Econometric Analysis of Financial Risk and Correlation

Abstract

The contribution of this dissertation is threefold. First, econometric procedures to test for the occurence of asset price bubbles ex post and in real time are proposed. Real time monitoring procedures represent an additional tool for financial agents to gauge whether or not a bubble is building up in a financial market at the date of measurement. Second, we consider the problem of risk assessment and performance measurement. Risk assessment is essential for determining the amount of required capital. It is also important to counterbalance expected profits. The focus here lies on the economic index of riskiness proposed by Aumann and Serrano (2008). <br /> New theoretical properties of the index are established and estimation techniques are proposed. It is brought to application as a counterweight to expected returns to measure the perfomance of mutual funds and hedge funds.<br /> The last part of the dissertation is of more basic econometric interest. It addresses the issue of the validity of standard inference procedures in fixed effect panel data models. Ordinary least squares inference about model parameters can be misleading if shocks to cross-sectional units are correlated. Existing tests of cross-section error dependence aim at determining whether or not there is cross-section error correlation per se. In this dissertation, a procedure is developed that aims at testing whether there is cross-section error correlation that invalidates ordinary least squares inference.