Job search with related information and wage signalling (Q1092781)

The paper uses multi-armed bandit methods to characterize the optimal policy of the job search problem with belated information, that is, if the non-pecuniary aspects of the job is discerned only after the job has been tested for one period. There is given a sufficient condition such that the `reservation wage property' for the optimal search policy remains preserved even if a job searcher believes in a positive resp. negative correlation between the wage rate \((=\) pecuniary aspect) and the job quality \((=\) non-pecuniary aspect) offered by a firm.