Papers Worth Reading: Chen (2026) on Digitalization’s Uneven Effects in India
Yutong Chen’s recent paper in the Journal of Development Economics deserves attention from anyone working on information technology and development. Using India’s 2016 demonetization as a natural experiment, Chen shows that digital transition’s effects are sharply uneven across sectors. Service firms in digitally-prepared districts saw productivity growth while manufacturing firms in those same districts suffered almost perfectly offsetting declines. Why? Because there weren’t enough tech workers to go around. That’s a wrinkle worth thinking about before assuming digital infrastructure investments are always a good idea.
On the evening of November 8, 2016, Modi appeared on television and announced that at midnight (four hours later) all 500 and 1000 rupee paper notes would become worthless. These were the largest bills in circulation, together representing 86% of India’s currency. Citizens had until December 31 to deposit their old notes in banks, but they couldn’t simply swap them for new cash. If you possessed large amounts of undeclared cash, you could deposit it, but then the tax authorities would notice. So people with illegal or undeclared stacks faced a choice: (a) lose the money or (b) deposit it and face scrutiny. ATM withdrawals were capped, and the new replacement notes (redesigned 500s and new 2000 rupee bills) trickled out slowly because the Reserve Bank hadn’t printed enough. For weeks, Indians faced a severe cash crunch. The stated goals were fighting corruption, flushing out undeclared “black money,” and pushing the country toward digital payments.
Before this change, cash dominated Indian transactions. Firms handled salaries, supplier payments, and customer transactions primarily in cash. The sudden shortage forced a rapid switch toward digital payments, and this is where Chen’s research design gets clever. Suddenly everyone needed to go digital, giving Chen a clean before-and-after comparison. But some places had good digital infrastructure and some didn’t, providing cross-sectional variation. The sudden shock hit everywhere at once, but some places were ready and some weren’t.
Chen compares firms in districts with good digital infrastructure to those in districts with poor infrastructure, before and after 2016. Statistical controls account for the fact that firms differ in all sorts of ways and that some industries faced their own shocks over this period.
Chen’s punchline challenges the assumption that digital infrastructure helps all firms. After 2016, service firms in districts with strong digital readiness gained 1.3% in income and 8% in productivity. Manufacturing firms in those same districts lost 2.5% in income and 8% in productivity. Here’s the thing: if you just pool all firms together and look at average effects, you find basically nothing—which would lead you to conclude that local digital infrastructure doesn’t matter. The sectoral breakdown tells a completely different story.
What drives this divergence? To figure out why, Chen turns to household survey data on workers and wages. The demonetization increased demand for tech workers as firms scrambled to digitalize. But ICT labor supply is relatively fixed in the short run. Given their consumer-facing nature, service firms faced stronger pressure to digitize. So they outbid manufacturing firms for scarce tech talent. Wages for tech professionals rose 10% relative to other workers in high tech districts, while tech workers became more likely to work in services and less likely to work in manufacturing.
The downstream effects follow naturally. Service firms that attracted tech workers increased investment in software, computers, and IT systems. Manufacturing firms, losing the workers needed to implement and maintain digital systems, cut back on such investments. The productivity gap compounds from there.
For ITD researchers, several implications stand out. First, the findings challenge the assumption that digital infrastructure investments benefit everyone. Much ITD research asks whether connectivity, mobile coverage, or digital payment systems improve outcomes for firms and households. Chen’s evidence suggests we need to ask “for whom?” The same infrastructure that enables service sector growth may hurt manufacturing by intensifying competition for complementary inputs.
Second, the labor market mechanism deserves more attention than it typically gets. Too much research focuses on whether firms or households have access to IT. Chen shows that access isn’t enough; you also need the human capital to exploit it. Districts with excellent digital infrastructure but thin tech labor markets may see limited benefits, or may even lose tech workers to better-paying opportunities elsewhere. Where the tech talent is located may matter as much as where the fiber optic cables are.
Third, the findings highlight tensions between different development objectives. India’s government simultaneously pursued digitalization (through demonetization and the Digital India campaign) and manufacturing promotion (through Make in India). Chen’s work suggests these goals conflicted: the digitalization push accelerated service-led growth at manufacturing’s expense. ITD researchers advising policymakers should watch for such tensions rather than assuming digital transformation lifts all boats.
Chen’s research design offers a template for studying digitalization elsewhere. The key ingredients are a shock that accelerates digital adoption, pre-existing variation in digital readiness, and data allowing sectoral breakdown.
The finding that matters most is that averages can hide dramatic winners and losers. That should shape how we design and interpret future studies. Disaggregate by sector from the start. Treat null average effects with suspicion, since winners and losers may be canceling out. And when advising policymakers that digital infrastructure investments are beneficial or harmless, remember that the aggregate picture can obscure significant harm.
The paper also raises a deeper question for ITD researchers: can governments actually steer digital transitions where they want them to go? The Indian government pursued demonetization partly to promote digitalization, yet the policy produced consequences that undercut its manufacturing ambitions. Complex economies respond in ways that planners cannot fully anticipate.
For now, the practical takeaway is simple: read Chen’s paper. It is carefully executed, and the findings are important. The assumption that digital infrastructure uniformly helps everyone deserves serious scrutiny.
Citation: Chen, Yutong. 2026. “Digitalization as a Double-Edged Sword: Winning Services and Losing Manufacturing in India.” Journal of Development Economics 179: 103618.
Downloadable here: https://www.sciencedirect.com/science/article/pii/S0304387825001695

Thanks, Matt!
Nice! This isn’t even a topic I’d have much interest in on my own, but your summary made it interesting and meaningful. Cheers.