NEW UPDATE: Completed trials may now upload and register supplementary documents (e.g. null results reports, populated pre-analysis plans, or post-trial results reports) in the Post Trial section under Reports, Papers, & Other Materials.
Counseling the Unemployed in France (OPP/CVE)
Initial registration date
December 23, 2015
December 23, 2015 5:43 AM EST
Other Primary Investigator(s)
PSE-École d’économie de Paris
PSE-École d’économie de Paris
Additional Trial Information
This project reports the results of a large-scaled randomized controlled experiment comparing the public and private provision of counseling to job seekers. The intention-to-treat estimates of both programs are not statistically different, but more workers were enrolled in the private program, implying an effect per beneficiary that is twice as large under the public as under the private program. We find suggestive evidence that the private firms may have insufficiently mastered the counseling technology, and exercised less effort on those who had the best chance to find a job. This highlights the incentive problems in designing contracts for these services.
All jobseekers involved in the experiment initially passed through the main public unemployment agency ANPE. People in the experiment are jobseekers identified by caseworkers as "at risk of long-term unemployment" and were classified into three different categories:
Population 1: Newly unemployed at risk of long-term unemployment - people who have just entered unemployment and are eligible for more than 120 days of unemployment benefits;
Population 2: Stock Unemployed at risk of long-term unemployment - those who were already unemployed at the beginning of the experiment;
Population 3: Ineligible Newly Unemployed at risk of long-term unemployment - people that have just become unemployed, but are eligible for less than 120 days of unemployment
Two separate programs were designed to support job seekers who have been, or are at risk of long-term unemployment: (1) Cap Vers l'Entreprise (CVE), provided directly by ANPE; (2) Opérateurs Privés de Placement (OPP), supplied by private firms mandated by UNEDIC. Each program offered similar forms of assistance and support which consisted of weekly personal counseling and monitoring by a counselor who worked with a limited number of jobseekers. The counseling lasted for at most six months and aimed to assist jobseekers in finding stable employment. This included an intensive follow-up, with at least a weekly contact (email, phone) and a monthly face-to-face meeting between the job seeker and his personal counselor. Compared to the usual track, where a contact is supposed to take place every month, this is a significant increase in human resources dedicated to follow the job seeker. The support also included continued monitoring of the individuals in their new employment.
The OPP program only accepted ineligible newly unemployed people (population 1). The CVE program accepted all three populations (population 1-2-3)
Intervention Start Date
Intervention End Date
Primary Outcomes (end points)
Data on employment status was collected for three, six and nine month intervals from administrative records and telephone surveys on a sub sample of jobseekers. In ongoing analysis, researchers are interested in estimating not only the average duration of unemployment, whether the counseling affects long term unemployment levels through a better matching of individuals to jobs.
The main outcome variables are employment outcomes (especially at six months) but also at 9 and 12 month
Primary Outcomes (explanation)
Our analysis mainly focuses on employment outcomes at six months because it corresponds to the end of the counseling service. As such, the objectives of both programs are to bring job seekers back to employment within six months, and payment to the private contractors is conditional on this. However, we also look at employment outcomes at other horizons: three months and 12 months. We consider outcomes at three months to measure a fast impact on employment, which is desirable in particular if the objective is to save on unemployment benefits. We also look at employment outcome 12 months after random assignment so as to be able to evaluate medium-term effects of the counseling programs.
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Registered jobseekers of population 1 were randomly assigned into each of the three groups: CVE, OPP, or the normal ANPE track (control group), which included much less intensive support.
Jobseekers from the "stock unemployed" (population 2) and "ineligible newly unemployed" (population 3) were not eligible for the OPP program, and so they were randomly assigned to either the CVE program or the normal ANPE track.
Experimental Design Details
Used Extranet program to randomly assign.
individual (job seeker)
Was the treatment clustered?
Sample size: planned number of clusters
Number of observation was not planned. Up to 40,000 slots were available for OPP and up to 40,00 for CVE. Entry in the program was maintained over a year. All eligible job seekers were registered in the experiment and allocated to treatment groups based on probabilities regularly up dated so as to meet capacity constraints and requirements.
Sample size: planned number of observations
same as clusters
Sample size (or number of clusters) by treatment arms
Population 1: OPP 45460; CVE 9875; Regular 13055
Population 2: CVE 29257; Regular 34684 Population 3: CVE 33315; Regular 53387
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
The probabilities of assignment to each group varied locally so as to maximize the statistical power of the evaluation while complying with the quantitative objectives of each program (each local area had targets in terms of recipients of the two programs). This often implied very high probabilities of assignment to the private program (up to 85 percent), and much lower probabilities of assignment to the public program (down to 6 percent) and control (down to 9 percent).
INSTITUTIONAL REVIEW BOARDS (IRBs)
Post Trial Information
Is the intervention completed?
Intervention Completion Date
September 30, 2008, 12:00 AM +00:00
Is data collection complete?
Data Collection Completion Date
March 30, 2009, 12:00 AM +00:00
Final Sample Size: Number of Clusters (Unit of Randomization)
Was attrition correlated with treatment status?
Final Sample Size: Total Number of Observations
The overall attrition rate, collapsing administrative records, and survey information, was limited to 9 percent, and well balanced among the three experimental groups.
Final Sample Size (or Number of Clusters) by Treatment Arms
Total : 43977
Control group 4565
Assigned to public program 3385
Assigned to private program 36027
Administrative data with some surveys on sampled subjects (where administrative data is missing)
see paper table 1 page 149
Reports, Papers & Other Materials
The paper reports the results of the experiment for jobseekers of population 1 (job-seeker newly unemployed with at least 120 days of unemployment benefits) in areas were private and public services were jointly offered
This paper reports the results of a large-scaled randomized controlled experiment comparing the public and private provision of counseling to job seekers. The intention-to-treat estimates of both programs are not statistically different, but more workers were enrolled in the private program, implying an effect per beneficiary that is twice as large under the public as under the private program. We find suggestive evidence that the private firms may have insufficiently mastered the counseling technology, and exercised less effort on those who had the best chance to find a job. This highlights the incentive problems in designing contracts for these services.
Behaghel, Luc, Bruno Crépon, and Marc Gurgand. 2014. "Private and Public Provision of Counseling to Job Seekers: Evidence from a Large Controlled Experiment." American Economic Journal: Applied Economics, 6(4): 142-74.
REPORTS & OTHER MATERIALS