Manufacturing work beyond manufacturing industries: Rethinking structural change in Germany

Manufacturing employment in advanced economies is widely viewed as being in long-term decline. Using German administrative data from 1975 to 2019, this column shows that much of the decline measured at the industry level reflects a shift of manufacturing tasks into service sector firms rather than a disappearance of manufacturing work. When workers move across sectors while keeping their manufacturing occupation, their long-run earnings losses are modest. By contrast, workers who must switch into service occupations experience substantial and persistent penalties. These findings suggest that occupation-specific skills, rather than sectoral affiliation, shape workers’ adjustment to structural change.

Manufacturing employment in advanced economies is widely seen as being in long-term decline. In Germany, conventional statistics show manufacturing’s share of total employment falling from 47% in 1975 to 23% in 2019 – a dramatic collapse that has fuelled anxiety about job losses and the future of industrial economies.

But the common way we measure structural change – based on the number of jobs in manufacturing industries – misses an essential part of the story. In a new study using German administrative data from 1975 – 2019, we show that many ‘lost’ manufacturing jobs have not fully disappeared from the economy (Boddin and Kroeger, 2025). Instead, they are offset in service sector firms, often with remarkably small consequences for affected workers’ long-run earnings and employment trajectories. The conventional narrative of manufacturing decline appears thus overstated and at times misleading when based solely on industry classifications.

Our findings suggest a more nuanced pattern of structural change: workers’ occupations matter far more than their employers’ industry for understanding structural change and its consequences for workers.

Why traditional measures overstate manufacturing decline

The standard measure of structural change classifies workers according to the industry of their establishment. This approach, used widely in both research and policy discussions, implicitly assumes that all workers inside a manufacturing firm are ‘manufacturing workers’ and all employees in service firms are ‘service workers’, regardless of the nature of their tasks.

Germany’s linked employer–employee data allow us to classify workers by occupation instead: whether their tasks involve producing or transforming physical goods (manufacturing occupations) or providing intangible services (service occupations). This occupation-based approach reveals patterns invisible in conventional industry-based statistics, while we are still able to control for industry of employment.

The distinction reveals a large and growing gap. By 2018, 1.8 million more workers held manufacturing occupations than were employed in manufacturing industries – an average of 851,000 over the entire period. Put differently, 42% of all manufacturing workers in 2019 worked for service-sector firms, up from 25% in 1975. As a share of total employment, manufacturing jobs fell by 23.8 percentage points when measured by industry, but only 17.8 percentage points when measured by occupations between 1975 and 2019.

Two-thirds of ‘lost’ manufacturing jobs were replaced

Perhaps the most striking pattern is shown in Figure 1. Between 1975 and 2019, manufacturing industries lost about 1.5 million manufacturing jobs, while service industries gained about 1.1 million manufacturing jobs. Thus, 67.5% of manufacturing jobs ‘lost’ in manufacturing industries reappeared in the service sector.

The net decline in manufacturing work, defined by occupations and hence, tasks performed, was only around 500,000 jobs, far smaller than the perceived 1.5 million.

Figure 1 Manufacturing jobs by sector, 1975-2019

Workers are not just reclassified ex post; they actively transition across sectors. By the 2000s, half of all manufacturing workers leaving manufacturing industries found new jobs in service industries while keeping manufacturing occupations, up from 36% in the late 1970s.

What happens to displaced workers? Their outcomes differ sharply by occupation but much less by sector

These patterns raise a critical question: When manufacturing workers transition to service firms, how costly is the move? Conditional wage gaps show service sector manufacturing workers earn 14% less on average. But this conflates voluntary and involuntary moves, and workers selecting into services may differ systematically.

To isolate causal effects, we focus on workers displaced through mass layoffs, which are plausibly exogenous to individual workers’ employment (Jacobson et al. 1993, Schmieder et al. 2023). To gain a clearer understanding of how occupation and sector of employment shape outcomes for manufacturing workers impacted by structural change, we track manufacturing workers hit by mass layoffs and compare their outcomes based on where they land. Importantly, we construct matched sets consisting of one non-displaced control worker and four displaced workers who each move into one of the four possible occupation–sector combinations: same occupation and sector, same occupation but different sector, different occupation but same sector, or both changed. This novel matching approach allows for direct comparisons of average treatment effects across destination job types.

