The role of technological change in labor market polarization has been subject to recent critique. This paper finds that changes in production technology played an important role in wage and job polarization in the US. It also demonstrates that such technological change is consistentwith the timing of labor market polarization in the US, including the stagnation of the 50/10 wage percentile ratio and the slowdown of employment growth in high-wage jobs from the 2000s. The paper does so using a model with two key ingredients: 1) directed search and 2) two-sided multi-dimensional heterogeneity. Estimation results show that the complementarity between cognitive skill and task increased while that between manual skill and task did not. The full model can fullyaccount for the rise and fall of the 90/50 and 50/10 wage percentile ratios respectively. It also generates 72.6 percent of the rise in employment share of high-paying jobs relative to middling jobs and 69 percent of the fall in employment share of middling jobs relative to low-paying jobs. The paper suggests that the stagnation of the 50/10 wage ratio may be due to rank-switching between workers across the wage distribution from the 2000s, while the slowdown of employment growthin high-wage jobs may result from the trade-off between the returns to applying for high-wage jobs and the likelihood of being hired.

This paper studies aggregate labor market dynamics when workers have heterogeneous skills for tasks which are subject to non-uniform labor demand shocks. When workers have different skills, movements in aggregate wages partly reflect a reallocation of different workers across tasks and into employment. This ensures that there nearly always exists some combination of task-specific demand shocks that induce aggregate employment and wages to negatively comove even in a frictionless economy. Furthermore, such reallocations would be interpreted either as a labor wedge or as a shift in an aggregate labor supply curve in representative agent economies. Developing a method to estimate the multidimensional skill distribution, I show that a frictionless model with realistic heterogeneity can replicate the mean wage increase and employment collapse of the Great Recession. Reduced-form composition-adjustment methods recover positive co-movements between employment and wages in recent periods suggesting an increasing role for composition effects through time, which the model rationalizes through changes in the skill distribution and composition of sectoral shocks.

This volume contains original research articles which analyze the linkages between education and skills and the causes and consequences of different types of skill mismatch. The volume yields new insights regarding overeducation, underskilling, graduate jobs, wages returns to skills, aggregate productivity, job complexity and skill development.

In my thesis, I study the effects of agents' heterogeneity on labor market outcomes, with particular focus on sorting, performance, wages, and inequality. Chapter one studies multidimensional matching between workers and jobs. Workers differ in manual and cognitive skills and sort into jobs that demand different combinations of these two skills. To study this multidimensional sorting, I develop a theoretical framework that generalizes the unidimensional notion of assortative matching. I derive the equilibrium in closed form and use this explicit solution to study biased technological change. The key finding is that an increase in worker-job complementarities in cognitive relative to manual inputs leads to more pronounced sorting and wage inequality across cognitive relative to manual skills. This can trigger wage polarization and boost aggregate wage dispersion. I then estimate the model for the US during the 1990s. I identify a significant increase in complementarities of cognitive inputs and in cognitive skill-bias in production. Counterfactual exercises suggest that these technology shifts can account for observed changes in worker-job sorting, wage polarization and a significant part of the increase in US wage dispersion. Chapter two develops a theory that links differences in men's and women's social networks to disparities in their labor market performance. We are motivated by our empirical finding that men's and women's networks differ. Men have a higher degree (more network links) than women, but women have a higher clustering coefficient (a woman's friends are also friends among each other). In our model, a worker with a higher degree has better access to information. In turn, a worker with a higher clustering coefficient faces more peer pressure. Both peer pressure and access to information can attenuate a team moral hazard problem in the work place. But whether peer pressure or access to information is more important depends on the work environment. We find that, in environments where uncertainty is high, information is crucial and, therefore, men outperform women / in line with findings from sectors with high earnings' uncertainty like the financial or film industry.

My dissertation focuses on the heterogeneity in labor markets. The first chapter proposes an explanation for the unemployment rate difference between skill groups. Low skill workers (workers without a four year college degree) have a higher unemployment rate. The reason for that " ... is mainly because they (low skill workers) are more likely to become unemployed, not because they remain unemployed longer, once unemployed" (Layard, Nickell, Jackman, 1991, p. 44). This chapter proposes an explanation for the difference in job separation probabilities between these skill groups: high skill workers have lower job separation probabilities as they are selected more effectively during the hiring process. I use a labor search model with match specific quality to quantify the explanatory power of this hypothesis on differences in job separation probabilities and unemployment rates across skill groups. The second chapter analyzes the effects of one channel of interaction (job competition) between skill groups on their labor market outcomes. Do skilled workers prefer unskilled jobs to being unemployed? If so, skilled workers compete with unskilled workers for those jobs. Job competition generates interaction between the labor market outcomes of these groups. I use a heterogeneous agents model with skilled and unskilled workers in which the only interaction across groups is the job competition. Direct effects of job competition are reducing skilled unemployment rate (since they have a bigger market) and increasing the unskilled unemployment rate (since they face greater competition). However number of vacancies respond to job competition in equilibrium. For instance, unskilled firms have incentives to open more vacancies since filling a vacancy is easier if there is job competition. Thus how unskilled unemployment and wages are affected by job competition depends on which effect dominates. The results for reasonable parameter values show that job competition does reduce the average unemployment rate. It reduces the skilled unemployment rate more, generating an increase in unemployment rate inequality. However, the employment rate at skilled jobs is unaffected. The third chapter focuses on skill biased technological change. Skill biased technological change is one of the explanations for the asymmetry between labor market outcomes of skill groups over the last few decades. However, during this time period there were also skill neutral shocks that could contribute to these outcomes. The third chapter analyzes the effects of skill biased and neutral shocks on overall labor market variables. I use a model in which skilled and unskilled outputs are intermediate goods, and final good sector receives all the shocks. A numerical exercise shows that both skilled and unskilled unemployment rates respond to shocks in the same direction. The response of unemployment rate to skill neutral shocks is bigger than the response to skill biased shocks for both skill groups. However, the unskilled unemployment changes more than the skilled unemployment rate as a response to skill neutral shocks. Thus, skill neutral shocks reduce the unemployment rate gap between skill groups.

Since applying for jobs is costly, workers prefer applying where their employment probability is high and, therefore, to jobs attracting fewer higher quality applicants. Since creating vacancies is expensive, firms create more vacancies when job-seeking is high. Our model captures these ideas and accounts for worker heterogeneity by assuming three types of nearly identical workers. These infinitesimal quality differences generate a discrete wage distribution. For some parameter values lower quality workers have discretely lower wages and higher unemployment than better workers. Moreover, increasing the number of the lowest quality workers can make all workers better off.