In this paper I introduce a novel source of residual wage dispersion. In the model, workers are heterogenous in productivity and randomly apply to ex ante identical posted vacancies. Each employer simultaneously meets several applicants, offers the position to the best candidate and bargains with her about the wage. Since the outside option of the employer is to hire the second-best worker, the wage paid to the best applicant decreases in the productivity of her closest competitor. Because the assignment of workers to vacancies is random in equilibrium, each worker faces a nondegenerate distribution of wages given her productivity before applying to a job. The framework suggests that the capability of search models to generate residual wage dispersion must be restricted to match-specific shocks.The model also predicts (i) residual wage dispersion of level wages increasing in productivity; (ii) residual wage dispersion of log wages decreasing in productivity; (iii) a negative relation between unemployment and residual wage dispersion and (iv) positive relation between productivity dispersion and residual wage dispersion. To assess these empirical predictions, I calibrate the model to match the mean and variance of the log wages in CPS data 1985-2006. The model`s predictions are strongly supported in the data.
Keywords: Nonsequential search, Recruiting selection, Wage Dispersion