduce a platform’s advantage. However, digital economy platforms are unconventional. They exhibit a fundamental paradox—the most disaggregated data is often the most useful to them. That is, data that identifies users through basic and personal characteristics provides business insights that are hard to match with pre-arranged data sets. Such data is most likely to be protected by privacy rules or anonymisation protocols, such as those prescribed under the proposed Personal Data Protection Bill.
Conversely, the limits to the usefulness of pre-arranged or derivative datasets can be understood through the lens of organisational economics. According to Stanford University professors Erik Brynjolfsson and Paul Milgrom, successful firms create clusters of complementary business practices. They optimise their operations so that various business processes merge seamlessly. Therefore, firms would find it hard to suddenly incorporate large datasets into their processes, especially datasets that are optimised for another firm. Very few will be able to hire or train staff, rethink existing processes, and reorder operational priorities.
In fact, most traditional Indian firms don’t even exploit the data they have ready access to. Kirana stores seldom use database solutions, even if they are cheap. Such stores are privy to valuable demand and supply data that could be organised using software, and leveraged for improved sales or inventory management. The rates for digital payments adoption remain similarly low, even among mid-sized retailers. This is despite the states’ long-drawn emphasis on awareness programmes around associated benefits of digitisation, such as easier access to loans based on transaction histories. A 2019 survey of 1,000 retailers in Jaipur found that around 60% had not adopted digital payments, despite the financial and infrastructural capacity to do so.
Data governance must also address the question of data ownership. Not only do digital platforms invest in organising data but also they are the only channel of bringing certain data into existence. There would be no data about user behaviour on social media if platforms with data-centric business models themselves did not exist.
Mandating data-sharing could also violate privacy rights since computer databases enjoy protection under Indian copyright law. It helps that data is a ‘non-rival good’ though. This means that one party’s use of data does not automatically preclude its use by another. This is different from most economic goods. For example, one firm’s office typically cannot be used by another. Therefore, economist Hal Varian argues, that “instead of focusing on data ownership—a concept appropriate for private goods—we really should think about data access”.
Platforms can be proactive and volunteer access to some of their anonymised datasets. This would signal what sort of data they are comfortable sharing, and remove uncertainty about what they consider commercially sensitive. For instance, taxi aggregators could share traffic data, telemedicine platforms could share types of health queries, e-retailers could share geographical consumption data, and so on. Such openness may obviate the need for regulation. Moreover, it may encourage traditional rivals to build capacity for using data to increase their sales. The key is for India to map use cases for different types of datasets, before creating new rules for the road.
Patnaik is fellow and Sharan an advisor, Esya Centre, a think-tank in New Delhi. Views are personal