"Risk and Uncertainty in Labor Markets"

In this dissertation I analyze the effects of various uncertainties on aggregate economy, especially on unemployment. For that reason I build a general equilibrium model with heterogeneous firms with Diamond-Mortensen-Pissarides style search-and-matching in the labor market. Firms can hire more than one worker and hiring decision is partially irreversible due to linear hiring costs. First, I analyze the effect of an increase in time-varying idiosyncratic volatility, which is a mean-preserving increase in the standard deviation of firm-level productivity. I show that the model that contains idiosyncratic volatility shock in addition to aggregate productivity shock explains 60% of the variation in unemployment and 54% of the variation in vacancies, while also performing well in consumption and investment dynamics. In that model if idiosyncratic volatility rises, unemployment rate increases. However, workers become more productive due to the reallocation of resources from low productive to high productive firms. Thus, even though there are fewer people working, total wage bill and consumption is larger. Output and capital also increase with volatility because the increase in capital demand of large and highly productive firms dominates the reduction in capital demand of small and less productive firms when volatility goes up. The irreversible search costs in the labor market by itself are not large enough to induce large and counteracting option value effects of volatility.

Secondly, I solve the model with low and high ambiguity aversion levels to understand the role of model uncertainty. I show that since low-volatility states are the ones with low utility, ambiguity-averse households distort the conditional expectations by putting more weight on low-volatility states when higher ambiguity aversion is considered. In addition, the distortion creates a correlation between aggregate productivity and idiosyncratic volatility shocks, which are independent in the benchmark distribution. However, the additional effect of this distortion and the correlation it induces on the dynamics of aggregate variables is negligible.

Lastly, I add disasters in terms of capital depreciation into the baseline model and analyze effects of an increase in probability of a disaster on the economy. When probability of a disaster increases without an actual realization of disaster, the rate of return on capital net of depreciation becomes riskier and lower on average, thus household reduces investment today. Rental rate of capital goes up, capital and output decline over time. On the separation margin, since marginal benefit of a worker to a firm depends negatively on the rental rate of capital, separation thresholds increase for all firms, expanding the separation region, thus increasing the total separations. The rise in number of unemployed lowers the labor market tightness. The effect on the hiring margin is non-trivial. Lower labor market tightness makes hiring less costly for all firms, however higher disaster probability also reduces the marginal benefit of a worker to a firm. Responses of hiring threshold at various productivity levels show that the reduction in marginal benefit outweighs the reduction in marginal cost at lower productivity levels, and the opposite result holds at higher productivity levels. On aggregate matches are increasing because the number of hires made by highly productive firms is greater than the number of hires less productive firms decided not to make after the increase in disaster risk.