Johan Sæverud, University of Copenhagen (Job Market Seminar)

"Subjective Earnings Risk"

Abstract

Earnings risk is in essence subjective. Yet, it is typically inferred from the cross sectional distribution of realized earnings growth observed in administrative data by making strong assumptions about homogeneity within groups of people. We collect new data using a custom survey designed to elicit subjective earnings expectations in order to characterize the actual nature of subjective earnings risk. The distinguishing feature of our survey instrument is that it elicits the entire subjective probability distribution over future earnings conditional on possible job transitions, i.e., whether people stay in their current job, quit or are laid-off. We create individual measures of overall earnings expectations by aggregating subjective probability distributions over future earnings across job transition states. By pooling individual data on subjective earnings expectations and comparing with the administrative data, we first show that the survey data is consistent with the cross sectional distribution of realized earnings growth that is observed in administrative data. We then turn to consider the subjective expectations and find that individual earnings risk inferred from administrative data is highly distorted: subjective earnings risk is on average far lower, and it is heterogeneous both within and across demographic groups. Furthermore, we document that overall subjective earnings risk is dominated by the job transition risk. We close the paper by showing how implied earnings expectations in a life cycle model of search and matching calibrated to administrative data on job transitions over-states how people actually perceive earnings risk. Our findings show that people tend to have more information about earnings risk than what current estimates of search and matching models imply. Overall, our results suggest that subjective risk has been massively overstated when inferred from data on realized earnings and job transitions.

Contact person: Mette Ejrnæs