Theses Doctoral

Essays in Macro-Labor

Dorn, Agnieszka

In the first chapter of this dissertation, I estimate the cyclicality of real wages for job stayers, and hires from both employment and from unemployment, using an administrative matched employer-employee dataset from Germany. I find that the wages of new hires appear to be lessprocyclical than the wages of job stayers. I propose an explanation based on countercyclical selection on match quality: when aggregate productivity is low, worker-firm matches have to be unusually productive to warrant job creation. The presence of the match quality selection effect is supported by the relationship between the initial aggregate conditions and the subsequent risk of separation: jobs started when unemployment is high are at a decreased risk of ending with a separation to unemployment, which suggests that they are positively selected.

Motivated by the findings of the first chapter, I build a Diamond-Mortensen-Pissarides search and matching model with match-specific productivity and turnover costs. The model-generated wages and job durations have cyclical properties empirically established in the previous chapter: the wages of new hires are less procyclical than the wages of job stayers, and jobs started when productivity is high are at a higher risk of subsequent separation. I show that the relative cyclical properties of wages are generated by changes in average match-specific productivity for new hires relative to job stayers. Match-specific productivity is subjected to countercyclical selection: when aggregate productivity is low, match-specific productivity has to high to justify creating or maintaining a match. Due to turnover costs, countercyclical selection for new hires is stronger than for job stayers. Low match-specific productivity of matches started when aggregate productivity is high generates the positive relationship between initial aggregate productivity and subsequent risk of separation.

In the third chapter, I examine the behavior of wages within employment spells, before separation from a job and after movement between jobs in order to evaluate hypotheses concerning job-to-job transitions. Using German administrative microdata, I establish three empirical findings. First, the properties of wage changes within employment spells and associated with job-to-job transitions are broadly similar. Second, wages deteriorate in the year preceding separation from a job, for all separations, including job-to-job transitions. The wage deterioration manifests as slower wage growth and lowering of real wages expected given workers' characteristics. Third, for job-to-job transitions wage growth after accession is faster if the initial wage is lower than the last wage in the previous job. This effect is not present for job-unemployment-job transitions. The second finding supports the notion that some job-to-job transitions are induced by the worsened job situation. The third suggests that, to some extent, workers might voluntarily make job-to-job transition that decreases their wages in expectation of higher wage growth in the future.


  • thumnail for Dorn_columbia_0054D_15459.pdf Dorn_columbia_0054D_15459.pdf application/pdf 702 KB Download File

More About This Work

Academic Units
Thesis Advisors
Drenik, Andres Pablo
Ph.D., Columbia University
Published Here
October 7, 2019