A theoretical and empirical investigation of banks' risk taking behavior

This dissertation provides a theoretical and empirical investigation of banks' risk-taking behavior. The conventional wisdom believes that banks' capitalization is negatively correlated with risk, because the option value of deposit insurance decreases with capitalization. Chapter I intends to shed new light on this issue by expanding existing models in two dimensions. First, by incorporating into a single model the three different incentives of three agents; and second, by introducing four assumptions on the risk-return profiles. By combining these two factors, the model demonstrates that banks' risk can either decrease or increase with capitalization depending on relative forces of differing incentives of agents and on various parametric assumptions about risk-return profiles.; Chapter II provides an empirical study for the theoretical analysis in Chapter I. The regression equation is modeled in such a way that the differing incentives of the three agents are allowed to interact with each other. The regression results show apparent differences in risk-capitalization relationships across high and low capital banks, which indicates the presence of regulator's incentives in determining risk for low-capital banks. For high-capital banks, it is shown that the positive relationship between risk and capitalization strengthens as the shareholders' incentives gain relative forces, and weakens as the entrenched managers' incentives gain relative forces.; Chapter III studies a manager's risk-taking behavior in the banking industry when the manager has career concerns in the labor market. It is well known that there exists an investment distortion away from the first best outcome when the manager has career concerns in the labor market. For example, a risk-averse manager will prefer the riskiest investment project available when the manager's choice of a risky project is observable. This paper extends previous studies by allowing that the manager's choice of a risky project is not observable. Using the second best reputational equilibrium, it is shown that a risk-neutral manager can choose the least risky asset available. An empirical test shows no discernible relationship between career concerns and asset risk, implying that the manager is roughly risk-neutral.