Can information correct optimistic wage expectations?

Forward-looking expectations are central to job search but often inaccurate. To test whether public information can help correct beliefs, we embed an experiment in a longitudinal survey of Mozambican graduates. We quantify responses of own-earning expectations to information about peers’ current earnings, sent by SMS.

Optimistic beliefs were revised downward by a larger margin in the treatment group. But, consistent with a theoretical on-the-job search model, responses to news do not conform to a linear updating framework, with strong evidence for limited responses to negative news. This explains the moderate impact of our intervention and why optimistic expectations remained persistent.