Abstract
This dissertation explores job matching process in the labor market and its implications for job match quality. The first chapter develops a theory of a firm's recruitment activities where the firm optimally chooses the quality of its workforce through worker selection, and studies its policy implications. The second chapter studies the cyclical behavior of match quality using a micro-level dataset.
In the first chapter, I incorporate worker selection into a random matching model with multi-worker firms. Unlike the standard random matching model, the worker selection model is compatible with establishment-level behavior of the hires-to-vacancy ratio, which (i) steeply rises with the employment growth rate, (ii) falls with establishment size, and (iii) rises with worker turnover rate. I calibrate the worker selection model to match the salient features of the U.S. labor market and compare it with the standard matching model without worker selection. I show that accounting for these patterns has both aggregate and firm-level implications for labor market policies. A hiring subsidy reduces aggregate unemployment substantially in the worker selection model, whereas the reduction in aggregate unemployment is very small in the standard model. Similarly, a firing tax reduces aggregate unemployment more in the worker selection model. At the firm level, labor market policy changes have a relatively bigger impact on fast growing and high worker turnover firms in the worker selection model. In contrast, the standard model implies that slowly growing and low worker turnover firms are affected relatively more by labor market policy changes. The existing empirical evidence supports the predictions of the worker selection model about the firm-level effects of labor market policies.
In the second chapter, I study the cyclical behavior of employment duration, a proxy for match quality. Models with on-the-job search predict that jobs created during a recession have shorter spells, because workers are more likely to accept low-quality jobs during a recession. In contrast, models with endogenous separation predict that jobs created during a recession endure longer, because firms contact a larger group of applicants and are able to hire high-quality workers. I test these competing predictions using data from the National Longitudinal Survey of Youth 1979 cohort. I estimate a proportional hazard model under the assumption that job terminations due to different reasons are competing risks. My results support the predictions of both models with on-the-job search and endogenous separation. A higher unemployment rate at the start of an employment relationship increases the probability that the worker quits to take or look for another job, but it decreases the probability the firm fires the worker. The net effect of these opposing forces on the overall duration of the employment is negative, but small, implying that match quality is weakly pro-cyclical. Furthermore, an increase in the current unemployment rate reduces the probability that the job spell ends by the worker's quit decision, but it increases the probability that the firm fires the worker. These findings are consistent with pro-cyclical quits and counter-cyclical firings.
