Digital Tools to Facilitate Job Search

We conduct a country-wide field experiment with unemployed job seekers in Denmark. Our experiment makes use of a new online information tool through which the Danish employment agency provides online job search assistance. The online tool - a dashboard containing different information cards - is embedded on the landing page reached by job seekers after they log in to the public portal of the Danish employment agency (jobnet.dk). In our experiment, we exogenously vary the content of the information tool across four groups of job seekers: a control group receives generic information on the functionalities of
the job portal. A second group receives additional information about the number of vacancies available in the occupations stored by job seekers in their personal search profile. The third treatment group receives referrals to occupations that are related to the ones specified in their personal search profile. A fourth group of job
seekers receives both vacancy information and occupational referrals. We will analyze the effects of our treatments using administrative data
on individuals' job search behavior and subsequent labor market outcomes. For more details see pre-analysis plan.

2019-03-16

2021-08-01

The overarching aim of our experiment is to understand how providing job seekers with information regarding alternative occupations as well as the number of vacancies in their occupational consideration set affects individuals' job search behavior and labour market outcomes (such as employment, earnings, occupations). For more details see section 3 in the pre-analysis plan

For more details see section 3 in the pre-analysis plan

Our randomized controlled trial focuses on providing online job search assistance to unemployed workers in the Danish labor market. The intervention aims (i) at providing information about the number of vacancies available in the "occupational consideration set" of treated individuals and (ii) at encouraging job seekers to broaden their focus in
terms of the occupations they consider. Our experiment makes use of a new online information tool through which the Danish employment agency provides online job search assistance. The online tool - a dashboard containing different information cards - is embedded on the landing page reached by job seekers after they log in to the public portal of the Danish employment agency (jobnet.dk). In our experiment, we exogenously vary the content of the information tool across four groups of job seekers: a control group receives generic information on the functionalities of
the job portal. A second group receives additional information about the number of vacancies available in the occupations stored by job seekers in their personal search profile. The third treatment group receives referrals to occupations that are related to the ones specified in their personal search profile. A fourth group of job seekers receives both vacancy information and occupational referrals. We will analyze the effects of our treatments using administrative data
on individuals' job search behavior and subsequent labor market outcomes. For more details see section 2 of the pre-analysis plan.

For more details see section 2 of the pre-analysis plan.

in office done by a computer

Individual level with regional stratification: We assign each municipality to one of three treatment intensities: 1) super-control regions, 2) low-intensity regions and 3) high-intensity regions. In the super-control regions all unemployed workers are assigned to the control group (C). In low and high intensity regions, unemployed workers will randomly be assigned to either the control group (C) or one of the three treatment groups (V, R or VR). Specifically, in low-intensity regions, job seekers are assigned to the control group with a probability of 40%. They are assigned to treatments (V), (R), (VR) with a probability of 20% each. In high-intensity regions, job seekers are assigned to each of the three treatments with probability 30%, whereas a
job seeker's probability of being assigned to the control group is 10%.

Yes

Two level randomization design. Treatment status is assigned on the individual level with varying treatment intensities at the municipality level (98 municipalities).

Estimate: about 140.000 individual job seekers

Participants in our experiment include the universe of unemployment benefit recipients and "job-ready" welfare recipients at the beginning of the intervention and the corresponding inflow of individuals entering unemployment between March 16 and July 31, 2019. Estimate: about 140.000 participants.

See section 4.4 of the pre-analysis plan.