Low-skilled jobs and the return to labor market experience and language training among immigrants in Sweden

The aim of the study is to test whether language training, i.e., successfully finishing SFI, and/or having experience from a low-skilled occupation is associated with a higher call-back rate from employers compared to not having gone through such training and having no such experience, respectively, for Middle-Eastern immigrants in Sweden when applying for other jobs with low formal qualification requirements. Theoretically, we expect both experience and language training to signal having more human capital and, thus, higher productivity.

The foreign-born fictitious applicants were born in a Middle-Eastern country in 1996 and have a high-school degree from the capital of that country. They received their Swedish residence permits in 2016 and are living in a suburb of Stockholm at time of application. This particular suburb does not stand out as either a high- or low-income area in relation to other suburbs. We chose to make the applicants relatively young in order to avoid having to define very detailed employment and unemployment histories before coming to Sweden. Half of all applicants are male and half are female. All applicants report having registered with the Swedish Public Employment Service (Arbetsförmedlingen) in August 2016.

There are four different types of applicants and one male and one female of each type. These eight applicants all have a unique (Arabic-sounding) name. For each vacancy, we randomly select one of the eight applicants. The first type has no work experience and does not claim to have finished Swedish For Immigrants (SFI). The second type has no work experience but claims to have finished the SFI program, while the third one does not mention anything about SFI but has since November 2017 been employed as a restaurant assistant at a well-known fast food chain in Stockholm. Finally, the fourth type both claims to have finished SFI and has since November 2017 been working at the same restaurant chain as the third type.

2019-01-16

2019-12-31

The primary outcome variable is an indicator variable which takes the value 1 if the employer contacted the fictitious applicant either via phone or email—i.e., if the applicant received a call-back—to set up an interview, to make a job offer or asking for more information.

We will utilize two secondary outcome variables. The first only takes the value 1 if the fictitious applicant was contacted regarding setting up an interview or being offered a job. Moreover, we will use another secondary outcome where the reason for the call-back is not considered. In other words, all responses induce an indicator value of 1, including those that inform the applicant that he/she did not get the job.

The correspondence test includes job vacancies from all over Sweden, which are applied to electronically. The population consists of all relevant job vacancies that are posted during the study period on the Public Employment Service’s (Arbetsförmedlingen) job board, which is called Platsbanken. Relevant job vacancies are those that: do not require previous experience; do not require any qualifications that are not stated in our fictitious resumes, such as language proficiency apart from Swedish, a driver’s license et cetera; do not require additional information that cannot be provided, such as social security numbers, a photo of the applicant et cetera. Moreover, only vacancies that are classified as elementary occupations according to the Swedish version of the International Standard Classification of Occupations are applied to. In practice, we are applying to vacancies in cleaning, home care, restaurant assistance, newspaper delivery and packaging in manufacturing. Lastly, we only apply to jobs in the private sector since applications to public-sector vacancies require social security numbers.

For each relevant vacancy at Platsbanken, we randomly choose one applicant (with the corresponding resume and personal letter) and send the application using an e-mail address specifically created for that applicant.

One of the eight different applicants are randomly selected to apply to each vacancy.

No

3,000 applications/vacancies

3,000 applications/vacancies

In expectation, approximately 750 applicants will have neither language training nor experience from a low-skilled job, 750 will have training but no experience, 750 will have experience but no language training, and 750 will have both experience and training.

Based on a call-back rate of five percent and 3,000 observations, 80 percent statistical power and a critical limit of 5 percent, we will be able to detect effect sizes of just below 2.5 percentage points. (See the analysis plan for details regarding the power calculations.)