Financial risk management and value-at-risk:

Abstract

Value-at-Risk (V@R) is a new, simple and yet informative measure of portfolio risk. This dissertation explores the impact of assumptions about an asset return-generating process on portfolio risk measurement and management within this new framework of enterprise financial risk management. The major contribution of this study is an identification of potential problems and consequences of model misspecification on risk measurement and ultimately on the corporate hedging decision. Chapter One reviews the normative and positive literature on financial risk management. Chapter Two explores the impact of model misspecification if the true (simulated) model returns are from a mixture of normal distributions with feasible parameters. The distributional properties of several assets, such as stocks, currencies and commodities, are examined in Chapter Three. The Maximum Likelihood Method and Method of Moments are used to estimate posited models. parameters and fit empirical distributions. V@R measures calculated from a mixture-of-normals model are then compared to measures from models commonly used by practitioners (e.g., RiskMetrics.), who assume either Gaussian, or some form of an ARCH (EWMA) process. The last chapter empirically explores characteristics of companies using V@R systems with a focus on the benefits and uses of Value-at-Risk systems in financial risk management.