Latest / The Productivity Podcast with Fexingo: Output, Efficiency, and Long-Term Economic Growth / How Sweden's Industrial Robot Fleet Reshaped Manufacturing
Transcript
- Lucas: If you've ever wondered whether industrial robots actually make a country more productive, Sweden is the natural place to look. Luna: Sweden? Not Japan or Germany? Lucas: Right, Japan and Germany have the highest absolute robot populations. But per capita, Sweden is a different story. According to the International Federation of Robotics, Sweden had about 290 industrial robots per ten thousand manufacturing workers in 2024. That is the third highest density in the world, after South Korea and Singapore, and well ahead of Germany and Japan. Luna: So Sweden's not just using robots — they're basically the densest robot workforce among large industrial economies. Lucas: Exactly, and the productivity figures back that up. Swedish manufacturing output per hour has grown at an average of about 2.5 percent annually since 2010, which significantly outperforms the European Union average of about 1.4 percent. The question is how they pulled that off without the kind of social friction you'd expect. Luna: And that's where the ABB story comes in, right? Lucas: It does. ABB is headquartered in both Sweden and Switzerland, and their robotics division has been central to this transformation. There's a specific factory in Västervik, about three hours south of Stockholm, that makes electric motors. In 2019, they introduced collaborative robots — cobots — to work alongside humans on assembly lines. These aren't the caged-off industrial arms you see in car plants. Luna: Right, cobots are designed to be safe around people. They have sensors that stop them if they touch a human. Lucas: Precisely. At Västervik, the cobots handle repetitive tasks like inserting coils and tightening screws, while human workers focus on quality checks and customisation. The result, according to ABB's own data, was a 30 percent increase in throughput on the lines that adopted cobots within the first six months. And here's the key: the factory didn't lay off anyone. Luna: That seems hard to believe. How did they avoid job losses? Lucas: Sweden has what economists call a 'flexicurity' model — flexible hiring and firing combined with strong social safety nets and active retraining programs. In Västervik specifically, the union was involved from the start. They agreed that any worker displaced by a cobot would be retrained for a higher-skilled role within the same factory. In practice, most of those workers ended up programming and maintaining the robots. Luna: So the robots didn't replace people, they upgraded their jobs. Lucas: That's the narrative, and the data mostly supports it. Sweden's manufacturing employment has actually stayed roughly flat since 2010, while output has risen by about 25 percent in real terms. That's a pure productivity gain — more value per worker, not fewer workers. Luna: But Sweden's overall unemployment rate has been around 7 to 8 percent in recent years, which is higher than the US. How does that square? Lucas: It's a fair point. The manufacturing gains are real, but they're concentrated in a shrinking share of the workforce. Manufacturing's share of total employment in Sweden has fallen from about 18 percent in 2000 to around 12 percent today. The robots helped preserve those jobs that remain, but they didn't reverse the long-term shift to services. The productivity gains show up in higher wages for manufacturing workers, though. Luna: Higher wages — is that measurable? Lucas: Yes. Between 2010 and 2024, real average hourly earnings in Swedish manufacturing grew by about 15 percent, after adjusting for inflation. That's roughly two percent per year. In the US over the same period, real manufacturing wage growth was closer to one percent per year. So the productivity gains in Sweden really did translate into better pay. Luna: What role did the collective bargaining system play in that? Lucas: A big one. Sweden has industry-wide collective bargaining agreements that cover about 90 percent of manufacturing workers. Those agreements typically include clauses about productivity-sharing — if a factory's output per worker goes up, a portion of that gain goes into higher wages. It's not automatic, but it's a strong norm. In Västervik, the union negotiated that any productivity gains from cobots would be split fifty-fifty between reinvestment and wage increases. Luna: So the workers had a direct incentive to accept automation. Lucas: Exactly, and that's the institutional piece that's often missing in the American debate about robots taking jobs. In Sweden, the labour market institutions were designed to make automation a win-win. It's not perfect — there are still displaced workers in sectors that can't be automated — but the manufacturing sector has managed a remarkably smooth transition. Luna: Is there a downside? Are there any Swedish industries where robot adoption backfired? Lucas: The most cited cautionary tale is the paper and pulp industry. Sweden is a major paper exporter, and they automated heavily in the 2000s. Productivity soared, but employment in that sector fell by about 30 percent between 2000 and 2020. The workers who lost those jobs were often older and harder to retrain. So the flexicurity system worked for some, but not for all. Luna: So the lesson is not that robots are always benign. It's that the institutional context matters a lot. Lucas: Right. The technology itself is neutral. Sweden shows that if you have strong unions, active labour market policies, and a willingness to share productivity gains, robot adoption can raise both output and wages. If you lack those institutions, the same robots can concentrate wealth and displace workers without compensation. Luna: That feels like a really practical insight for policymakers elsewhere. Lucas: It is. And it's especially relevant right now, as many countries are trying to reshore manufacturing. You can't just drop robots into a factory and expect productivity gains to translate into shared prosperity. The Swedish model suggests you need the social infrastructure to match. Luna: If this episode made something click about how automation and productivity really work, we hope you found it useful. And if you did, the reason this show stays ad-free is that listeners who value it sometimes choose to support it. Lucas: Yeah, that's exactly it. If you'd like to be part of keeping that going, you can find us at buy me a coffee dot com slash fexingo. No pressure, genuinely — just if the show has been helpful to you. Luna: And if today's episode got you thinking about automation policy, that's the best outcome. We'll be back next time with another deep dive. Lucas: Until then, it's worth asking: what kind of institutions does your own country have to share the gains from technology adoption? Because that might determine whether the next wave of automation lifts all boats or just a few.