Four essays in financial economics

Abstract

This dissertation is composed of four pieces of independent but related works. The first is on optimal risk sharing and CEO compensation. Using a principal agent model, I addressed theoretically the optimal tradeoff between the inefficient risk sharing and a high incentive, to conclude that the optimal CEO pay performance level should be negatively related to the level of idiosyncratic (undiversifiable) risk in the firm. I empirically tested the hypothesis and found strong support. The second paper is on the intermediation of labor income risk. From the perspective of optimal risk sharing, it looks at the recent trend of the rise intermediated labor force and its relationship with the riskness of human capital. I built an overlapping generations model to address the following issues: 1) what types of laborers get to work in conventional firms versus in labor intermediation firms? 2) how will the labor intermediation business be endogenously determined in equilibrium? 3) what effect does labor intermediation have on the ex ante decision of laborers to acquire specialized human capital? The third paper is a joint work with Stewart Myers. Empirical researches document that more developed financial systems seem to have higher level of idiosyncratic risk. Also, in the United States, idiosyncratic risk level goes up as the level of economic activity increases. This paper uses the framework developed in Myers (2000 JF) to explain such phenomena. The fourth paper is a joint work with Andrew Lo. It is on data-mining, and its role in explaining the financial anomalies documented in the literature. We developed a series of statistical and econometric methodologies to differentiate spurious statistical artifact from a real anomaly. We also give an extensive literature review of the landscape of the financial anomalies.