Working Paper: NBER ID: w22188
Authors: Fernando E. Alvarez; Katarna Borovikov; Robert Shimer
Abstract: We develop a dynamic model of transitions in and out of employment. A worker finds a job at an optimal stopping time, when a Brownian motion with drift hits a barrier. This implies that the duration of each worker's jobless spells has an inverse Gaussian distribution. We allow for arbitrary heterogeneity across workers in the parameters of this distribution and prove that the distribution of these parameters is identified from the duration of two spells. We use social security data for Austrian workers to estimate the model. We conclude that dynamic selection is a critical source of duration dependence.
Keywords: duration dependence; job finding; unemployment; heterogeneity
JEL Codes: E24; J64
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| the distribution of unobserved heterogeneity (D39) | the population distribution of the parameters of the inverse Gaussian distribution (C46) |
| the model has testable implications (C52) | the model can be rejected if the true data generating process diverges from their model (C52) |
| the observed evolution of the hazard of exiting nonemployment can be decomposed into contributions from structural duration dependence and unobserved heterogeneity (C41) | a mixed proportional hazard model can be proposed (C41) |
| the model can infer the size of the fixed cost of switching employment status using duration data and wage dynamics (C41) | even small fixed costs can significantly impact job search durations (J65) |
| the model cannot be rejected when testing against real data (C52) | substantial individual heterogeneity in job finding rates is revealed (J68) |
| the distribution of unobserved heterogeneity (D39) | the sign of the drift in the underlying Brownian motion (C69) |
| the true data generating process is characterized by a constant hazard of finding a job (C41) | their model will be rejected (C52) |
| the true data generating process follows a lognormal distribution (C51) | their model will be rejected (C52) |