#	Pagina
attuale pagina	/open-h2020/projects/196020/results.html

Report

Teaser, summary, work performed and final results

Periodic Reporting for period 1 - ALMP (The Effect of Active Labour Market Policies on the Behaviour and Employability of Benefit Claimants)

Teaser

Summary

Unemployment is dramatic both at the individual level and for society. Long-term unemployment can lead to a permanent reduction of mental and physical health, can reduce earnings potential and a high unemployment rate is a burden for public finances. Moreover, we know that a long unemployment spell sends a very bad signal to potential employers, even if one remains motivated to look for a job. Employers are less likely to invite an applicant for an interview who has been unemployed for a year than a similar applicant who has only been unemployed for a short time.

It is hence important to find effective strategies to get people back into work and to detect those who are at risk of becoming long-term unemployed as soon as possible.

While the literature has shown mixed evidence on the effectiveness of resource-consuming government policies to help the unemployed into work (Active Labour Market Policies), this project found that cheap and light-touch counselling programmes right at the start of the unemployment spell are effective to improve job finding rates. This is of societal interest as in times of austerity, governments are looking for cost-effective strategies to reach their policy goals. The project also adds to the debate, embedded in the era of austerity, on substituting human coaching with self-service online platforms: while the latter might be useful, a minimum of human face-to-face interactions can be very cost-effective.

This project also found ways to better predict which people are at risk of becoming long-term unemployed, which can help to better target such groups right from the beginning. Public Employment Services are increasingly interested in profiling their unemployed using big data. Usually they rely on available administrative data, and socioeconomic variables such as age, education, and work experience have proven to be good predictors of oneâ€™s success on the labour market.

We found that beyond all these factors, measures of peopleâ€™s cognitive skills (notably numeracy skills) can predict how well people will fare on the job market. These skills can be measured with simple survey questions that are not time-consuming: one can ask interviewees to perform a few small calculations, for example. Since conducting surveys has become much easier thanks to online tools, we are hopeful for employment offices to include such questions in their surveys in order to refine their profiling models as to be able to target their clients more effectively.

Work performed

This project is mainly empirical. It involved cleaning and analysing both administrative data and data from surveys. The project also involved a significant methodological component. Indeed, there are beliefs that repeatedly interviewing people might have an effect on how people answer questions (e.g. they might give more accurate answers) or it could even have an effect on their behaviour. In this project, we performed methodological work to find out how we can measure such effects and, if present, how we can make sure that they do not lead to misleading conclusions. The project has led to three main deliverables, which are academic reports. Two of them are still being improved and a third one has been sent to an academic journal for publication. Results have also been presented at academic conferences, but also at several meetings with European Public Employment Services and at events for the wider public. A few internal technical reports for policy makers have been produced as well, which intend to lay the groundwork for future research related to this project.

Final results

It has been proven to be surprisingly hard to pinpoint the most effective interventions to get people (back) into a satisfying job. Changes in the unemployment rate do not only depend on labour market policies but can be caused by many other things such as a financial crisis. We might hence observe that in regions, or time periods, with many labour market programmes in place, the unemployment rate is high, not because the programmes do not work, but simply because a high unemployment rate leads to the implementation of such programmes. Similarly, people who are taking part in a certain activity (such as training, coaching, collective information sessions) are often found to be less successful on the labour market than those who do not participate. But often governments target those individuals who are most at risk of not finding a job, or individuals who decide to take part in such programmes are those that feel they might struggle to find a job on their own. Hence, participants might even have been worse-off if they would not have participated.

In order to be able to measure clean impacts, this project has been involved in Randomized Controlled Trials, well-established in Medicine but still relatively novel in the social sciences. A new policy is rolled out gradually: a random group of eligible individuals is subject to the new programme, while for the others, business-as-usual remains in place. This strategy helps to understand whether a new policy works better than the old one, and to make sound policy conclusions. Throughout the project, we have been able to talk about this strategy to several Public Employment Services in Europe for which such strategies are still very new, and the project might well help to make sound policy evaluations through trials more common in Europe.

A second innovative aspect of this project entails the use of administrative data on individualsâ€™ labour market history matched with survey data. Profiling models of Public Employment Services, used to better target the unemployed, are generally making use of administrative data. This project, which is based on a relatively small dataset, has found that information on soft skills, that can be measured through surveys, can significantly improve these profiling models. This finding is likely to have a direct impact on policy making. At this moment, for example, we are involved into commissioned research which advises a European Public Employment Service on how to improve its profiling model. Obviously we are incorporating insights from our h2020 project and are helping them to put together a survey that can help to achieve a better profiling model.