Essays on factor demand adjustment

Thesis (Ph. D.)--University of Rochester. Department of Economics, 2015.This dissertation studies two aspects of firm factor demand decisions. The first chapter explores the excess labor reallocation at the firm level. The second chapter constructs a new measure of firm-level uncertainty and tests theories of investment under uncertainty. In the first chapter, I study the importance of match quality heterogeneity and costly information in excess labor reallocation at the firm level. Churning, defined as the sum of separations at expanding establishments and hires at shrinking ones, accounts for two thirds of gross worker flows at quarterly frequency. Moreover, churn rate is increasing in employment growth rate; more aggressively hiring firms also separate more. To explain these facts, I develop a search and matching model where firms do not observe candidates' qualities upon meeting. Given a costly screening technology, firms choose when to resolve the uncertainty: immediate discovery or gradual learning on the job. Faster growing firms screen less intensively to reap short-term profits, only shaking out their poorer matches later. This explains the positive association between churn and employment growth rates. The model is calibrated to match job finding and loss rates, the cross sectional employment growth distribution, and the distribution of job tenures. It is able to account for the cross sectional relationship between hiring/separation rates and employment growth, and about 50% of the observed churn. Understanding the exact nature of churn is important. For example, given the uncertainty in match quality, churning allows firms to grow faster, and as a result, enhances the efficiency of worker reallocation. A lower yield per vacancy may reflect increased screening intensity. The outward shift in the Beveridge Curve during the recent recession, often viewed as a decrease in matching efficiency, should therefore be interpreted with caution. In the second chapter, based on the risk neutral densities recovered from stock option prices, we construct a new measure of firm-level uncertainty for a quarterly panel of US public companies from 1996 to 2010. We find that this new measure is highly correlated with other widely used measures--realized stock return volatility and Black-Scholes implied volatility. However, the correlations between the skewness and kurtosis of the risk neutral densities and those of realized stock returns are very close to zero, suggesting "misperception" of high order moments by the market. The q model and the "real option" model of investment under uncertainty are then estimated using all three measures. For the q model, we find that both our new measure and the implied volatility have sizable negative impacts on investment, conditional on q, while the realized stock return volatility has a weak positive impact. For implications of the real option model, we do not find strong evidence of sluggishness in response to high uncertainty using any measure. Moreover, we find that investment responds concavely to positive demand shocks, a feature incompatible with the supermodularity in production technology