Understanding Occupational Mobility

We design a new online survey and administer it to a large sample of 5,000 German employees (age 20 to 50), oversampling workers in occupations that are more at risk of being displaced by structural changes in the labor market – automation, green transition. The survey elicits workers’ beliefs about the transferability of their skills, the costs of moving to a different occupation – search costs, skills requirements, need and costs of re-training, need and cost of acquiring a license – and their perceptions of the occupations that are “adjacent” to theirs, probabilities of moving across occupations – and the benefits of making a career change – median wages in other occupations and starting wages if they were to move to different occupations. There are also additional questions aimed at understanding the main factors workers weigh in their decision to change occupations. The survey will have a panel structure. The first wave is going to be administered in November-December 2024 and a second wave is planned six months after. The main survey features multiple information experiments where random subsamples of respondents receive information on the transferability of their skills to other occupations paired with information on the costs or benefits of occupational mobility, and/or information on exposure to technological change or the green transition. For respondents who will provide their consent, we will be able to match survey records with administrative data on pre- and post-survey employment histories.

2024-11-04

2024-12-15

Beliefs about: skills transferability, close occupations in terms of skills similarity, common transitions between occupations, median wages and job availability in other occupations, own starting wage in other occupations, likelihood of needing re-training and licenses to access other occupations and effort to acquire these licenses/training, search costs to find a job in other occupations (applications, success rate, time costs, mental costs), probability of receiving an offer for a job in own or another occupation (arrival rate of offers), share of automatable and environmentally harmful tasks in own occupation, probability that automation or green transition will affect own earnings or employment.

Labor market intentions: likelihood to fook for a new job in own current occupation, likelihood to look for a new job in another occupation, likelihood to look for a new job in specific other occupations, likelihood to ask current employer for a raise.

We ask questions about the respondents' labor market intentions after the treatments. We elicit the beliefs listed above before the treatment. After the treatments, we elicit again respondents' beliefs about wages in selected occupations, how easy/difficult it would be to find a job in selected occupations, similarity (in terms of tasks) between own and selected occupations, likelihood of needing a license or training to access selected occupations.

We will also measure again respondents' beliefs and labor market intentions in the follow-up survey and ask about actual labor market behavior (whether respondents have changed jobs/occupations or have tried to do so, whether they asked their employer for a raise).
We plan to:
- study the effect of the treatments on respondents' beliefs and labor market intentions measured in the main survey
- study the effects of the treatments on beliefs and intentions as well as actual labor market behavior measured in the follow-up survey
- study the effects of the treatments on actual labor market transitions that we can observe in post-treatment administrative data over different horizons
- explore the heterogeneity in the treatment effects depending on respondents' characteristics (e.g., age, education, gender, region of residence, family situation, occupation type), employment histories (from the administrative data), and other variables and personality traits measured in the survey (e.g. having recently done training, how much importance respondents give to different job attributes, risk aversion, openness to new experiences).

The main survey features multiple information experiments where random subsamples of respondents receive information on the transferability of their skills to other occupations paired with information on the costs or benefits of occupational mobility, and/or information on exposure to technological change or the green transition.
More specifically, we split respondents into two groups depending on their occupation: respondents in occupations at risk of being displaced by automation or by the green transition -- i.e., those who work in occupations performing a high share of automatable task or environmentally-damaging tasks -- and respondent in occupations not at risk. At-risk respondents are then split into 6 mutually exclusive groups: 1) Control; 2) Benefit treatment; 3) Cost treatment; 4) Displacement risk (from automation or green transition) treatment; 5) Displacement risk + Benefit treatments; 6) Displacement risk + Cost treamtents. Not-at-risk respondents are split into 3 mutually exclusive groups: 1) Control; 2) Benefit treatment; 3) Cost treatment.

All treated respondents receive personalized reccomendations on two occupations that are close to theirs in terms of task similarity. Then, depending on the groups they are randomized in, they receive attitional information on:
- Benefit treatment: some of the benefits of moving to the two aforementioned occupations -- median, 25th and 75th percentile of the wage distribution, and qualitative information on the number of vacancies available relative to jobseekers
- Cost treatment: sone of the costs of moving to the two aforementioned occupations -- likelihood of needing to re-train or acquire a licences
- Displacement risk treatment: respondents randomized in the Displacement risk treatment are informed that people in their occupation perform a high share of automatable tasks or environmentally-damaging tasks and are provided with similar information on the two aforementioned alternative occupations.
- Benefits + displacement risk treatment: information on displacement risk (as in the previous point) plus information on the benefits as above
- Costs + displacement risk treatment: information on displacement risk (as in the previous point) plus information on the costs as above.

Participants in the control group do not receive any piece of information and move directly to the next block of the survey.

If respondents are in an occupations for which there are no close alternative occupations, they do not receive any treatment.

Not available

As they take the survey, depending on their self-declared occupations, respondents are first classified into two groups: respondents in occupations at risk of being displaced by automation or by the green transition -- i.e., those who work in occupations performing a high share of automatable task or environmentally-damaging tasks -- and respondent in occupations not at risk. At-risk respondents are then split into 6 mutually exclusive groups: 1) Control; 2) Benefit treatment; 3) Cost treatment; 4) Displacement risk (from automation or green transition) treatment; 5) Displacement risk + Benefit treatments; 6) Displacement risk + Cost treatment. Not-at-risk respondents are split into 3 mutually exclusive groups: 1) Control; 2) Benefit treatment; 3) Cost treatment.

Individual respondent.

No

1 country (Germany)

5,000 employees (wage and salary workers) who are working full-time at the time they take the survey.

Within group (at-risk vs. not-at-risk), respondents are evenly assigned to the groups outlined above
Not-at-risk respondents: we expect to have about 700 respondents per treatment group (or in the control group). At-risk respondents: we expect to have about 350 respondents per treatment group (or in the control group).