Conventional DSGE models that include labor search-and-matching frictions assume that only unemployed workers are job seekers. Without taking the behavior of employed job seekers into account, these standard models overestimate the true magnitude of fluctuations in the entire population of job seekers. Hence, our first chapter proposes augmenting the standard DSGE model with on-the-job search by employed workers. Using Bayesian estimation methods, I show that this augmented model not only well explains the 25 percent decline in hiring that was seen during the Great Recession, but also successfully predicts the subsequent actual five-year-long recovery period of hiring. In contrast, the standard models, which do not incorporate employed workers' on-the-job searches, fail to explain the sharp declines in hiring seen during the Great Recession and the slow job recovery afterward. Furthermore, I find that if on-the-job-search is incorporated in the model, the decline in matching efficiency of the unemployed workers would explain 54 percent of the increase in U.S. unemployment during the Great Recession, as opposed to 27 percent suggested by the standard DSGE models without on-the-job search.The second chapter uses Bayesian methods to estimate a search and matching model that permits job rationing featuring demand and matching efficiency shocks. In such a model, unemployment can exist even in the absence of search frictions. These models have been used to show that jobs rationing, rather than search frictions, causes unemployment to rise during recessions. Previous studies relied on calibration methods to show the extent of job rationing, but it remains unknown how sensitive this result is to the choice of parameters. Based on the parameters in literature, I set the prior probability that job rationing to occur as 1/2, and estimate these models with jobs rationing to update parameters according to observed data. Using these updated parameters (posterior means),I found no instances of unemployment due to jobs rationingin the United States between 1964:Q1--2015:Q3.Although the estimation results do not support the existence of job rationing, I found that the features such as wage rigidity and diminishing marginal productivity of labor thatcould lead to job rationing make the model do a better job of fitting the data than the fully flexible wages model.In the United States, there is substantial heterogeneity in labor market outcomes across demographic groups. Not only do young workers, non-white workers, and those without college degrees have persistently higher unemployment rates than other demographic groups, but these groups also experience substantially larger increases in unemployment rates during recessions. To understand the source of these differences, in the third chapter we decompose the unemployment rate for each demographic group into flows between unemployment, employment, and out-of-the-labor-force.We find that the gap in unemployment rates between these disadvantaged groups and other demographic groups can be primarily attributed to the fact that disadvantaged groups have higher rates of exit from employment, although hiring also plays a small role.In contrast, we find that the cyclical increase in unemployment is primarily driven by a reduction in hiring, which can explain 2/3 to 3/4 of the cyclical fluctuations in the unemployment rate across all demographic groups.Thus, policies that can reduce exit rates from employment for these disadvantaged groups can reduce both the persistent and cyclical disparities in the unemployment rate across demographic groups.
PDF
download
878987797
028-85220240
