Are jobseekers looking too narrowly for jobs?


This page in:

I had a close friend in grad school who was doing a PhD in Medieval French Literature. We used to joke that instead of Job Openings for Economists, the way she would find job openings would be through the obituary pages – the only openings for professors in this field were to replace those who had died. Twenty years later she is to be found writing and editing dialog for story-driven video games, after a stint working as an asset analyst in a community bank.

Her story has to mind to me recently as an inspiring exception to a body of recent work that argues that people focus too much on the stereotypical occupation for a field of study when both choosing what to study, and then when looking for jobs – suggesting the need for policy interventions that help encourage/steer people towards considering looking beyond jobs that they were specifically trained for. Most of this evidence is from developed countries as far as I know, and it is unclear how much this is also an issue in many developed countries – but I’ll discuss what I know and then welcome any other suggestions of related literature.

When choosing college major, U.S. undergrads have incorrect beliefs about how majors translate into occupations

One paper that got me thinking about this recently is the very nice job market paper by John Conlon (with Dev Patel). They do surveys and an experiment with first-year college students at the Ohio State University. They hypothesize that students stereotype majors, so that when considering a major, they imagine the career they would have if they pursued the major is the most obvious one associated with the major, whereas this is very often not the case.  For example, they find 65% of prospective art majors expect to be artists (only 17% are), 63% of biology majors expect to be doctors (23% are), 42% of communications/journalism majors expect to be writers or journalists (4% are), 62% of psychology majors expect to be counselors (21% are), etc. In contrast, unless they are pursuing business or education majors, few students expect to end up in business or to become teachers. Moreover, for some majors, if you don’t end up being one of the rare people working in the stereotyped profession, the other alternatives tend to be low-paying (e.g., fine arts, humanities, communications, psychology), whereas other majors like STEM, business and STEM either have well-paying alternatives, or it is much more common to end up in the stereotyped job.

They then do a light-touch information intervention with a sample of 814 first-year students enrolled in the Exploration program for students who have not yet declared a major. They get asked their beliefs about their likely career in their most probable major, as well as in their second-ranked and two other majors. Half then get randomly selected to receive statistics on the actual joint distribution of majors and careers. The authors find that students who were considering “risky” majors like humanities, psychology, art, and communications and who overestimated the most common career paths in these majors then had a 7.5 percentage point lower likelihood of wanting to pursue that major, and, less precisely, appear less likely to actually enroll in classes in that major in the next semester.

Once people have studied in a major or worked in an occupation, they may only consider looking for jobs in that occupation

A second line of papers suggests that job-seekers often search too narrowly for jobs, only considering occupations they have worked in before or trained in, and not alternative occupations that could use their skills. Here are some recent experiments in developed countries that test this idea and interventions to get job-seekers to search more broadly:

1.       Belot et al. (2019, ReStud) – they do an RCT with 300 unemployed job-seekers in Edinburgh, Scotland. They invite both treatment and control to 12 weekly sessions to search for real jobs on their web interface. The control group uses their own search criteria, whereas the treatment group has an algorithm based on representative labor market statistics recommend additional relevant alternative occupations where skill transferability is high. They find that exposing jobseekers to jobs from a broader set of occupations leads to them applying to a broader set of jobs, and getting 44% more job interviews – from 0.61 interviews per week to 0.89 interviews per week. However, they note one of the big difficulties facing such studies is the challenge of tracing this through to impacts on employment- only 2 percent find a job each week, and to pick up a 44% increase in job finding would require a sample size of 3,794 per treatment! Another point to note is that over time people start broadening their search – so perhaps the treatment just speeds up the learning and exploring that would eventually happen anyway – their treatment effect is equivalent to what happens naturally over 9 weeks of job search.

2.       Dhia et al. (2022, WP) – this is a large-scale, low-intensity intervention. They work with the French employment agency and a sample of 212,277 individuals. They evaluate the fun-named “Bob Emploi”, an online website that aims to assist and motivate job seekers by providing personalized and data-informed advice on sectors and locations to target; offering step-by-step planning assistance; sending regular reminders and encouragement messages; and providing general tips. So it bundles broadening job search with other types of job search assistance. The website is publicly open, but new, and the paper is a good example of an encouragement design working well. 56.3% of the sample get assigned to treatment, and get invited to an information session to learn about Bob Emploi, and then three emails encouraging them to create an account. Take-up is 27.2% in treatment vs 0.2% in control – a really successful encouragement! Using admin data the authors can then follow the sample for 18 months, coupled with one survey. The authors find pretty precise null effects – no impacts on time spent looking for a job, occupational (or geographic) search scope, or employment. “Considering the upper bound of the 95 percent confidence interval, we can reject any effect higher than 0.4 percentage points on experiencing some employment episode within the 18 months following the intervention and any reduction of the time spent unemployed larger than two days”.

3.       Belot, Kircher and Muller (2022, WP) – the team follow-up the first paper listed above with an attempt to get impacts on employment. They conduct an experiment with around 1500 long-term job-seekers  (most unemployed for more than a year, many with a disability) in the UK. A private company contracted by the UK government to help these job-seekers worked with the researchers to integrate advice into the job search engine. Individuals could specify occupations they were searching for as “job goals” and would be shown vacancies in their geographic area. 60% were randomly assigned to treatment. Treated individuals were suggested up to 10 additional occupations based on data on common occupational transitions, and/or on whether they largely require the same skills. About 40% of the job-seekers never use the platform, and this is unrelated to treatment status. If they use the full sample, there is a positive but insignificant impact of around 2 percentage points on finding a stable job over the next year. Restricting to those who use the platform, the impact is 3-5 p.p. relative to a control mean of 16 percent, and is significant at the 10 percent level in some months. In pre-registered heterogeneity analysis, impacts are stronger for those with above median unemployment duration who used the platform (7p.p.).

It can be hard to get people to consider other jobs

The above evidence suggests that many job-seekers do not look broadly at jobs in a range of relevant occupations, and that pushing them to consider other occupations may have some impact on getting more interviews and jobs – but the effects are often modest in magnitude. One reason for these modest effects is that job-seekers may be very reluctant to consider these other career paths. In a previous experiment in Jordan (Groh et al, 2015), my co-authors and I worked with young educated job-seekers and provided a job-matching service based on skill tests and psychometric assessments. We matched job-seekers to jobs suited for their skills, but found they rejected the opportunity to even have an interview in 28 percent of cases, and when a job offer was received, graduates rejected this offer or quickly quit the job 83 percent of the time. When we examined the reasons why youth turned down potential job openings, the main reason given was not that the salary was too low, but instead, recent graduates explained that the job is “unsuitable”, or “not on the right career path.” Educated youth appear unwilling to take on certain types of jobs. This connects us back to Conlon and Patel’s stereotype work – and perhaps the need to sensitize youth early on, and continually during study, to the many possible career paths from one field of study.

Any other evidence on too narrow an occupational job search from developing countries? 


David McKenzie

Lead Economist, Development Research Group, World Bank

Join the Conversation