India’s digital success is generally known. It has enabled the country to handle the pandemic and its attendant consequences better. By any metric, India’s performance is far ahead of many so-called advanced countries. This success is based on digital public infrastructure (DPI).
Classically, “infrastructure” is anything that supports the economy, and traditional examples of infrastructure will be roads, railways, bridges and electricity, perhaps even water supply and sewage treatment. I think the first time the Economic Survey had a separate chapter on infrastructure was in 1980–81. In the 1980s, there was of course a lot of interest in the USA on what we will now call physical infrastructure. That 1980–81 chapter covered power, coal, railways and shipping/ports.
Today, no one will mention coal, and other items of infrastructure will be included. With fast advances in technology, “digital” is bound to be included. Infrastructure can be developed by the public sector or the private sector. Since there is market failure in some areas, some of this digital infrastructure will have to be publicly developed. Thus, the word “public” in DPI.
In October 2022, ORF (Observer Research Foundation) published a brief paper on creating “good” DPI. This was authored by Kriti Mittal, Varad Pande and Aishwarya Viswanathan.
To quote from this, “DPI comprises foundational population-scale technology systems on which the digital economy operates, such as identity systems, payment systems, data exchanges, and social registries…Today, DPI is increasingly being built using open-source and modular technologies that enable ‘interoperability’, which facilitates the exchange of information between different arms of the public and private sector, thereby, vastly improving the speed and scale of service delivery. This represents a paradigm shift from older end-to-end siloed systems, wherein governments provided end-to-end services through monolithic tech systems, to building minimal digital infrastructure that allows multiple actors to build solutions on top. DPI designed in this way can mean significant time and cost savings.”
We get the general idea and hence, there is the notion of India Stack, a set of open APIs (Application Programming Interfaces) and digital public goods (with public goods not being used in the classic economist’s sense). What is this India Stack? Most people will mention Aadhaar (identity) and UPI (United Payments Interface) because we use them the most and are most familiar with them. But there is more to India Stack—CoWin, DigiLocker, Aarogya Setu, Diksha, Umang, e-Sanjeevani, GeM, API Setu, e-Office and e-Hospital, not to forget GSTN. Aadhaar and UPI took some time to take off. So will these, with a lag. We need to recognise that all of these are part of the same overall jigsaw of DPI.
If we have done 2.21 billion vaccinations, that has been facilitated by CoWin. DigiLocker has 148 million registered users and 5.6 billion documents have been issued. Personally, I may not be that interested in digital versions of marksheets I can access anywhere, but I remember the days when I had to run around to get photocopies of marksheets attested. With the pandemic having eased and contact tracing no longer that important, Aarogya Setu has lost some of its relevance. But in its day, it was extremely important in fighting the Covid battle. Unless you are a student (or a teacher), you may not be that interested in Diksha. Nevertheless, the fact remains that 5.2 billion learners have used the Diksha infrastructure. Most of these learning sessions have been in Rajasthan, UP, MP, Odisha and Karnataka, driving home perhaps an obvious point. When there is an asymmetry in distribution of access to physical infrastructure, there is greater demand for the digital. In that sense, DPI empowers the relatively disadvantaged. There have been 50 million registrations and 3.4 billion transactions under Umang.
Relatively speaking, e-Sanjeevani probably lags. Even then, 93 million patients have been served. GeM dashboards, for government procurement, are no less impressive. API Setu is very specific, as is e-Office, for G2G transactions. Finally, like e-Sanjeevani, e-Hospital also seems to lag a bit.
Naturally, Aadhaar had a head start. It was the first piece in the building block. So far, 1.36 billion Aadhaar numbers have been generated. Unique Identification Authority of India (UIDAI) uses a concept of “saturation”, to gauge what percentage of the population has been issued Aadhaar numbers. Naturally, this is highest for adults (age more than 18), lower for those between five and 18, and lowest for those who are under five years of age.
Since Census numbers are for 2011, any population figure after that is a projection or estimate. With that qualification, for adults, saturation is almost 95%. I can understand saturation being low in the northeast, Ladakh and J&K (by the way, low means in excess of 80% or even 85%). I don’t understand why it should be 87% in Bihar. Therefore, for UIDAI, new Aadhaar generation tapers off and stabilises at a steady number of around 30 million a year. (The peak was in 2013–14.) The focus shifts to updating details and use of Aadhaar for authentication.
We may not realise it, but a staggering 70 million authentications are done every day—primarily for banking, telecom, PDS and MGNREGA. The traditional authentication mode has been biometry (fingerprints), other than OTPs and demographic data. As technology evolves, perhaps we will switch more to biometry through the face. If CISF is going to use facial recognition cameras at airports, why not elsewhere? It is difficult to quantify benefits of something like DPI, or Aadhaar in particular. But it should be obvious that DBT has curbed leakage, and financial inclusion and social security have become easier. Without this umbrella, portability of welfare benefits would have been difficult during the pandemic. Perhaps Aadhaar will eventually be linked with birth and death records, illegal residents and crime.
It shouldn’t be surprising that India’s success with DPI figures in the G-20 agenda. Nor should it be surprising that several countries wish to learn from India’s experiment with DPI and Aadhaar, seeking to replicate it. It has been a remarkable success, in a relatively short period of time. I suspect most people have forgotten the name of Ranjana Sonawane. She was the first person to get an Aadhaar number (at that time, a card). That was in September 2010, not very long ago. The USA started its social security number system in 1935. Technology helped, but we used it through DPI.
Bibek Debroy
Chairman, Economic Advisory Council to the PM