Few diffusers and many bottlenecks: the French labor market according to statistical physics

How long have you been doing your current job? Have you ever thought about trying a new profession? How difficult does change seem to you? The current rapid transformation of the labor market is putting many workers to the test: they struggle to keep up and move into new roles, while at the same time companies are having difficulty finding qualified personnel. A new study has analyzed the French labor market using methods from statistical physics, and found that over 90% of jobs today function as bottlenecks: they are easily accessible, but once entered, they become traps from which it is hard to move elsewhere—even when other opportunities are available.

The study, conducted by Max Knicker, Karl Naumann-Woleske, and Michael Benzaquen of École Polytechnique in Paris, and published in the Journal of Statistical Mechanics: Theory and Experiment (JSTAT), provides a detailed mapping of accessibility and transferability characteristics within the French occupational network. It reveals strong structural rigidity in the overall labor system and offers a basis for understanding what kinds of interventions and policy decisions might help to break this deadlock.

Technology, lifestyle changes, migration, and other issues are rapidly transforming the labor market: professions that were in high demand just ten years ago may now be obsolete, while new and growing jobs struggle to find skilled workers. This affects people’s lives directly, and also impacts economic development. To find effective strategies, it is crucial to understand the detailed mechanisms of access and transition between occupations. Doing so requires the analysis of large-scale data, which is where statistical physics, developed precisely to manage large volumes of dynamic information, becomes particularly useful. Knicker and his team applied network analysis tools to data from the last ten years of the French labor market, revealing structural rigidities and vulnerabilities.

One of the study’s strengths is its ability to view the system in its entirety. The researchers did not rely on a sample or projections, but on real, comprehensive data from across France. “We used official data provided by the National Institute of Statistics and Economic Studies through the service of Secure Data Access Center (CASD). In total, we had access to information on about 30 million workers and employers in France, whom we tracked over a 10-year period,” explains Max Knicker, a PhD candidate at the École Polytechnique in Paris, affiliated with the EconophysiX lab and lead author of the study. The team worked with high-resolution administrative data from INSEE (the French National Institute of Statistics), specifically the BTS-Postes (Base Tous Salariés – Postes).

“We then assigned each occupation a score for two key metrics: transferability and accessibility,” Knicker explains.“On one hand, Transferability captures how diverse the set of occupations is that people move into from a given occupation. Accessibility, on the other hand, measures how diverse the origins are of people entering a given occupation, indicating how broadly accessible it is from across the labor market.”

Knicker and colleagues mapped all occupations onto a two-dimensional space defined by these variables, identifying four main clusters or categories of jobs. “Diffuser occupations are those with high transferability but low accessibility, they’re harder to enter but offer a wide range of exit opportunities,” Knicker explains. “Channel occupations are both hard to enter and offer few onward transitions. Hubs are both widely accessible and highly transferable, acting as central nodes in the transition network. Finally, the most common type are condensers — occupations that many workers can enter from diverse backgrounds, but which offer limited options for moving onward.”

Hubs, condensers, diffusers, and channels

“Hubs include jobs like retail sellers, which require a broad but not highly specialized skill set,” says Knicker. “Condensers include caregiving roles: technically, there aren’t many barriers to entry, but once someone enters a condenser occupation it’s hard to transition to something else. As for diffusers—where few occupations enter but from which it’s easy to move elsewhere—we’re talking about roles like technical flight managers or merchant navy specialists: jobs that require specific training to enter, but that training enables transitions to many other areas.”

“Lastly,” Knicker concludes, “channels are jobs that are hard to access and hard to leave. They often involve highly specialized skills, such as industrial welding machine operators.”

“For now, it’s a descriptive analysis. We’re essentially looking at the past, not building predictive models yet. But even this descriptive framework helps us understand how transitions happen,” Knicker explains. “While the broader labor market is undoubtedly undergoing structural shifts due to technological and economic change, we found that the observed occupational transition patterns have remained relatively stable over the past decade. This empirical stability allows us to use the current structure as a meaningful baseline, enabling our metrics to highlight occupations where policy efforts might most effectively ease reallocation bottlenecks.”

According to Knicker, the insights from this and future studies could help guide efforts to promote smoother transitions within the labor market. “With our work, we aim to identify the occupations with the greatest potential to act as levers or bridges, facilitating people’s movement from one job to another.”

Knicker and his team are making their methodology available to anyone who wants to apply it to other contexts—for example, to other European countries or even across the entire EU. One current challenge, however, he explains, is data standardization: some countries have extensive datasets comparable to France’s, while in others the picture is more fragmented. Still, the study just published is only a first step. In the future, Knicker and colleagues hope to track individual career trajectories and integrate other types of data, such as information on specific vocational training.