Job separation, wage rigidity, and bargaining power in search models

This dissertation focuses on job separation, wage rigidity in search models and workers’ bargaining power in the US labor market. In the first chapter, I use CPS duration data to measure job finding and separation rates and quantify their contribution to unemployment variability. Since 1948, the rate at which workers separate from a job in the United States has averaged 4.5 percent a month, contrary to the frequently used 3.5 percent figure. Contemporaneous fluctuations in the separation rate explain about 12 to 47 percent of unemployment variability, depending on how the data are detrended. Over the last two decades, fluctuations in the separation rate have become increasingly irrelevant in explaining unemployment fluctuations, explaining only about 4 − 12 percent. The job finding rate, on the other hand, continues to explain about 90 percent of unemployment fluctuations. The second chapter explores the wage negotiating decision in order to understand this phenomenon, provide a model based foundation for the degree of wage rigidity and understand the role wage rigidity plays in accounting for volatility in labor market fluctuations. I develop a search model where firms pay different wages to incumbent workers and new hires. I extend prior work by generating an endogenous degree of wage rigidity through the choices of workers and firms in the model. My model, calibrated to quarterly US data over the 1951-2018 period, does well in matching the volatility of US labor market variables. Specifically, I find that the model can explain over 70 percent of the volatility in unemployment and vacancies. Unlike similar models, the model also produces the Beveridge curve relationship between unemployment and vacancies. Furthermore, I use the model to show that the interaction between wage rigidity, decreased macroeconomic volatility, and decreased job separation can contribute to explaining the wage-productivity gap in the United States. In the third chapter, I estimate workers’ bargaining power over time from a wage equation derived from a search model. The estimates show that workers’ bargaining power declined in the United States between 1975 and 2018. Standard OLS computations show a drop of 33 to 62 percent between the periods pre- and post-1997. Using time-varying parameter regression estimation, I compute quarterly workers’ bargaining power series for 1975Q1-2018Q4, which show a consistent decline. This is the first time series of bargaining power with potential to advance our understanding of the behavior of bargaining power in the US and the factors that affect it. Caution is needed in interpreting these results because a key variable (the vacancy rate) had to be estimated for the period before 2001, while another one (unemployment benefits) is potentially endogenous. Thus the empirical results should be regarded as preliminary.