A Test of the Signalling Hypothesis - Evidence from Natural Experiment

The paper proposes an alternative methodology for testing signalling hypothesis based on chances to get a job in a particular class of the job market. The individuals are ranked and matched by an external mechanism, based on preferences of employers in respect to actual observable and perceived unobservable attributes of individual. This paper tests existence of a relation between the set of observable and revealed attributes and the outcome of the game, specifically: whether signals associated with attained education plays a significant role in determining chances of the individual to get a job. The proposed model is empirically tested by applying a unique dataset from a natural experiment, conducted in Poland in years 2002-2005, where a relatively large set of job market candidates are offered a chance to get a paid internship at an attractive employer, with considerably great chances of getting a permanent job thereafter. Results support the hypothesis, that in the absence of revealed attributes, employers decisions depend upon signals on education. Whenever information is available, the significance of the signals diminishes.