Broadband on the factory floor

In 2010, fast broadband in Colombia reached Bogota. Then the government began to lay fibre to hundreds more municipalities. Factories in those towns gained the tools city firms already had: a website, online orders, job ads, video calls, bank transfers in seconds. Would it grow firms, and would it take jobs?

Gaurav Chiplunkar (University of Virginia) followed about 12,000 Colombian manufacturing plants from 2008 to 2019 to find out. The firms that got broadband grew, but not in the way the conventional wisdom predicted. They hired more people for the production line and bought more machines, but they hired no more managers or technical staff. The larger firms appear to have grown fastest. 

Chiplunkar presented the research at STEG's sixth Annual Conference in Nairobi in January 2026. He tells Tim Phillips how he separated the effect of broadband from everything else happening in Colombia's economy, why rising wages both helped and held back the firms that adopted it, and what the changes in Colombia can tell us about AI.

The research behind this episode

Chiplunkar, Gaurav. 2026. "General Purpose Technology and Firm Production: Evidence from Internet Adoption." Working paper, presented at the STEG Annual Conference, Nairobi, January 2026. Draft available from the author on request.

To cite this episode

Phillips, Tim, and Gaurav Chiplunkar. 2026. "Broadband on the factory floor." Conversations on Transformation (podcast).

About the guest

Gaurav Chiplunkar is Assistant Professor of Business Administration in the Global Economies and Markets area at the University of Virginia's Darden School of Business, which he joined in 2019. His research spans labour market frictions, the barriers women and young people face in work and entrepreneurship, industrial policy, and how digital technologies reshape firms and jobs. He holds a BA in Economics from Delhi University, an MSc from the London School of Economics, an MPhil from the University of Cambridge and a PhD in Economics from Yale University. He is a research affiliate of BREAD, J-PAL, the International Growth Centre, IZA and Y-RISE.

Research cited in this episode

General purpose technologies are innovations that spread across the whole economy, improve over time and make further innovation possible. Timothy Bresnahan and Manuel Trajtenberg gave the idea its name in "General Purpose Technologies: 'Engines of Growth'?" (Journal of Econometrics, 1995). Electricity is the classic example; the internet and AI are the recent ones. Chiplunkar's point is that such a technology does not remove one barrier but several at once. It opens markets, cheapens inputs and changes how a firm organises itself, and studies that measure one channel at a time can miss how the channels interact.

Mobile phones and Kerala's fish markets are the subject of Robert Jensen's paper "The Digital Provide: Information (Technology), Market Performance, and Welfare in the South Indian Fisheries Sector" (Quarterly Journal of Economics, 2007). When mobile phone coverage spread along the Kerala coast between 1997 and 2001, fishermen could check prices before they chose where to land their catch. Price differences between beach markets collapsed and waste disappeared. Chiplunkar cites it as the clearest picture of what better information does to markets in a low-income economy.

Vive Digital and the National Fibre Optic Project supplied the rollout that Chiplunkar studies. Colombia's ICT ministry, MinTIC, made the Proyecto Nacional de Fibra Óptica the backbone of its Vive Digital plan. The contract, awarded in November 2011, committed the operator to connect 788 municipalities to the fibre network, chosen mainly for coverage and for the technical feasibility of laying cable. Because the fibre reached different towns at different times, otherwise similar firms gained access to broadband in different years.

The Annual Manufacturing Survey (Encuesta Anual Manufacturera, or EAM) is run by DANE, Colombia's national statistics agency. Chiplunkar uses a panel of about 12,000 formal manufacturing establishments. It records production and inputs, with separate counts of managers, professional and technical staff, and production workers. A module on information technology asks firms whether they have a website, sell online, take orders online, hire online or use internet calls. That module shows what firms did with broadband, not only whether they grew. Informal firms are not in the survey; Chiplunkar plans to study them next.

Shift-share instruments deal with a basic problem: the towns that got fibre, and the firms that adopted broadband first, were not chosen at random. Chiplunkar combines two sources of variation, how fast the number of broadband providers grew in a region and how heavily a firm's industry already relied on the internet. A consumer electronics maker and a rubber tyre maker in the same town face the same new supply of broadband, but the electronics firm has far more to gain. Comparing them removes anything that affects the whole town; comparing electronics firms across towns removes anything that affects the whole industry. The approach draws on work by Kirill Borusyak, Peter Hull and Xavier Jaravel, including "Quasi-Experimental Shift-Share Research Designs" (Review of Economic Studies, 2022).

Skill-biased and scale-biased technical change frame the jobs question. In the 1990s and 2000s, economists asked whether computers and the internet would raise demand for highly skilled workers and push out the less skilled. A scale-biased change instead lets firms grow by doing more of what they already do. Colombia's factories fit the second pattern; they added production workers and machines, not managers or technical staff. Chiplunkar cautions that the survey records job categories, not skills, so treating production work as less skilled is an approximation.

General equilibrium effects are the consequences that come back to firms once the whole economy adjusts. When broadband raises wages, workers spend more, which lifts demand for firms' products. But firms must also pay their workers more, which makes them hire fewer. In Chiplunkar's preliminary estimates, higher incomes explain about a quarter of the gain in firms' sales, while higher wages cut the gain in labour use by about a fifth. In the results he presented in Nairobi, broadband raised Colombia's manufacturing output by 7% and real incomes by 1.5%.

More Conversations on Transformation episodes

Digital technologies in developing countries. Robert Townsend asks how digital tools such as blockchain can reproduce, and sometimes improve on, the social structures that support trade in rural Thailand, a different view of what technology lets firms do.

A Microeconomic Perspective on Misallocation. Eric Verhoogen explains why the most productive firms in low-income countries often stay small. Chiplunkar finds that broadband helped the larger firms grow fastest.

Further reading

Mobile Internet, Collateral and Banking, a STEG working paper by Angelo D'Andrea, Kangni Kpodar, Nicola Limodio, Andrea Presbitero and Patrick Hitayezu, finds that 3G coverage in Rwanda moved borrowers from microfinance to commercial banks on better terms, partly because it helped them obtain land titles under a land reform and use them as collateral.

Technology Adoption and Late Industrialization, a STEG working paper by Jaedo Choi and Younghun Shim, uses historical data on South Korean firms' contracts to adopt foreign technology after the war and finds productivity gains for adopters and spillovers to nearby firms.