The emerging picture is clear: switching occupations is costly. Relative to matched controls, displaced workers experience long-run average wage declines that vary substantially depending on where they end up.

Figure 2 shows the dynamic effects on annual income over time. The vertical line at year 0 marks the layoff event. Before displacement, all groups follow parallel trends, after displacement, trajectories diverge sharply by destination type.

Workers who find a new job in the manufacturing sector while keeping a manufacturing occupation experience average annual income losses of about €876 (in 2015 euros), or about 2.2% of their wages, while those taking service occupations in manufacturing lose around €590 or a 4.6% decline in their wage. Workers who retain a manufacturing occupation but transition into the service sector face larger wage losses of roughly €2,247 or 5.9% per year. The most substantial losses of approximately €9,488 (or 29.9%) annually are incurred by workers who switch both occupation and sector and end up in service occupations within the service sector.

Figure 2 Dynamic treatment effects on annual income by post-displacement destination

These effects are substantially smaller than canonical displacement cost estimates from the US (15–25% long-run earnings losses), partly because German displaced workers often preserve their task-specific skills. Workers who retain manufacturing occupations – even in service firms – face modest income losses. Those forced into service occupations suffer dramatically larger penalties.

Why are penalties small? Establishment quality matters

When we account for firm quality using establishment wage premiums (Abowd et al. 1999), the sector penalty nearly vanishes. The wage penalty for workers who move to the service sector while retaining a manufacturing occupation shrinks from -5.9% to -0.5% and becomes statistically insignificant. In contrast, switching occupations remains costly even after accounting for establishment quality, with workers who end up in service occupations within the service sector still facing wage losses of roughly -17%.

These results imply that manufacturing occupation workers who transition into service firms generally enter lower-paying establishments, rather than lower-paying occupations. The sector wage gap reflects firm characteristics rather than the nature of work itself.

This interpretation aligns closely with evidence from Helm et al. (2023), who also find that much of the earnings loss associated with displacement reflects movement into lower-paying establishments rather than pure sectoral effects.

The main driver of long-run wage losses is the loss of occupation-specific human capital, not the act of moving from manufacturing to services.

Workers who must change occupations, particularly those pushed into lower-skill service jobs, experience substantial and persistent losses. This mirrors well-known patterns of occupational specificity in human capital (Kambourov and Manovskii 2009, Gathmann and Schönberg 2010). The key message, however, is that sector transitions per se are not the main risk. What matters is whether workers can preserve their occupational skill set.

Implications for policy

Labor market policy should prioritise the preservation and development of occupational skills rather than sectors. Since manufacturing work increasingly takes place outside of traditional manufacturing firms, and since workers’ task-specific skills are far more important than sectoral transitions, policies that aim to protect specific industries are likely to be less effective than those that help workers maintain or update their occupational competencies.

A growing set of advanced services – including logistics, IT, quality control, design, analytics, and other production-related activities – can absorb manufacturing talent as firms outsource these functions (Houseman 2007). The service sector is not homogeneous; it contains high-quality, capital-intensive segments that rely heavily on manufacturing-related skills.

Retraining programmes should be aligned with these realities. Workers who retain manufacturing occupations, even when they move into service sector firms, experience far smaller losses than those who switch into service occupations. Employment agencies should expand job matching beyond industry boundaries, recognising that service firms increasingly need manufacturing expertise. Effective retraining should therefore strengthen manufacturing task capabilities and support moves into advanced services that use similar skills, rather than steering displaced workers toward unrelated service jobs where earnings losses are substantial.

Conclusion

Germany’s experience over nearly half a century shows that the decline of manufacturing employment, when measured by industry classifications, overstates the extent of change. Structural change is less disruptive than commonly believed. Much of what appears as a loss of manufacturing jobs reflects a shift of manufacturing occupations into service sector firms, rather than a disappearance of manufacturing work itself (see also Smeets et al. 2016 and Bernard et al. 2017).

In transition, the typical fear is that displaced manufacturing workers must either accept low-pay service jobs or exit the labour market altogether. Our results offer a more optimistic narrative: manufacturing work increasingly takes place in service firms and workers frequently follow this shift without switching occupations. These transitions carry relatively small long-run penalties.

This insight changes how we understand structural change. It shifts attention away from industries and toward tasks and occupations (Duernecker and Herrendorf 2022). It reveals that workers who retain their occupational set of tasks can often transition smoothly across sectors at relatively modest costs. 

Source : VOXeu