Abstract

We run a large-scale field experiment in which we provide information to students at randomly selected schools about the job opportunities and hourly wages of a small set of occupations they are interested in. The experiment takes place on an online career guidance counseling platform that is widely used in the Netherlands, and involves 28,267 pre-vocational secondary education students in 243 schools over a period of 2 years. We find that the information improves the accuracy of students' beliefs, both in the short run (for job opportunities and hourly wages) and in the long run (for job opportunities only). Students who receive the information also tend to change their favorite occupation 0.88 to 2.16 percentage points more often, and switch towards an occupation with better labor market prospects if they do so. Last, and most importantly, they select secondary school specializations related to occupations with better labor market prospects (1.5% and 0.3% -- €0.05 an hour -- higher than the control group mean for job opportunities and wages, respectively) and choose post-secondary education programs with higher expected wages (2.5% -- approximately €0.40 an hour -- higher than the control group mean).

Abstract

Using data from the New York Federal Reserve's Survey of Consumer Expectations, we study how the United States Bureau of Labor Statistics' Employment Situation Reports (Jobs Reports) affect individuals' expectations about the likelihood of losing their own job. We do this in two steps. First, we estimate the information shocks of the Jobs Reports on expectations about the development of the national unemployment rate in the next twelve months. We do this by comparing survey responses shortly before and after publication of the reports. Second, we estimate how these shocks affect individuals' expectations about losing their own job in the same time frame. The results show that when a report is estimated to increase beliefs about the likelihood of the unemployment rate increasing by 1 percentage point, beliefs about the likelihood of personal job loss during that time increase by up to 0.22 percentage points. We further find that the information shock negatively affects individuals' beliefs about the likelihood of finding a new job if they were to lose their current one, but (surprisingly) positively affects individuals' beliefs about the likelihood of voluntarily leaving their job. Our results are robust to the use of different bandwidths around the reports' publication dates and placebo treatments provide reassurance that the information shock is indeed the mechanism driving the result.

We study the impact of online information provision to job seekers who are looking for work in occupations with poor labor market prospects. The information is provided through a personalized email containing suggestions about suitable alternative occupations and how the prospects of these alternatives compare to the job seekers' current occupation of interest. We additionally include a link to a motivational video for parts of the treatment group. We evaluate the interventions using a randomized field experiment covering all registered job seekers in the target occupations, where two thirds are treated. Our email is opened by the vast majority of job seekers, revealing the alternative suggestions. The motivational video link is rarely used. Effects on unemployed job seekers in structurally poor labor markets are large: over the 20 months following the intervention, treated job seekers are 2.5 percentage points more likely to be employed, work over 50 more hours, and earn €800.- more. Additionally, they are more likely to end up in a different occupation than they initially targeted. There is little impact on job seekers in occupations which did well prior to and after the Covid-19 lockdowns.

 We study the impact of published student satisfaction scores (ranging from 1 to 5) on enrollment of first-year students for the near universe of higher education programs in the Netherlands between 2011 and 2019. We use pageview data from the largest Dutch educational information website to determine each programs' closest substitutes. This allows us to not only analyze the impact of changes in a program's own published student satisfaction score, but also the impact of changes in the student satisfaction scores of its substitutes. We analyze the impact of these satisfaction scores using fixed effects Poisson regressions and exploit rounding discontinuities to identify causal effects. Our findings show that student satisfaction scores matter for enrollment. An increase in a program's student satisfaction score leads to higher levels of enrollment, whereas an increase in the student satisfaction scores of substitutes leads to lower levels of enrollment. Point estimates of the impact of a program's student satisfaction score being rounded up to the next tenth on first-year enrollment range between 1.70% and 3.52%, depending on the bandwidth around the threshold we consider. Conditional on being above the rounding threshold, a program being rounded up over at least one of its closest substitutes increases first-year enrollment by up to 4.37%.