Michigan Job Seekers Study

Job seekers are assigned to Treatment Group A, Treatment Group B, or the control group. Treatment Grop A provides job-seekers with access to a website and smartphone app that provides tailored information about occupations providing relatively high wages and strong employment growth, along with nudges to search and apply for jobs. Treatment Group B receives access to a website and smartphone app that provides non-tailored information about occupations, along with nudges to search and apply for jobs. The control group receives general information.

Setting and Sample

Our partner is Michigan Works! Southwest (MWS), which operates federally-funded American Job Centers in Branch, Calhoun, Kalamazoo, and St. Joseph counties. MWS provides a variety of services under the Wagner-Peyser Act Employment Services, Workforce Innovation and Opportunity Act (WIOA), Trade Adjustment Assistance (TAA), and cash assistance programs. Only the five percent of clients who qualify for WIOA, TAA, or cash assistance programs receive intensive, one-on-one job search assistance. The remaining 95 percent, who qualify only for Employment Services, receive little tailored support. MWS staff believe that these clients' limited information about job opportunities hampers their job search. We focus on individuals eligible only for Employment Services, as the intervention is likely to have the largest impact and interpreting the results is more straightforward for this group.

Intervention

The RCT randomly provides some MWS clients with access to a website and smartphone app that provides tailored information about occupations providing relatively high wages and strong employment growth, along with nudges to search and apply for jobs. The control group receives general information. The main steps of the intervention are:

1) Job seekers who visit MWS complete an intake survey as part of the standard operating procedure. (Details on all surveys are provided below.) All individuals receive a username and link to the website/app via email and text message.

2) Individuals sign into the website/app using a MWS computer or their own computer or smartphone. Individuals complete a short baseline survey that measures their perceptions of labor market opportunities.

3) After individuals complete the baseline survey, we randomly assign them into one of two treatment groups or the control group. For treatment group A, the website/app provides tailored information, based on answers to the intake survey, about the occupations estimated to offer relatively high wages and strong employment growth. (We describe the statistical model that produces these estimates below.) We provide a range of predicted wages that individuals might earn in each occupation, plus links to current job postings from Indeed, one of the largest job sites in the US. For treatment group B, we provide information about the same occupations as in treatment group A, but we do not provide predicted wages in each occupation. For both treatment groups, we only provide information on \textit{relevant} occupations, defined as those in which a sizable number of workers with the same background characteristics are employed (to take an extreme example, we will not provide information about physician jobs to individuals with only a high school degree). Having two treatment groups allows us to identify whether the effects stem from narrowing down the set of all possible occupations to this relevant set (as is done for both treatment groups) or from providing quantitative predictions about wages in each job (done only for treatment group A). Individuals in both treatment groups can use the website/app whenever they want. The website/app sends regular email and text messages that encourage individuals to search for a job.

For the control group, the website/app provides general information -- a list of websites where jobs are posted, such as Indeed and Pure Michigan Talent Connect -- currently provided to MWS clients.

4) Short weekly text message surveys ask individuals about job search activity. An endline survey asks more questions about their employment situation.

2019-07-01

2019-10-21

The four key outcome variables are:
i) earnings information bias (i.e. the bias in beliefs about how much different occupations pay)
ii) whether or not individuals switch two-digit occupation categories
iii) weekly earnings 12 weeks after filling out the baseline survey and starting the intervention
iv) days nonemployed within first 12 weeks after filling out the baseline survey and starting the intervention

i) will be measured based on surveys asking people to rank 6 randomly chosen occupations form 1-6, with 1 offering the lowest wages for the individual and 6 offering the highest. We will answer this both before and 12 weeks after the information intervention. We will measure information quality as the within person correlation of the responded occupation ranking with the true occupation ranking.

i) Number of jobs-applied for per week
ii) Number of jobs-applied for in new occupations (i.e. different than previous occupation) per week

Job-seekers coming into four Michigan one-stop job centers will fill out an intake survey the first time they come in, as part of the normal process of beginning to receive services at the one-stop job center. Individuals completing this intake survey will be asked if they want to participate in a research study about job-search tools. If they provide their informed consent, they will be invited to go to the app/website to fill out a baseline survey on their labor market beliefs. After completing this baseline survey, job-seekers will be randomly assigned to be offered website and app Treatment A, Treatment B, or the control treatment, all described in more detail above.

Randomization will be done in the job-search app/website by computer. Individuals are randomized individually when they fill out the baseline survey on the job-search app/website.

Individual job-seeker randomization into Treatment A, treatment group B, and the control group.

No

4 Michigan Works Southwest Ofices

2500 job-seekers

834: Treatment Group A (tailored job information)
833: Treatment Group B (general job information)
833: Control Group

i) earnings information bias: unit is correlation, baseline mean is unknown, so we can only say that we estimate that the minimum detectable effect size will be .17 standard deviations ii) whether or not individuals switch two-digit occupation categories: unit is share switching two digit occupations, the minimum detectable effect size will be .08 percentage points, or .15 standard deviations, which is 33% of the expected baseline mean. iii) weekly earnings 12 weeks after filling out the baseline survey and starting the intervention: the minimum detectable effect size is $61 per week, which is .12 standard deviations and 11% of the baseline mean iv) days nonemployed within first 12 weeks after filling out the baseline survey and starting the intervention: the minimum detectable effect size will be 2.4 days nonemployed within the first 12 weeks of filling out the baseline survey, which is .12 standard deviations or 4.7% of the baseline mean.