Persistent unemployment across OECD countries has led to increasing investments in activation programmes and, as a consequence, rigorous evaluations of the effectiveness of these programmes. The results of these evaluations have been mixed at best. To improve the effectiveness of the activation programmes, it is important to know why we observe these unsatisfactory results. One possible explanation that has been largely underexplored is the signal these programmes send to prospective employers. We investigate this signalling effect in the context of a job-vacancy referral system. To this end, we conduct a state-of-the-art vignette experiment in which HR professionals make hiring decisions concerning fictitious job candidates who apply either under a job-vacancy referral system or directly (without referral). By analysing the experimental data, we provide first causal evidence for a substantial adverse effect of referral on the probability of being hired. In addition, our experimental design allows us to explore whether this effect is heterogeneous by job candidate and recruiter characteristics and what exactly is signalled by the job-vacancy referral. In particular, we find that employers perceive referred candidates as being less motivated than other candidates.