By James Albrecht, Bruno Decreuse and Susan Vroman
When vacancies are filled, the ads that were posted are generally not withdrawn, creating phantom vacancies. The existence of phantoms implies that older job listings are less likely to represent true vacancies than are younger ones. We assume that job seekers direct their search based on the listing age for otherwise identical listings and so equalize the probability of matching across listing age. Forming a match with a vacancy of age a creates a phantom of age a and thus creates a negative informational externality that affects all vacancies of age a or older. The magnitude of this externality decreases with a. The directed search behavior of job seekers leads them to over-apply to younger listings. We calibrate the model using US labor market data. The contribution of phantoms to overall frictions is large, but, conditional on the existence of phantoms, the social planner cannot improve much on the directed search allocation.
This is a cool model, a great way to teach graduate students about the state space and how to write the model in recursive form. If I were still teaching, it would have been an exercise or exam question… Also: interesting question, despite the conclusion that not much can be done about it.