We develop a partial equilibrium stochastic job matching model of the labor market to examine whetherthe counseling of unemployment workers displaces unemployed workers not offered the counseling. Inthis model, the improvement of the reservation utility of counseled job seekers induces them to refusejob offers that they would have accepted if they were not counseled. This behavior, which exerts anegative spillover on job creation, reduces the arrival rate of job offers to the unemployed workers whodo not benefit from counseling. The model is estimated on data concerning intensive counselingschemes that are provided to about 12.5 percent of the unemployed workers since the 2001 Frenchunemployment policy reform (PARE). We find significant favourable effects of counselling on the exitrate out of unemployment of counseled workers. We also find that counselling reduces the exit ratefrom unemployment of workers not offered the counseling. This effect is sufficiently large to imply thatcounseling reduces the average exit rate from unemployment of counseled and non counseled workerswhen a small share of unemployed workers are counseled. This result puts to the fore that evaluationsrelying only on comparisons of differences in outcomes between a treatment group and a control groupcan lead to misleading conclusions even when a small share of the population is treated.