BY SAKSHAM KHOSLA
Saksham Khosla is a second-year International Development concentrator at SAIS and a senior editor at SAIS Perspectives. He previously conducted research on social protection, governance, and financial inclusion in India at the Carnegie Endowment for International Peace and Carnegie India.
Dr. Alan Gelb, Senior Fellow at the Center of Global Development, discussed the landscape of digital public service delivery in developing countries at the second Development Roundtable of the 2019-20 academic year. His discussion drew on the experience of the Krishna district of the Indian state of Andhra Pradesh, a leading implementer of digital reforms.
In a recent survey, Dr. Gelb and his colleagues sought to understand the impact of such reforms from the perspectives of users of digitized programs. They documented largely positive responses driven by wide access, clear accountability, user choice, and real-time feedback. However, challenges like inefficient beneficiary selection processes and still-imperfect biometric authentication highlighted remaining issues with the Krishna model.
Discussing the Krishna model’s generalizability in other Indian states and beyond, Dr. Gelb emphasized the necessity of having a technological infrastructure like the Aadhaar platform in place, high preexisting levels of state capacity, and a political economy committed to the sustained implementation of digital reforms.
Following his presentation, SAIS Perspectives spoke to Dr. Gelb about implementation strategies for digital governance, and the data security and privacy safeguards required by the increased use of technology in government. This interview has been edited and condensed for clarity.
Perspectives: The story told by your presentation was one of process improvements, while another narrative around Aadhaar emphasizes its role in empowering beneficiaries at the last-mile. What specific problem is digital governance is trying to solve in India?
Dr. Gelb: I think that there are several objectives. One is the improvement of a process or a delivery interaction; whether to improve customer experience or to save resources. Another is empowerment, in other words, enabling people to do things they were not able to do otherwise. Empowerment, as well as inclusion, is partly a matter of transaction costs. Both objectives come together, but when we see how different systems are applied, there may be an emphasis on one or the other objective. In India you can both see efforts to improve benefit systems and other programs as well as to reduce transaction costs through the IndiaStack, the Unified Payments Interface, and the use of Aadhaar for e-KYC.
Perspectives: How can variation in state capacity across regions or levels of government be explicitly recognized when undertaking digital governance reforms?
Dr. Gelb: Not all elements of the Krishna model are applicable, at least quickly, in all jurisdictions, but some parts probably are. For example, quite a number of countries have managed to clean their beneficiary rolls through the use of ID technology, but authentication at point of service is harder. Some countries may find it easier to use a ration smart card than fingerprints, which are difficult to read for come people. One lesson from Krishna is that that digitized programs will generate huge volumes of administrative data. A capable administration can build on this by analyzing it and using this information to fix problems as they arise. But this is not an easy or simple thing to do.
When I see delivery systems that are in principle digital, one question I ask is if they have a feedback mechanism to know how well they’re doing. The answer often is no, but many systems could. These systems need to have very quick feedback mechanisms in place, whether it be from people or from administrative data.
Perspectives: Along with getting the policy design and the institutional environment right, does digital governance also demand investments in privacy and data security?
Dr. Gelb: Many governments, not just those in developing countries, are grappling with the implications of vastly increased amounts of personally-identifiable digital data. As far as I know, Estonia is the only government in the world that has implemented a comprehensive approach towards structuring of government data. In the Estonian model, each service or facility is responsible for its own data, so there’s no one massive database. Data is securely stored and encrypted, there is a log of every time it is accessed or altered, and so on. There also needs to be an organized way for secure data exchange between silos; this is the X-Road platform in Estonia. Citizens can see who has accessed their data.
If you simply lock up data in silos, you lose the efficiency gains that are the whole point of having a digital system. In the Estonian system, the principle is that you only give one piece of information once. If I were to ask for a pension, for example, I wouldn’t have to give my age – it would be pulled automatically from the system. The triggering of data sharing in the Estonian system is done automatically according to the needs of different services or programs as you access them. I call that “implied consent.” In other words, if I go and ask for a pension, by that action I am saying to the government that I know you will want to know my age and I give you permission to pull it from where it is. I don’t need to separately tell you, or authorize its release. This is also what India is doing with e-KYC, which pulls data to reduce the costs of onboarding clients.
Data-sharing compounds the problems of privacy, for which you need systems that have both a legal and a technical component. Indeed, mechanisms for both implied and explicit consent to share information need to be available in systems for digital governance. Estonia was able to start more or less from scratch as a newly independent country, but Andhra Pradesh, and most administrations, have to deal with the legacy systems. So you have to set up an infrastructure and migrate your systems into it, as you are able to do. I personally believe that there is one conceptual model underlying the management of public data which is derivative of Estonia. Most administrations need to make a strong effort to build something of that sort or they will run into some really serious problems.
Perspectives: Citizens around the world are being exposed to varying standards for data protection. From their point of view, which models do you see as being particularly effective?
Dr. Gelb: I don’t think that an explicit consent model for data-sharing is practical for all purposes. There are two reasons -- one is the difficulty of understanding what I am consenting to, and the other is the time and the effort that it would take for me to have to give consent to every piece of data-sharing.
I think you need a clear understanding of what implicit consent is, and then mechanisms for people to increase choice. I could imagine a choice-based system, where I’m told that if you want this service we’re going to pull this information for you, and you can tick the box and tell us that’s okay, and if not, you can spend 6 months filling in things – it’s up to you.
In terms of an implicit consent architecture, there should also be a governance structure which regulates sharing. It should be perfectly transparent, so that I know if I’m applying for a pension it’s going to ask my age, let’s say, and maybe it’s going to ask how many years I’ve lived in the country if the pension is dependent on something like that. But it should be clear – and based on that, I can choose whether to permit sharing or go through the laborious process of managing individual pieces of data. But in practice I think most people would just tick the box.
SAIS Perspectives is grateful to Dr. Gelb for taking the time to speak with us, and to the International Development program for organizing the Development Roundtable. To read about other Development Roundtable events, click here.
PHOTO CREDIT: Biswarup Ganguly, from Wikimedia Commons, licensed under CC BY-SA 3.0.
