Making India’s numbers count again

E-PaperSubscribeSubscribeEnjoy unlimited accessSubscribe Now! Get features like How do you know if your data are crap? You go and look at the toilets. In 2018 Rajasthan, a state in north-western India beloved by tourists, declared that it was “open defecation free”. Separately, government auditors visited dozens of villages in the state to inspect homes built under a government housing-subsidy scheme. “Interestingly”, the report notes, 49% of houses had no toilets. Despite the disclosure, data dashboards continued to show potty penetration at 100%. Interesting indeed. Another big project is to prepare the country’s statistical infrastructure for AI (Unsplash)It is not uncommon, anywhere in the world, for one wing of government to have no idea what another is doing. But that vignette is not petty snark from a columnist. It is contained in a recent report from Niti Aayog, the government’s own think-tank, that laments the quality of statistics in India. The risk of inaction, it said, is that “policy rides on numbers no one fully trusts.” That was just the starkest of six risks it listed. For decades after independence India boasted a statistical infrastructure far superior to what its level of development would suggest. But the system suffered from both neglect and political interference. Things reached a nadir in the 2010s under governments led first by a distracted and scandal-plagued Congress party and then by an inexperienced and ideological Bharatiya Janata Party. Controversies in 2018 and 2019—over GDP calculations, unemployment rates, payroll data and consumption—sapped confidence, at home and abroad, in India’s numbers. After one incident, two officials at the National Statistical Commission, a regulator, resigned in protest. Any competent political outfit should know how to spin data to make itself look good, or at least make itself look less bad. India’s ham-fisted approach of withholding surveys or using indefensible methodologies made it look worse than the actual numbers would have. Eventually the government realised that dodgy data were harming it more than helping it. In 2024 it installed Saurabh Garg, a veteran of the civil service, as the top official at the statistics ministry to turn things around. Among the terminally nerdy crowd that pays attention to these things—policy wonks, researchers, journalists—Mr Garg is spoken of in glowing terms. He has all the right credentials: a degree from one of India’s elite universities, an MBA from its most prestigious business school, and a doctorate from Johns Hopkins University in America. He is well-read and avoids ideological debates. He says things like “if credible data is not available there will be non-credible data to fill the vacuum.” Print that slogan on a T-shirt and, you can be sure, some in India would buy it. Since Mr Garg took over he has created a calendar of data releases, cleared the backlog of pending surveys, sped up the pace of data-gathering and publication, and organised dozens of workshops and consultations. “If you had asked me in 2023 if all these things will happen in the next three years, I would have said it is impossible,” says Pramit Bhattacharya of Data for India, a non-partisan website. Having cleaned up the mess he inherited, Mr Garg is now focussed on producing output numbers for India’s 800-odd districts, a level down from states, to make economic data more granular. Another big project is to prepare the country’s statistical infrastructure for AI . That will involve the fiddly work of harmonising definitions, setting metadata standards, and pushing other departments to release datasets that AI can process at scale. Too many still put out their data as PDFs or, worse, as images. Yet the praise being heaped upon him is a sign of a deeper problem with how India is run. Despite nearly eight decades of independence, a population of 1.45bn people, and an economy that is growing ever more sophisticated, administrative capacity is a fraction of what the country needs. The “steel frame” of governance, as the elite civil service is sometimes known, is made up of just 5,577 officers. It has grown by less than 1,100 in the past 15 years, a period in which the population surged by 250m and the economy doubled in size. Britain, whose population is a 20th of India’s, has over 7,500 senior civil servants. With such a small pool, depth of talent is a problem. And Indian bureaucracy can be an ad-hoc affair that relies more on individual capability than established process. Mr Garg is already past the absurdly low official retirement age of 60, having received extensions for this role. He will eventually move on. If he has done his job well, India’s statistical system may even rise above the individuals who run it.